Study better. Learn faster. Get the grade you want.
Discover why millions of students use us to learn better.

- StudyBlue
- Iowa
- Iowa State University
- Psychology
- Psychology 301
- Cooper
- PSYCH 301 Study Guide (2013-14 Cooper)

Join StudyBlue to buy this Study Guide
Occur when the participant changes his or her behavior because he or she identifies the manipulations in the study.

this occurs when the ordering of values on the scale represents the ordering of the trait being measured

the logic of experimentation breaks down if you have a confound because it could either be the confound or the IV responsible for the change in the DV

occurs when the effect of one independent variable depends on another, inferential stats, when graphed lines touch or are not paralell.

as one variable increases, then the other variable also tends to increase. or as one decreases the other also tends to decrease

In comparing Mean and Median, if the Mean is bigger than the Median, what kind of distribution is this?

In comparing Mean and Median, what kind of distribution is it when the Mean is smaller than the Median?

Every member of the population to which we would like to generalize the results has an equally likely chance to participate

Demand characteristic that occurs when subjects change their behavior due to unintentional cues from the researcher

Goals of Behavioral Research

1. describe behavior 2. predict behavior 3. determine cause of behavior 4. understand/explain behavior

Scientific approach

best way to avoid arguments, requires everyone play by the rule of science

rules of science

1. systematic empiricism 2. public verification 3. solvable problems

systematic empiricism

conclusions and arguments based on data gathered from actual observations, data collected in a planned/organized/systematic way

public verification

data, analyses, and procedures must inspected by others so they can check and evaluate basis of arguments

solvable problems

needs to be something that can have an answer and can be studied

Theories

used to predict and explain behavior, specify the relationships and effects that exist between concepts

a theory is never...

considered "true", model of how people tend to behave, predict results of different studies

theory

explanation for a phenomenon that can be falsifiable and involves entities that cannot be directly observed

Theory

Is an explanation for a phenomena that can be falsified and that involves entities that cannot be directly observed

Theory

An explanation that has been supported by many, many experiments

Theory

Explanation for a phenomena that can be falsified and involes entities that cannot be directly observed

characteristics of a theory

1. specifies relationships among concepts/variables 2. entails propositions (how one variable effects another) 3. can be used to generate a hypthoeses 4. is falsifiable

hypotheses

prediction about what will occur or why something did occur

priori hypotheses

prediction made before the scientific test

post hoc hypothesis

explanation for why something that already happened might have occured

whats viewed more favorably?

priori hypotheses that are supported by data are view more favorably, but post hoc hypotheses have their place

what is post hoc hypothesis place?

scientists come upon an unexpected phenomenon or observation they've made no predictions about

deduced

logically derived

induced

inferred from empirical data/observations

conceptual definitions

communicates generally what we mean by these terms

operational definitions

how those concepts are manipulated or measured

methodological pluralism

when a theory is tested in many different ways, using different measures and manipulations (supported this way=good theory)

Strong inference

study designed in a way that one outcome with support one theory and another outcome will support a different one

who decides whats is scientifically true/meaningful?

everyone...investigator and their team, editor of a scientific journal, other scientists, then the general population

types of research

1. descriptive research 2. correlational research 3. experimental research

descriptive research

research that simply characterizes one or more variables w/o investigating the relationship

correlational reasearch

explores realtionship between two or more variable without experimental manipulation

experimental research

kind of research the investigators exercise great control over (random assignment, different conditions)

naturalistic observation

researcher observes participants

naturalistic observation

research design in which the researcher simply observes and describes behavior

naturalistic observation

a research technique in which the researcher observes and describes behavior

Naturalistic Observation

A research technique in which the researcher simply observes and describes behavior

Naturalistic Observation

the observer becomes a participant in the group being observed

naturalistic observation

a naturalistic observation in which the observer becomes a participant in the group being observed

Naturalistic Observation

A research technique in which the researcher simply oversees and describes behavior

Naturalistic Observation

A research technique in which the researcher simply observes and describes technique.

Naturalistic Observation

Researcher simply observes/describes behavior

Naturalistic Observation

A research technique in which the researcher simply observes and describes behavior

Best used to study questions that do not involve a relationship between variables

participant observation

a kind of naturalistic, research participates in the environment

Participant Observation

nat. observation in which the researcher becomes a member of the group being observed

Participant observation

research technique in which the researcher infiltrates a group of individuals in order to study their behavior and relationships

Participant Observation

A naturalistic observation in which the researcher becomes a member of the group being observed

participant observation

the researcher takes on a role in the social situation under observation,\

Participant observation

a naturalistic observation in which the researcher becomes a member of the group being observed

Adv: allows access to groups otherwise inaccessible

Disadv: the participation of the researcher may alter the behavior being studied

problems with participant observation

1. researchers might unintentionally influence the behavior of the real participants 2. it might influence the researcher's objectivity

contrived observation

researcher sets up environment in which they observe the participant, may be an artificial setting (lab room, etc.)

difference between naturalistic v. contrived

naturalistic researcher tries to observe participants in real settings while contrived the researcher views in settings made up

disguised v. nondisguised observation

whether or not the participants are aware

reactivity

problem with nondisguised observation, participant behaves differently when they know they're being watched

partial concealment

participants are aware they are being observed but they do not know what the observer is looking at

disguised observation

participant does not know they are being observed

knowledgeable informants

strategy involves asking other people who have info about participant (talking to family, friends, etc.)

unobtrusive measures

observe something that is related to the behavior you want to study but not directly related.

behavioral recording (how researcher aquires data)

1. narratives 2. field notes 3. checklists 4. temporal measures (time related) 5. behavioral rating scales 6. physiological measures 7. self-report questionnaires

narratives

diaries, memoirs, etc. not that useful because they may not be reliable enough

field notes

researcher observes participant and takes notes

checklists

list of behaviors that might occur, simply record each time that behavior is observed

temporal measures (time related)

how long it takes for something to start occuring after a specific point in time

behavioral rating scales

used when we ant to know how much a behavior occured or how intensely something was experienced

physiological measures

blood pressure, sweating, heart rate, etc.

physiological (measure)

heart beating fast, blood pressure, etc

self report questionnaires

used a lot in psych, present a list of questions/statements

self report biases

social desirability, acquiescence, nay-saying

social desirability

when the participant answers in ways that they think are appropriate

acquiescence

participant tends to agree with questions, not matter how they are asked

nay-saying

opposite of acquiescence, disagreeing with everything

archival data

data that has been collected by someone else and is stored somewhere (census, death certificates, etc.)

scales of measurement

1.ratio 2.interval 3. ordinal 4. nominal

Scales of Measurement

-Identity

-Magnitude

-Equal Intervals

-Absolute Zero

ratio

scale has an absolute zero point and equal intervals between the scale scores correspond to equal intervals in the variable being measured

ratio

has identity, magnitude, equal intervals, and absolute zero (weight in lbs, number correct on test, Kelvin)

all arithmetic operations are meaningful

F ratio

measures how much greater MS treat is than MS within and thus whether the IV is having an effect on the DV

interval

scale has equal intervals between the scale scores which correspond to the equal intervals in the variable being measured, no true zero

interval

has only the identity, magnitude, and equal intervals properties (Fahrenheit, Celsius)

addition and subtraction are meaningful

multiplication and division are not

intervals

have only the identity, magnitude, and equal intervals properties.

Examples: Fahrenheit temp and Celsius temp

Examples: Fahrenheit temp and Celsius temp

Interval

have only magnitude, identity, and equal interval properties

ex. Fahrenheit Temp

Celsius Temp

addition and subtraction are meaningful, but multiplication and division are not

ordinal

only tell us the order of variable values, not telling us the intervals

ordinal

has only the identity and magnitude properties (bball rankings)

no arithmetic operations are meaningful

ordinal

have only the identity and magnitude properties

example: basketball rankings, class rankings

No arithmetic operations are meaningful.

example: basketball rankings, class rankings

No arithmetic operations are meaningful.

nomial

all we know is that the "scores" are different, scores have no mathematical meaning except that they are not equal

nomial

has only the identity property

( zip codes/ jersey numbers)

no arithmetic operations are meaningful

types of measuring

observational, physiological

observational

researchers watch participants and make ratings

converging operations

measure it in several different ways

reliability

degree to which a given measure gives you the same score over and over again

reliability=

true score variance/total variance in scores

observed score=

true score+measurement error

total variance score=

true score variance+error variance

reliability is assessed...

through correlations

true scores will..

true scores will not..

measurement error will not..

true scores will not..

measurement error will not..

..correlate with true scores

..correlate with measurement error

..correlate with measurement error

..correlate with measurement error

..correlate with measurement error

test-retest reliability

when you administer a measure once, then you wait and administer it again later

in addition to assessing reliability across time (w/ test-retest) we can also assess..

reliability among items within a single administration of a measure

kinds of inter-item reliability

1. item-total reliability 2. split half reliability 3. cronbach's alpha

item-total reliability

examine the correlation of each item to the average of the rest of the items on the scale. useful for finding particular items that might not be measuring the same thing as the other items

split half reliability

items for a measure are divided in half and the average score for 1 half is correlated with the average score for the other half. shows if the 2 halves are measuring the same thing or not

Cronbach's alpha

A correlation-based statistic that measures a scale's internal reliability.

validity

how much the measure reflects the construct it was intended to measure

types of validity

1. face validity 2. content validity 3. construct validity 4. criterion related validity

face validity

it seems like it should measure the construct we want to measure

content validity

physiological construct that often entails a variety of behaviors or characteristics

construct validity & 2 types

corresponds to the measure correlating with other measures that you would expect it to correlate with; 1. convergent validity, 2. discriminant validity

convergent validity

measure correlate with related measures (ex. anxiety correlating positively with depression and negatively with optimism.)

discriminant validity

measure has this when it is unrelated to measures that it shouldn't be related to. (ex. anxiety and intelligence close to 0)

criterion related validity & 2 types

degree to which a measure correlates with some observable behavioral outcome; 1. concurrent validity, 2. predictive validity

concurrent validity

measure correlates with a behavioral criterion that is observed at the same point in time

predictive validity

measure correlates with a behavioral criterion that is observed at some future point

test bias

two groups of people get diff. scores on a measure but they should actually have the same regard to the construct of the question, hard to know if the scores are lower from reason or from bias

how to test test bias

use scores to predict future criterion that both groups have an equal opportunity to do well with, if the 2 groups score the same the test was probably biased

descriptive stats

used to describe or summarize behavior/characteristics of a group of people

inferential stats

what we use to make decisions, or to decide what conclusion to make about a result of a study

inferential stats allow us to..

draw conclusions about the data we've received, if the relationship differences are significant they are unlikely due to the chance differences that always affect results in research

some stats

mean, range, deviation score, variance, standard deviation

mean

average

mean

The mean is the sum of all the scores divided by the number of scores in the data. The average

Mean

the average

provides consistent results

range

subtract smallest number from biggest number (ex. 1 and 9...9-1=8)

deviation score

how much the score deviates from the mean.. (score=9...mean=5...deviation score=4)

variance

measure of how much the scores vary about the mean (square deviation score)

variance

pop variance= SS/N

sample variance= SS/N-1

standard deviation

take the square root of the variance

Standard deviation

square root of sample variance

population

every possible person that your research question applies to

data based on an entire population then stats are called

population parameter

sample stats

portion of the population that we measure

systematic variance

some of the variation in the dependent variable is systematically related to another variable like the IV

error variance

random effects that influence behavior

total variance=

systematic variance+error variance

what does it mean if systematic and error are pretty much equal?

you cannot be sure the behavior was influenced by the IV

if systematic is much greater than error?

it is very unlikely that the variance associated with the IV arose from random influences. we can conclude that the IV had an effect

approaches to ethical decisions

1. deontological or universal

2. ethical skepticism

3. utilitarian (cost/benefit)

2. ethical skepticism

3. utilitarian (cost/benefit)

deontological/universal

associated with the belief that there is a single, universal code for deciding what is right and wrong

ethical skepticism

ehtical decisions are left up to the individual because it is believed that all ethical guidelines are relative

utilitarian (cost/benefit)

considers the consequences of an action and the harm or benefit that might emerge

Institutional Review Board

any institution that receives funds must have an IRB oversee and approve research activities, must be a diverse board, not all scientists,

informed consent

proved the potential participant with the info about a research study. needs to be: written in plain language, describe general terms, inform individuals they have the right to refuse to participate, describe risks, provide info about the limits of confidentiality

3 conditions met=no informed consent

1. no risk to participants

2. right and welfare of participant will not be affected

3. not reasonably possible to obtain informed consent w/o ruining the study

2. right and welfare of participant will not be affected

3. not reasonably possible to obtain informed consent w/o ruining the study

When to use the between subjects variance test

used to compare the variances in between subjects designs that have one IV with exactly two levels

Within Subject Variance Test (when to use)

used to compare variances from within sub designs that have one IV with exactly 2 levels

Between subjects variance test

test of variances, 2 group and each one does one of the two groups

Within subjects variance test

2 IV with exactly 2 levels

Stupid Criticism #1

one cannot criticize an experiment for having some element that could have produced Type II error if an effect of IV on DV was found

Stupid Criticism #2

one cannot criticize an experiment for a subject assignment confound if the experiment did random assignment

Stupid Criticism #3

calling a nuisance variable a confound when the nuisance variable doesn't vary non randomly with IV

Latin Square

Method of counterbalancing - row: subject, column: treatment order

ABCD

DABC

CDAB

BCDA

Latin Square

ABCD

DABC

CDAB

BCDA

DABC

CDAB

BCDA

Latin Square

has as many columns and as many rows as there are IV levels. Rows represent subjects and columns represent treatment orders. Treatments are placed in every cell such that every treatment appears only once in every column and every row (Method of counterbalancing)

Latin Square

Columns represent treatment order and rows represent subjects.. each treatment appears only once in every column and every row

Use of a Latin Square

Has as many columns and rows as there are IV levels: columns represent treatment order and rows represent subjects

Counterbalancing

method of assigning subjects to treatment orders to that across subjects, practice effects are even

Counterbalancing

The practice of enlisting several groups in an experiment to offset the effects that may arise due to differences between them.

Counterbalancing

A solution to practice effects which assigns subjects to treatment orders so that across subjects, practice effects are even.

Counterbalancing

method of assigning subjects to treatment orders to a within SD so that, across subjects, practice effects are balanced

Counterbalancing

A method of assigning subjects to treatment orders in a within subjects design so that, across subjects, practice effects are balanced

Dependent Variable

the variable the researcher measures to determine effect of IV

Dependent Variable

The variable the experimenter measures to gauge the effects of the manipulations

dependent variables

the variable the researcher measures to determine the effects of the independent variable

Dependent Variable

Variable that the researcher measures

Dependent Variable

In an empirical study, the variable the experimenter measures to gauge the effects of his/her manipulations

Dependent Variable

Researcher measures to determine the effects of the IV

Dependent variable

the variable in an experiment the researcher measures to determine the effects of the IV

Dependent variable

the variable that the researcher measures to determine the effete of the IV on the DV.

When to use the correlational approach?

when manipulating variables would be impossible, difficult, or unethical

correlational approach

research technique in which the researcher determines relationship between variables without manipulating variables

when to use the correlational approach

when: 1) manipulating the variables would be difficult or impossible

2) manipulation the variables would be unethical

Correlational Approach

research technique in which the researcher determines relationship between variables without manipulating them

Correlational approach

a research technique in which the researcher determines the relationship between variables when manipulating variable would be difficult or impossible, or manipulating the variables would be unethical

Correlational Approach

research technique in which the researcher determines relationships between variables without manipulating the variables

1.manipulating variables would be difficult or impossible

2.manipulating variables would be unethical

Correlational Approach

Researcher determines the relationship between variables without manipulating the variables

When to Use Correlational Approach

1) Manipualating variables would be difficult or impossible 2) Manipulating variables would be unethical

Correlational approach

Research technique in which the researcher determines the relationship between variables with out manipulating the variable1. Manipulation is impossible

2. Manipulation is unethical

2. Manipulation is unethical

Correlation Approach

a research technique in which the researcher determines the relation between variance without manipulating the variable

Correlational Approach

Research technique in which the researcher determines the relationship between variables without maniuplating the variables

Correlational approach

a research technique in which the researcher determines the relationship between variables

Correlational Approach

a research approach in which the research determines the relationship between variables without manipulating the varibles

When to use correlational approach

used to determine the relationship between variables when:

1. Manipulating the variables would be difficult or impossible.

2. Manipulating the variables would be unethical.
Experiment

a study that: has random assignment, requires researcher to manipulate IV

experiment

a study that satisfies a criteria of random assignment, and the researcher manipulates the IV

Experiment

A research study that satisfies the following:1. Random assignment

2. Researcher manipulates IV

2. Researcher manipulates IV

Experiment

Must have random assignment and the researcher must manipulate the IV

Experiment

A research technique in which satisfies the following 2 criteria:

1. Random Assignment

2 Researcher manipulates the IV

Experiment

A research technique which satisfies the following two criteria

1. Random Assignment

2. Manipulation of the IV

(allows to infer causation)

Experiment

A research technique which satisfies the following criteria:

1. Random assignment

2. Researcher manipulates the independent variable

Positively Skewed Distribution

tail is to the right, few extreme scores, Mean>Median

in a positively skewed distribution the mean...

> median

Positively Skewed Distribution

Tail to right, few high scores

Mean>Median

Positively skewed distribution

distribution with a few extreme HIGH scores

Positively Skewed Distribution

Majority of the data falls to the left of the mean, the tail is on the right side of the graph

Positively Skewed Distribution

Few extreme high scores

Positively Skewed Distribution

a distribution with a few extreme high scores

ex: annual income, hours of sleep yesterday

positively skewed distribution

distribution with a few extreme high scores (tail to the right)

Positively Skewed Distribution

a distribution with a few extreme high scores

Mean>Median

Positively skewed distribution

a distribution that has a few extreme high scores.

Examples: annual income, hours of sleep for all people

Examples: annual income, hours of sleep for all people

Sensitization Effect

occur when subject changes their behavior because they realized what the manipulations are

**Most common reason why within subjects can't be used

Sensitization Effect

occur when subject changes their behavior because they realized what the manipulations are

sensitization effects

when the subjects become aware of the manipulation in a study and such awareness causes the subject to change his/her behavior

Sensitization Effects

occurs when the subjects become aware of the manipulation in a study and such awareness causes the subject to change his/her behavior

sensitization effects

occur when a subject changes their behavior b/c they realize what the manipulations are in a study

Sensitization Effects

Sensitization Effects

Subject realizes what the manipulations are and this awareness changes the behavior

Sensitization effects

occurs when the subject realizes what the manipulations are in a study and this awareness causes the subject to change his/her behavior

Sensitization Effects

Occur when a subject changes their behavior because they realize what the manipulations are in a study.

Must use between subjects design if this occurs..

Two methods of counterbalancing

1. use all combinations

2. use a latin square

what are the two methods of counterbalancing

use all possible treatment orders

use a latin square

Methods of Counterbalancing

a. Use all possible treatment orders (4 or fewer IV levels)

b. Use a Latin Scale (4 or more IV levels)

2 methods of counterbalancing

1. use all possible treatment orders (<4)

2. use a latin square (>4) column = treatment order, rows =subjects

4 properties of measurement scales

1. identity

2. magnitude

3. equal intervals

4. absolute zero

properties of scales of measurements

1) identify (occurs when different entities receive different scores

2) magnitude(occurs when the ordering of values on the scale reflects the ordering of the trait being measured

3) equal intervals(occurs when a difference of 1 on the scale represents the same amount of trait being measured everywhere on the scale

4) absolute zero (occurs when a score of zero indicates complete absence of the trait being measured

Properties of Scales of Measurement

1. Identity

2. Magnitude

3. Equal Interviews

4. Absolute Zero

Properties of Scales of Measurement

1. Identity: occurs when different entities receive different scores

2. Magnitude: occurs when the ordering of the values reflects the ordering of the trait being measured

3. Equal Intervals: occurs when a difference of 1 on the scale means the same amount everywhere on the scale

4. Absolute Zero: occurs when a score of 0 indicates complete absence of the trait being measured

4 properties of scales of measurement

(I'M EA)

1. Interval

2. Magnitude

3. Equal interval

4. Absolute zero

normal distribution

symmetrical bell shape, mean = median

in a normal distribution the mean...

= median

Normal Distribution

Mean=Median

Symmetrical

Normal distribution

a symmetrical bell-shaped distribution

normal distribution

symmetrical pattern of scores on a scale in which a majority of the scores are clustered near the center and a minority are at the extremes

Normal Distribution

Symmetrical and bell-shaped

Normal Distribution

a symmetrical, bell-shaped curve

ex: height, IQ

normal distribution

symmetrical, bell-shaped curve

Normal Distribution

a symmetrical bell-shaped distribution

X=median

Normal Distribution

a symmetrical, bell shaped distribution.

Examples: height (within sex), IQ scores

Examples: height (within sex), IQ scores

when to use the within subjects t-test

used to compare the means from within subjects designs that have 1 IV with exactly 2 levels

When to use between subjects t test

between sub design comparing means

When to use between subjects t-test

when comparing 2 sample means

-each sample has different people

when to use a b/w subjects t-test

used to compare the means for b/w subjects experiments w/ one independent variable that has exactly 2 levels

when to use between subjects t-test

between sub design comparing means, 1 IV, 2 levels

Between subjects t-test

One IV and 2 IV levels

WHEN TO USE: Within Subjects t-test

Comparing means that have one IV with two levels

internal validity

the extent to which your experiment provides a valid test of the relationship between IV and DV

Internal Validity

Extent to which your research provides a valid test of the relationship between the IV and the DV

Internal Validity

the intent to which your research provides valid tests of the relationship between the IV and DV

Type I error

finding an effect of IV on DV when in reality no such effect exists (False Positive)

type 1 error

finding an effect of the Iv on the DV when in reality no such effect exists

Type I Error

Occurs when experiment finds a relationship between the IV?DV variables that does not actually exist (False Positive)

Type 1 error

rejecting the null hypothesis when it is true

Type One Error

Finding an effect when in reality no such effect exists

Type Two Error

Failing to find an effect when in reality an effect does exist

Type 2 error

failing to find an effect of the IV on the DV when in reality an effect does exist (false negative)

Type 2 Error

Failing to find an effect of the IV on the DV when in reality an effect does exist

Type I error

Finding an effect of the IV on the DV when in fact no such relationship existed

Type 2 error

"false negative"

finding no effect of the IV on the DV but when in reality a effect exists

finding no effect of the IV on the DV but when in reality a effect exists

Type I error

Finding an effect of the IV when in reality there was no such effect

Causation

relationship is causal when the change of one variable results in a change of the other

i.e. supply and demand, NOT act and college gpa

causation

we say a causal relationship exists b/w 2 variables when a change in one results in a change in the other.

Causation

Causal relationship exists if a change in one results in an exchange in the other

Causation

We say a causal relationship exists between two variables if a change in one results in a change in the other.

Unlike experiments, the correlational approach does not allow a researcher to determine whether there is a causal relationship.

Unlike experiments, the correlational approach does not allow a researcher to determine whether there is a causal relationship.

Type II error

failing to find an effect of the IV on the DV when in reality an effect does exist (False negative)

Type II Error

Failing to find the effect of the IV on the DV when in reality the effect does exist (falso negative)

Type II Error

Occurs when experiment finds no relationship between the IV/DV variables when one exists in reality (False Negative)

Type II error

accepting the null hypothesis when it is false

Type II Error

Failing to find an effect of the IV on the DV when in reality there is an effect (false negative)

type II error

failing to find an effect of the IV on the DV when in reality an effect does exists

type II error

failing to find an effect of the IV on the when in reality an effect does exist

Type II error

Failing to find an effect of the IV on the DV. Alpha is usually set to .05 (False negative)

Type II errors

Finding no effect of the IV on the DV when in reality there was an effect,

Type II error

Not finding an effect of the IV when in reality there was one.

Hypothesis

prediction

hypothesis

a tentative statement about the relationship between observable variables

hypothesis

An educated guess made to explain an observation or to answer a question

Hypothesis

A tentative statement abou the possible relationship between observable areas

Hypothesis

A tentative statement about the possible relationship between observable areas

The falsifiability criterion for theories

there must be hypothetical facts the theory cannot explain

-must always be able to prove theory false

the falsifiability criterion for theories

for an explanation to be useful it must be able to generate predictions. as such there must be some type of hypothetical facts the theory would not be able to explain. it must always be possible to prove that a scientific theory is false.

The falsifiability criterion for theories

for a theory to be useful it must be able to generate predictions , as such there must be at least hypothetical facts that the theory would NOT be able to explain (falsified)

The falsifiability criterion for theories

-for an explanation to be useful it must be able to generate predictions.

-As such, there must be at least some hypothetical facts that would prove the theory false.

-it must always be possible to prove a scientific theory false.

-As such, there must be at least some hypothetical facts that would prove the theory false.

-it must always be possible to prove a scientific theory false.

Variability in DV

smaller variability gives more power

variability in DV

small variability gives more power

Variability in the DV

low variability in the DV gives more power

sample size

larger sample size gives more power

sample size

larger sample size giver more power

advantage of experiments

allows us to determine cause

Advantage of experiments

allows to infer causation

random selection

every member of the choice population has an equal chance to participate in the study

--Not absolutely necessary

Random Selection

Occurs when every member of the population to which we would like to generalize the results have an equally likely chance of being chosen to participate in the study

Random Selection

Everyone from the choice population has an equal chance of being involved in the study

Random Selection

Everyone in the chosen population has an equal chance of being selected

Random selection

Occurs when every member of the population to which we would like to generalize the results has an equally likely chance to participate in the research

Random Selection

Every member of the population to which we would like to generalize the results has an equally likely chance to participate

random selection

every member of the population to which we would like to generalize the results has an equally likely chance to participation in research

Random Selection

Occurs when every member of the population to which would like to generalize the results has an equally likely chance to particpate in the research

Random selection

occurs when every member of the population we would like to generalize to has an equally likely chance to participate in the research.

random assignment

each participant has an equally likely chance to be assigned to each IV level or treatment order

Random Assignment

Each participant has an equally likely chance to be assigned to each IV level (in a between subjects design) or to each treatment order (in a within subjects design)

random assignment

once the participant for the study have been chosen, random assignment occurs when each participant has an equally likely chance to be assigned to each IV level (in a b/w subjects design) or to each treatment order (w/in subjects design)

Random Assignment

Every subject has an equal chance to be assigned to each treatment or treatment order

random assignment

once the participant for the study have been chosen, random assignment occurs when each participant has an equally likely chance to be assigned to each IV level or to each treatment order

Random Assignment

Participants are randomly assigned to levels of the IV.

Random assignment

Once the participants for the experiment have been selected, random assignment occurs when each participant has an equally likely chance to be assinged to each IV level

random assignment

once the participants for the experiment have been chosen random assignment occurs when each participant has an equally likely chance to be assigned to each IV level

Random Assignment

once the participant for the experiment has been chose, random assignment occurs when each participant has an equally likely chance to be assigned to each IV level (in a subjects design) or to each treatment order (in a within subject design)

Random Assignment

once the participants for the experiment have been chosen, random assignment occurs when each participant has an equally likely chance to be assigned to each IV level (in a between subjects design) or to each treatment order (in a within subjects design)

random assignment

Once the participants for an experiment have been chosen, random assignment occurs when each participant has an equally likely chance to be assigned to each independent variable level (in a between subjects design) or to each treatment order (in a within subject design).

Random Assignment

once the participants for an experiment have been chosen. It occurs when each participants has an equally likely chance to be assign to each IV level (in Between) or to each treatment order (in Within)

carry over effects

occur when effects of one treatment persist when another treatment is introduced

Carry-Over Effects

the influence that is particular treatment or task has on performance in a subsequent treatment or task that follows it

Carry-over Effects

Occur when the treatment of one treatment persists when another one is introduced.

Carry-Over Effects

Effects of one treatment persist when another treatment is introduced

Carry-over effect

occur when effects of one treatment persists when another treatment is introduced

(e.g. disease is cured)

Carry-over Effects

Occur when the effects of one treatment persists when another is introduced.

Used to compare the mean of one sample of data to a standard value.

factors that increase type II errors

nuisance variables, floor/ceiling effects, narrow range of IV

factors that increase type II errors

nuisance variables or any variable other than IV that can have an effect on the DV.

sum of squares

SS=Σ(x-xbar)^2

Sum of squares

SS=∑(X-M)²

Sum of Squares

SUM: (x-xbar)^2

Sum of Squares

SUM: (x-meanx)^{2}

^{}

sum of squares

sigma(X-Xbar)squared

levels of the IV

specific values of the IV that the researcher chooses in an experiment

Levels of the IV

values the researches chooses

Levels of the IV

The specific values of the independent variable that the researcher chooses to use in a study

Levels of (IV)

the specific values of the IV that a researcher chooses to use in the study

Levels of the IV

The specific values of the IV the researcher chooses to use. (mg of dosage, red, blue)

between subjects design

research design in which each subject receives one 1 level of the IV

within subjects design

research design in which each subject recieves all levels of the IV

Between subjects design

A research design in which each subject is assigned to ONE level if the IV

Within Subjects design

Research design in which each subject is assinged to ALL levels of the IV

Between Subject Design

Each subject is assigned to one level of the IV

Within Subjects Design

Each subject is assigned to all levels

Between Subjects Design

Research designs in which each subject is assigned to one level of the independent variable.

Within Subjects design

Research designs in which subjects receives all independent variable levels

quasi experiment

research design in which researcher manipulates IV but fails to have random assignment

Quasi Experiment

Experiment without random assignment

Quasi-Experiment

Researcher manipulates the IV but fails to have random assignment

Quasi-Experiment

research technique in which the researcher manipulates the IV, but which fails to have random assignment

Quasi-Experiment

a research technique in which the researcher manipulates the IV, but fials to have random assignment

Quasi-Experiment

a research techniques, an experiment allows the researcher to infer that there is a causal relationship between the IV and DV the IV, but which fails to have random assignment

Quasi-Experiment

A research technique in which the researcher manipulates the independent variable but which fails to have random assignment.

frequency distribution

graph showing # of times each score occurred in data set

frequency distribution

a graph showing the number of times each score occurred in a data set

Frequency Distribution

Express how often a score occurs in a set of data

Frequency Distribution

graph showing number of times each score occured

Frequency distribution

a graph showing the number of times each score ocured in a data set

absolute zero

occurs when score of zero indicates complete absence of the trait being measured

Absolute Zero

Score of zero indicates complete absence of the trait

absolute zero

occurs when a score of zero indicates complete absence of the trait being measured

(Kelvin vs. Fahrenheit)

Absolute zero

occurs when zero on the scale represents a complete absence of the trait being measured

negatively skewed distribution

tail to left, few extreme low scores, mean<median

in a negatively skewed distribution the mean...

< median

Negatively skewed distribution

distribution with a few extreme LOW scores

Negatively Skewed Distribution

The data mostly fall to the right of the mean and the tail is to the left on the graph

Negatively Skewed Distribution

Few extreme low scores

Negatively Skewed Distribution

a distribution with a few extreme low scores

ex: test scores

negatively skewed distribution

distribution with a few extreme low scores (tail to the left)

Negatively Skewed Distribution

a distribution with a few extreme low scores

Median>Mean

standardized scores

z score, allows comparisons to be made between scores measured on different scales by placing all scores on common scale, z=x-xbar/s

standardized scores

allow comparisons to be made b/w scores measured on different scores by placing all scores on a common scale

standardized scores

Allow comparisons to be made between scores measured on different scales by placing all scores on a common scale.

types of scales of measurement (4)

nominal

ordinal

interval

ratio

types of scales of measurement

1) nominal scale has only the identity property

2)ordinal scale has only the identity and magnitude properties

3)interval scale has only the identity magnitude and equal intervals properties

4)ratio scale has identity, magnitude, equal intervals and absolute zero

Types of Measurement Scales

1. nominal (no numerical value only names)

2. ordinal (categories have a natural ordering)

3. interval (no absolute zero point)

4. ratio (has an absolute zero point)

Types of Scales of Measurement

1. Nominal

2. Ordinal

3. Interval

circumstances when the experimental approach cannot be used

when manipulation of variables is difficult, impossible, or unethical

when random assignment is not available

advantages of within subjects designs

fewer subjects needed to obtain same # of observations

even with equal observations, within subject designs have greater stat. power

advantages of within subjects design

1. fewer subjects needed to obtain the same number of observations

2. even with equal observations with/in subjects design have greater statistical power

Advantages of Within Subject Designs

1. Fewer subjects needed to obtain same number of observations

2. even with = observations, within subjects have greater stat power

Advantages of Within Subjects Design

1. Allows the use of fewer subjects to obtain the same number of observations

2. Allows for greater statistical power than between subjects design

*because of these advantages, it is always best to use a WSD when possible

Advantages of Within Subjects Design

-Fewer subjects; same number of observations

-Greater statistical power

Advantages of Within Subjects Designs

1. Allows the use of fewer subjects to obtain the same number of observations

2. They allow for greater statistical power than between subjects design

Advantages of within subjects design

1. allows the use of fewer subjects to obtain the same number of observations

2. allows for greater statistical power than between subjects design

*preferred over between subjects design

*preferred over between subjects design

take home message for within subject design

when using within subject designs, one must always counterbalance!

if sensitization or carry over would be a problem, between subjects must be used

nuisance variables

any variable other than IV that can influence DV

Nuisance Variables

Any variable other than the IV that can have an effect on the DV

Nuisance Variable

Anything other than the IV that can effect the DV

Nuisance variables

anything other than the IV that can affect the DV

Nuisance Variables

any variable other than the IV that can affect the DV

nuisance variable

Any variable other than the IV that can effect the IV

Independent variable

variable for which the researcher chooses the values

Independent Variable

in an experiment, the researcher chooses values to determine what effect the variable has on the behavior being studied

Independent Variable

Variable for which the researcher chooses the levels

Independent Variable

In an empirical study, the variable for which the researcher chooses values to determine what effect the variable has on the behavior being studied (the dependent variable)

Independent Variable

Researcher determines the value

Independent Variable

Variable for which the researcher determines the value

Independent variable

the variable in an experiment for which the researcher chooses values

Independent variable

The variable in which the researcher chooses to values to be manipulated

Independent Variable

the variable in which the researcher chooses to study

Independent variable

The variable that the researcher chooses the value

problems with within subjects designs

1. practice effects

2. sensitization effect

3. carry over effects

correlation between IV levels (Within subject designs only)

larger correlation gives more power

how to increase stat. power

1. choose IV levels that will max. effect size

2. try to lower variability in data

3. collect more observations (increase sample size)

***3 is most common/effective way

Floor/Ceiling effects

values of DV are so low (floor) or so high (ceiling) that they are unlikely to be affected by IV (causes you to find NO effects)

Floor and Ceiling Effect

Occur when the values of the DV are so low (floor) or so high (ceiling) that they are unlikely to be affected by the IV

floor and ceiling effect

occurs when the values of the DV are so low or so high that they are unlikely to be affected by the IV

Floor/Ceiling Effects

values of the DV are so low or so high that they are unlikely to be affected by the IV

Floor and Ceiling Effect

Values of the DV are so extreme that they are unlikely to be affected by the IV

Ceiling and Floor Effects

the values of the DV are so low (floor) or high (ceiling), that they are likely to be affected by the IV

Floor & Ceiling Effects

occur when the values of the DV are so low (floor effect) or so high (ceiling effect) that the DV is unlikely to be effected by the IV

Floor and Ceiling Effects

those that occur when values of the DV are so low(floor) or so high(ceiling) that the DV is unlikely to be affected by the IV.

Practice Effect and Solution

occur when subjects performance on experimental task changes for better or worse due to experience with task

**Counterbalancing

Narrow range of IV

occurs when values of the IV levels are so similar that effects can't be distinguished

narrow range of the IV

occurs when the values of the IV levels are so similar that their effects cannot be distiguished

Narrow range of the IV

occurs when levels of the IV are so similar that their effects on the DV cannot be distinuished

narrow range of the IV

when levels of the IV are so similar that their effects on the DV cannot be distinguished

Narrow Range of the IV

Occurs when the levels of the IV are so similar that their effects on the DV cannot be distinguished

Can't prove null hypothesis

it is impossible to prove that an IV has no effect on DV

Can't Prove the Null Hypothesis

it is impossible to prove that an IV has no effect on DV- never justified

Can't Prove the Null Hypothesis

You cannot prove that an IV has no effect on a DV

when to use a test of single sample mean

when comparing one sample mean to standard value for which sample is unavailable

WHEN TO USE: Test of a Single Sample Mean

Comparing the mean of a sample to a single standard value

Test of a single sample mean

single sample to standard sample

modus tollens

if p is true, then q is true. Q is not true, therefore p is not true

-both valid and useful to science, used to prove theories false

Modus Tollens

A valid argument form and rule of inference. If P implies Q, and Q is not true, then P cannot be true.

modus tollens

if p is true then q is true if q is not true therefore p is not true

Modus Tollens

if A is true, B is true; but B is false; therefore A is false

Modus tollens

"If p is true, then q is trueQ is not true

Therefore, p is not true"

Both valid and useful to science because it can be used to prove theories false

Therefore, p is not true"

Both valid and useful to science because it can be used to prove theories false

Modus Tollens

If P is true, then Q is true

Q is not True

Therefore, P is not true.

-Both valid and useful in science. Science works by using it to prove other theories false.

Q is not True

Therefore, P is not true.

-Both valid and useful in science. Science works by using it to prove other theories false.

Modus Tollens

If P is true then Q is true; Q is not true therefore P is not true.

Both valid and useful to science.

Science works by using modus tollens to prove theories false.

Both valid and useful to science.

Science works by using modus tollens to prove theories false.

Modus Tollens

If p is true, q is true.

q is not true, therefore p is not true.

Valid and useful to science

alpha

probability of making a type I error given an experiment found an effect of IV on DV

Alpha

the probability level for accepting the difference between the means is not due to number of variables. Usually alpha is set to .05

Alpha

The probability level for accepting that the value of a test statistic was not obtained simply by chance

Alpha

Probability of making a Type One Error

Alpha

the probability of making a Type 1 error given that our experiment failed to find an effect of the IV on the DV

Alpha

the probability level for accepting that a difference between the means is not due to nuisance variables alpha is usually set to 5%.

Alpha

Probability of making a Type I error given that our experiment found an effect of the IV on the DV. Alpha is usually set .05

alpha

the probability of making a type one error given that our experiment found an effect of the IV on the DV. (.05)

Alpha

the probability of making a Type 1 error given that our experiment found an effect of the IV on the DV

usually set to .05

usually set to .05

beta

probability of making a type II error given that our experiment failed to find an effect of IV on DV

Beta

Probability of committing Type II error

Beta

Probability of making a Type Two Error

Beta

the probability of makng a Type 2 error given that our experiment failed to find an effect of the IV on the DV

Beta

the probability of making a Type 2 Error given that our experiment failed to find an effect of the IV on DV

statistical power

probability that a given experiment will find an effect of IV on DV given that effect exists (1-beta)

statistical power

the probability that a given experiment will find an effect on the IV on the DV given that an effect exist

statistical power = 1- beta

Statistical Power

the probability that a given experiment will find an effect of the IV on the DV given that an effect exists

Statistical Power

The probability that the experiment will find an effect of the IV on the DV if an effect exists

LOWERS type II errors

Statistical Power

the probability that your experiment will find an effect of the IV on the DV if the effect exists

Is desirable in an experiment because it prevents Type ll errors

statistical power

the probability that an experiment will find an effect of the IV on the DV given that an effect exists.

**it is desirable for an experiment to have high power because it lowers type 2 errors

Affirming the consequent

if p is true, then q is true.

q is true, therefore p is true.

- may be outlying reasons but increases belief little by little

affirming the consequent

if p is true then q is true, q is true therefore p is true

Affirming the Consequent

If p is true, then q is true.

Q is true.

Therefore, p is true.

Therefore, p is true.

Invalid form of logic.

Affirming the Consequent

1)If p is true, then q is true 2)q is true 3)Therefore, p is true

Invalid form of argument

Invalid form of argument

Affirming the consequent

"If p is true, q is trueQ is true

Therefore, p is true"

Invalid form of argument because a correct prediction does not prove the theory to be true (it is logically impossible to prove a theory true)

Therefore, p is true"

Invalid form of argument because a correct prediction does not prove the theory to be true (it is logically impossible to prove a theory true)

Affirming the Consequent

If P is true, then Q is true

Q is true

Therefore, P is true

-invalid form of logic because it is logically impossible to prove that a theory is true.

Q is true

Therefore, P is true

-invalid form of logic because it is logically impossible to prove that a theory is true.

Affirming the consequent

If p is true then q is true

q is true. Therefore p is true

Logically impossible to prove a theory is true. However, if a theory makes a prediction and the prediction is true, then it increases the belief in the theory

Affirming the consequent

if P is true then Q is true: Q is true, therefore P is true.

Affirming the consequent is an invalid form of argument. Because it is invalid, this means it is logically impossible to prove that a theory is true.

However, if a theory makes a prediction, and the prediction is true, then it increases our belief in the theory a little bit.

Affirming the consequent is an invalid form of argument. Because it is invalid, this means it is logically impossible to prove that a theory is true.

However, if a theory makes a prediction, and the prediction is true, then it increases our belief in the theory a little bit.

stat power is desirable. why?

high stat power desirable because it prevents type II errors

effect size

the magnitude of relationship between IV and DV

-larger effect size = more power

Effect Size

Magnitude of Relationship between IV and DV

effect size

the magnitude of the effect the IV has on the DV

modus ponens

if p is true, then q is true. p is true, therefore q is true.

valid argument but not useful to science

modus ponens

if p is true then q is true. p is true therefore q is true

Modus Ponens

A --> B

A

therefore B

Modus ponens

"If p is true, then a is trueP is true

Therefore, q is true"

Valid form of argument but not useful to science because it assumes theory is true

Therefore, q is true"

Valid form of argument but not useful to science because it assumes theory is true

Modus Ponens

if P is true then Q is true

P is true

Therefore, Q is true.

-if the first two statements are true. then the conclusion must be true.

- it is a valid argument but not useful to science because it assumes that the theory is true.

P is true

Therefore, Q is true.

-if the first two statements are true. then the conclusion must be true.

- it is a valid argument but not useful to science because it assumes that the theory is true.

modus ponens

If P is true then Q is true; P is true, therefore Q is true.

Modus ponens is a valid argument, but is not useful to science because it assumes the theory is true.

Modus ponens is a valid argument, but is not useful to science because it assumes the theory is true.

ratio scale

Has ALL properties.

All math is meaningful.

example: weight in lbs, temp in K, # correct on exam

Ratio scale

has the identity, magnitude, equal intervals, and absolute zero identities

ex: lbs, exam scores, kelvin

factors determining stat power (5)

1. alpha level

2. effect size

3. variability in DV

4. sample size

5. correlation between IV levels (within subject only)

Factors determining stat power

Correlation between IV levels (within only)

Alpha

Variability in DV

Effect Size

Sample Size

ordinal scale

only identity and magnitude

No math is meaningful.

example) bball rankings

Ordinal scale

has only the identity and magnitude properties

ex: basketball rankings

Identity

occurs when different entities receive different scores (weights)

Identity

Different entities receive different scores

identity

occurs when different entities receive different scores

water bottle = 5 lbs does not equal a feather

Identity

occurs when different entities receive different scores.

magnitude

occurs when the ordering of values on the scale reflect the ordering of the trait being measured

Magnitude

Ordering of the values reflects the ordering of the trait being measured

magnitude

occurs when the ordering of the values reflects the ordering of the trait being measured

(6 lbs > 1 lb)

Magnitude

equal intervals

occurs when a difference of one on the scale reps the same amount of the trait being measured everywhere on the scale

Equal Intervals

Difference of 1 on the scale means the same everywhere

equal intervals

occurs when a difference of one on the scale means the same amount everywhere on the scale

(Bball league: NBA>>> 3rd grade>2nd grade)

equal intervals

occurs when a difference of 1 on the scale represents the same amount of the trait being measured everywhere on the scale.

nominal scale

only Identity.

no math is meaningful

example: zip codes, jersey numbers

Nominal Scales

only have the identity property

Ex: football jersey numbers, zip codes

alpha level

larger alpha gives more power, but can't increase alpha because it would increase type I errors

Alpha level

The probability level for accepting that the value of a test statistic was not obtained simply by chance.

alpha level

larger alpha gives more power - however we cannot increase alpha because it is the probability of making a type 1 error

interval scale

only identity, magnitude, equal intervals.

addition and subtraction meaningful.

example: degrees F or C

Interval Scale

has only the identity, magnitude, and equal interval properties

ex: fahrenheit, celcius

factors producing Type I errors

1. regression to the mean

2. confounds

Factors producing Type I errors

1. Regression to the mean – tendency for extreme values of a variable to fall closer to the group mean when retested.

2. Confounds – a nuisance variable that varies non-randomly with the IV

Confounds can produce both type 1 and 2 errors.

2. Confounds – a nuisance variable that varies non-randomly with the IV

Confounds can produce both type 1 and 2 errors.

regression to the mean

tendency of the extreme value to fall closer to group mean when retested

Regression of the Mean

The tendency for extreme values of a variable to fall closer to the group mean when re-tested

Regression to Mean

tendency of extreme scores to fall closer to group mean when retested

regression to the mean

tendency for extreme values of a variable to fall close to the group mean when re-tested

Regression to the mean

The tendency for extreme values of a variable to fall closer to the group mean when retested

control group

group of subjects in a between subject design that receives a treatment we know is ineffective at manipulating the DV --> Placebo

Control Group

A group of subjects in a between subjects design that receives a treatment we know is ineffective at manipulating the DV

Control Group

Receives a treatment we know is ineffective- allows us to see if treatment is really effective

Control Group

a group of subjects in a between subjects design that receives a treatment we know is ineffective at changing the DV

Makes sure effects are not due to regression to the mean

control group

group of subjects, in a between subjects design, that receives a treatment we know is ineffective at changing the DV

Logic of experimentation ruined by counfounds

the logic of experimentation breaks down if you have a confound because it could either be the confound or the IV responsible for the change in the DV

Types of Counfounds

a. confounds due to subject assignment (subs at different IV levels differ on some variable prior to the manipulation of IV)

b. confounds due to manipulation of IV (unanticipated changes that accompany the changes of IV)

External Validity

the extent to which the findings of a study can be applied outside the research situation

External validity

extent to which the results of experiment can be generalized

External Validity

The extent to which the results of a study can be applied outside the research situation

External Validity

The extent to which the results of your research can be applied outside the research situation

External validity

the extent in which your research can be applied outside of the research situation

External Validity

the extent to which the results of your research can be applied outside the research

Confound

nuisance variable that varies non randomly with the IV

Confound

A nuisance variable that varies non-randomly with the IV. Can produce Type I and Type II error.

Confound

a nuisance variable that varies non randomly with the IV

Produce both Type l and ll Errors

Confound

A nuisnace varialbe that varies nonrandomly with the IV

Confound

a nuisance variable that varies non-randomly with the IV

Confounds can produce both type 1 and 2 errors.

Confounds can produce both type 1 and 2 errors.

Random Factor

an IV whose levels were chosen randomly from a population of possible values (May be generalized to ALL values in population)

Random Factor

An IV whose levels were chosen randomly from a population of possible values.

Ex. Vitamin C on math scores 340, 678, 932

Ex. Vitamin C on math scores 340, 678, 932

random factor

an IV whose levels were chosen randomly from a population of possible values

Random Factor

A level of the IV chosen randomly from population of levels

random factor

an iv whose levels were chosen rendomly from a population of possible values

Random Factor

An IV whose levels were chosen randomly.

If a reliable effect of a random factor is found, the researcher is permitted to generalize the effect to all the levels in the pop;ulatio

Random factor

IV levels that are chosen randomly

Fixed Factor

IV whose levels were chosen non randomly (does not allow generalization to other levels)

Fixed Factor

An IV whose levels were chosen non-randomly.

Ex. Vitamin C on math scores; 0, 500, 1000

Ex. Vitamin C on math scores; 0, 500, 1000

fixed factor

an Iv whose levels were chosen non-randomly

Fixed Factor

Level of the IV that was not chosen randomly from a population of levels

Fixed Factor

an IV whose levels were chosen non-randomly - cannot be generalized outside levels tested

Fixed Factor

An IV whose levels were chosen non-randomly

Fixed factors does not allow generalization to other levels

Demand Characteristics

Aspects of study that indicate to subjects how they are expected to respond

Demand characteristics

Aspects of a study that indicate to a subject as to how they should respond

Between Subjects Experiment

Experiments in which each subject receives only one level of the independent variable

Subjects agree to participate; subjects are not coerced into participation; subject should be fully informed about purpose of experiment when deciding whether to participate; in deception experiments, subjects should be fully informed about true nature of experiment at conclusion; gains to science from experiment should outweigh harm done to subjects; subjects should be warned of any potential harmful effects of experiment; subject data should be kept confidential; all reports of the data must be accurate

Ethical principles to consider when conducting psychological research (4)

Levels of the independent variable

the actual values of the independent variable that the experimenter chooses to use in an experimental study

levels of independent variables

the specific values of the IV that a researcher chooses to use in an experiment

Levels of the Independent Variable

The actual values of the independent variable that the experimenter chooses to use in an empirical study

Levels of the Independent Variable

Specific values of the IV

levels of the independent variable

specific values of the IV that the researcher chooses to use in a study

Levels of the independent variable

the specific values of the IV that a researcher chooses to use in study

Levels of the independent variable

The specific values of the independent variable that a researcher chooses to use in an experiment.

Levels of the independent variable

specific values of the IV that the researcher chooses

Factors Determining Statistical Power

1. Alpha Level

2. Effect Size

3. Variability of the DV

4. Sample Size

5. Correlation between the IV levels

2. Effect Size

3. Variability of the DV

4. Sample Size

5. Correlation between the IV levels

Factors determining statistical power

C-correlational between the IV

A-alpha

V-variability in the DV

E-effect size

S-sample size

Factors determining statistical power (5)

Correlation between IV levels (within only)

Alpha

Variability in DV

Effect Size

Sample Size

5 Factors determining Statistical Power

1. Correlation between IV levels

2. Alpha Size

3. Variability in the DV

4. Effect Size

5. Sample Size

Factors Determining Statistical Power

1. alpha level

2. Effect size

3. Variability in the DV

4. Sample Size

5. In Within Subjects Designs: Correlation between the IV levels

2. Effect size

3. Variability in the DV

4. Sample Size

5. In Within Subjects Designs: Correlation between the IV levels

5 factors determining statistical power

S.ample size

Alpha level

Variability in the DV

Effect size

Correlation between IV levels

(SAVE C)

Experimenter Expectancy Effect (Rosenthal Effect)

A demand characteristic that occurs when subjects change their behavior due to unintentional cues from the researcher

Ex. Clever horse

Ex. Clever horse

Experimenter expectancy effect (Rosenthal effect)

a demand characteristic that occurs when subjects change their behaviors due to unintentional cues from the researcher

How confounds break logic of experiment

Logic breaks down if you have a confound because it could either be the confound or the IV that is responsible for any changes in the DV

Types of Confounds

Confound due to subject Assignment

types of confounds

1. during subjects assignment: when subjects at the different IV levels differ on some variable prior to IV manipulation

2. confounds due to the manipulation of the IV: occurs when additional unanticipated changes accompany IV manipulation

confound types

a. due to subject assignment

b. due to IV manipulation

Types of Confounds

1. Confounds due to subject assignment

2. Confounds due to manipulation of the IV

Types of Confounds

1. Confounds due to subject assignment: these occur when the subjects at the different IV levels differ on some variable prior to IV manipulation

2. Confounds due to manipulations of the IV

types of confounds

1. Confounds due to subjects assignment: occurs when the subjects at the different IV levels differ on the same variable prior to the IV manipulation.

Random assignment is the key to overcoming subject assignment confounds.

Random assignment is the key to overcoming subject assignment confounds.

Confound due to subject assignment

These occur when subjects at the different IV levels differ on some variable prior to IV manipulation

Investigation

A research technique that does not allow a researcher to infer causation

Investigation

A research technique that does not allow the researcher to infer causation (opposite if experiment)

Investigation

Not allowed to say causation

Investigation

a research technique that doesn't allow the researcher to infur causation

Modus Pones

P = Theory

Q = Hypothesis

IF P is true then Q is true

Valid argument but not useful to science because it assumes the theory is true

Q = Hypothesis

IF P is true then Q is true

Valid argument but not useful to science because it assumes the theory is true

Placebo Effect

demand characteristic that occurs when subjects change their behavior as a result of their expectation that change should occur

Placebo Effect

deman characteristic that occurs when subjects change their behavior as a result of their expectation that change should occur

Placebo Effect

A phenomenon whereby the expectation that a treatment will have an effect may produce the effect, even though the treatment has no therapeutic effect

Placebo Effect

a demand characteristic that occurs when subjects change their behavior because of their expectation that change should occur

Placebo effect

subjects change their behavior due to their own expectation that change should occur

Placebo Effect

demand characteristic that occurs when subjects change their behavior because of their expectation that change effect

Ways to overcome Placebo and Rosenthal Effect

1. Single Blind Study (Subject doesn't know which treatment they received) (Placebo Effect Solution)

2. Double Blind Study (Neither subject nor person administering the treatments knows which treatment the subject receives) (Solutions to both the Placebo and Rosenthal effect)

ways to overcome placebo and rosenthal effects

1. single blind study

2. double blind study

Ways to overcome Placebo and Rosenthal effects

1. single blind experiment: researcher but NOT the subjects knows which IV level the subject received

2. double blind experiment: neither subject nor researcher knows which IV level the subject receives

Ways to overcome Placebo and Rosenthal effects

- Single Blind Experiment: Researcher but NOT the subjects, knows which IV level the subject receieved
- Double Blind Experiment: Neither subjects nor researcher knows which IV level the subject receives

Ways to Overcome the Rosenthal and Placebo Effects

-Single-Blind Experiment

-Double-Blind Experiment

-Double-Blind Experiment

Ways to overcome Placebo and Rosenthal effects

1. Single blind experiment: an experiment in which the researcher but not the subject knows which condition the subject received

( overcome Placebo)

2. double blind experiment: an experiment in which neither the subject nor person giving the treatment to the subjects know which conditon the subject resolved

(overcome both Placebo and Rosenthal)

Ways to overcome Placebo and Rosenthal effects

1. single blind experiment-an experiment in which the researcher not the subject knows which condition the subject received

* this overcomes Placebo effect but NOT Rosenthal effect

Ways of Overcoming Placebo & Rosenthal Effects

1. Single Blind Study

2. Double Blind Study

3. Hawthorne Effect

Ways to overcome placebo and rosenthal effects

1. single blind study - researchers knows IV level - solves placebo effect

2. double blind - neither giver nor receiver knows IV level - solves both placebo and rosenthal effects

Ways to overcome Placebo and Rosenthal Effects

1. Single Blind Study- a study in which only the researcher knows what IV level the subject is receiving

2. Double Blind Study- a study in which neither the researcher or subject knows what IV they are receiving

2. Double Blind Study- a study in which neither the researcher or subject knows what IV they are receiving

Hawthorne Effect

demand characteristic that occurs when subjects change their behavior because they know they are being observed

Hawthorne Effect

The subject behaves differently when they know they are being studied

hawthorne effect

demand characteristic that occurs when subjects change their behavior bc they know they are being observed

Hawthorne effect

a demand characteristic that occurs when subjects change there behavior because they know they are being deserved

Hawthorne effect

subjects change their behavior because they are being watched

Novelty Effect

Occurs when the DV is influenced by the IV only because the IV is something new

novelty effect

occurs when the dv is influenced by the IV only bc the IV is something new

Novelty Effect

IV affects DV only because IV is something new

Novelty Effect

DV is influenced only because the IV is something new

Novelty Effect

Occurs when an IV is effective only b/c the IV is something new

Novelty effect

Occurs when an IV is effected only because it is something new.

when can the experimental approach not be used

1. difficult or impossible to manipulate the IV.

2. unethical to manipulate the IV.

3. Random assignment cannot be done.

when can the experimental approach not be used??

1. difficult/impossible to manipulate variables

2. unethical to manipulate variables

3. random assignment cannot be done

Experimental approach cannot be used when

1. Difficult/impossible to manipulate the IV (age, sex, SES)2. Unethical to manipulate IV (smoking, abuse)

3. Subjects cannot be randomly assigned (due to previous circumstances (fourth graders))

3. Subjects cannot be randomly assigned (due to previous circumstances (fourth graders))

advantage of the experimental approach

in contrast to all other research methods an experiment allows a researcher to infer a causal relationship b/w the IV and the DV

Advantage of experimental approach

Allows cause to be determined

Advantage of the experimental approach

allows a researcher to infer a causal relationship between the IV and the DV

advantage of the experimental approach

allows the researcher to infer causation

Advantage of experimental approach

An experiment allows the researcher to infer a causal relationship between the IV and DV

Advantage of the Experimental Approach

Allows researcher to infer a causal relationship

types of distributions

Normal: a symmetric, bell-shaped distribution, a positively skewed distribution

compare and contrast the mean and median

the mean is appreciated for more by extreme score than is the median. as such the median is often a more useful descriptive statistic than the median is skewed distributions

Compare and contrast mean and median

1. effects of extreme scores - mean is more affected by extreme scores

2. consistency across repeated samples - mean provides more consistent results across repeated samples

3. ease of calculations

sum of squares formula

2

Ex-(Ex)

____

N

Sum of Squares Formula

SS = Sum of (x - xbar)^2

how to increase the statistical power

collect more observations

How to increase Statistical Power

-IV levels maximizing the effect size

-Lowering variability in the DV

-Increase sample size

Increasing statistical power

1. Choose IV levels that will maximize effect size

2. Try to lower the variability of the DV

3. Increase sample size (most common)

how is the logic of experimentation ruined by confounds?

in experimentation we presume that the IV levels differ only in the Iv manipulations the logic breaks down if you have a confound b/c the IV that is responsible for any changes in the DV

how is the logic of experimentation ruined by confounds?

How is the logic of experimentation ruined by confounds?

If you have a confound in your experiment the logic of experimentation breaks down because it could either be the confound or the IV that caused any changes in the DV

how is the logic of experimentation ruined by confounds

it is impossible to know if it was the confound or the IV that had an effect on the DV

experimenter expectancy effect

a demand characteristic that occurs when a subject changes their behavior due to unintentional cues from the researcher

Sample a subset of population (group selected to participate in study).

Negative Distribution

Tail to left, few low scores, Mean<Median

Negative distribution

Distribution that has few low scores

Mean Vs. Median

Mean is affected by extreme scores.

Median often more useful descriptive stat than mean.

median

the score that falls exactly in the center of a distribution of scores

Median

The middle number in a set of chronologically ordered data

more useful descriptive statistic in skewed distributions

* mean is far more affected by extreme scores than the median is, so median is often a more useful descriptive stat in skewed distribution

* mean: always closest to the tail

* mean provides more consistent results across repeated samples than the median

* mean is usually easier to calculate than the median

mean vs. median

Mean vs Median

Mean: average value of a group of numbers

Median- the middle number in a group of numbers arranged from low to high

Mean is more affected by extreme scores

Practice Effects

occur when subjects performance on experimental task changes for better or worse due to experience with task

Practice effects

performance improving over time

Practice Effect

A carryover effect in which participants perform better on a task in later conditions because they have had a chance to practice

practice effects

occur when a subjects performance on the experimental task changes either for the better or worse as a result of experience with the task

Practice Effects

A subject's performance chances as result of experience

Practice Effects

Occur when the subject's performance on the experimental task changes (either for better or worse) as a result of experience with the test

Practice effects

occur when the subjects performance on the experimental task changes as a result of experience with the task

Function of Latin Square

Method of counterbalancing. Assigns treatment orders in grid.

function of a latin square

has as many columns and as many rows as there are IV levels. Rows represent subjects and columns represent treatment orders. Treatments are placed in every cell such that every treatment appears only once in every column and every row

function of Latin Square

Method of counterbalancing. Assigns treatment orders in grid. (solves practice effects)

Function of a Latin Square

What is the function of a Latin Square?

function: solve practice effects (counterbalancing)

Figure: Has as many columns and rows as there are IV levels.

Function of a Latin Square

used to control for Practice Effects

Rosenthal Effect

demand characteristic that occurs when subjects change behavior due to unintentional cues from researcher

Rosenthal Effect

A demand characteristic that occurs when subjects change their behavior due to unintentional cues form the researcher.

When to use within sub t test

within sub design comparing means

When to use between subject one way ANOVA

used to compare means from between sub design that have one IV with more than 2 levels

When to use within subjects one way ANOVA

used to compare means from within subjects designs that have one IV with more than 2 levels

Between subjects One-Way ANOVA

1 IV, more than 2 IV levels

Why do an ANOVA instead of multiple t tests?

1. multiple t tests are more work

2. Multiple t tests lead to type I error rate inflations

Why do an ANOVA instead of multiple t-tests? 2

1. More Work

2. Inflates the probability of type I error

Why do an ANOVA instead of Multiple t-tests

1. Multiple t-tests would be more work

2. Multiple t-test leads to type 1 error rate inflammation

Why do an ANOVA instead of multiple t-tests?

1. More Work

2. Inflates type I errors

Why do an ANOVA instead of multiple t-tests?

-Multiple t-tests are more work

-Multiple t-tests inflate the Type 1 Error Rate

-Multiple t-tests inflate the Type 1 Error Rate

Why do an ANOVA instead of multiple t-tests?

1. They are more work

2. leads to type I error rate inflation

MStreat measures?

measure of the variability in DV due to the IV and error

What MStreat measures

Variability in the DV due to the IV and error

MSw measures what?

the variability in the DV due to ONLY error

F measures what?

how much greater MStreat is than MSw and thus whether the IV is affecting the DV

What F measures

How much greater MStreat is than MSwithin and thus whether the IV is having an effect on the DV

Factorial Design

a design in which there is more than one IV, and every IV level is present at all levels of the other other IV's

Factorial Design

has more than one IV and every IV is present at all levels of the other IVs

Main Effect

The independent effect of one IV on the DV in the Factorial Design

Main Effect

Independent effect of one IV on the DV in a factoral design

Main Effect

Independent effect of one IV on the DV in a facotiral design

Interaction

occurs when effects of one IV on the DV change depending on the level of another IV

Interaction

when the effect of one IV on a DV change depending on the level of another IV

Interaction

Occurs when the effect of one IV on the DV changes depending on the level of another IV

Interaction

Interaction

the effect of an IV on the DV changes depending on the level of another IV

Between Subjects Factorial ANOVA

When to use: used to compare the means from between subjects designs that have 2 IV's

Between subjects Factorial ANOVA

2 IV's

when to use a correlational design

when manipulating the IV would be difficult, impossible, or unethical

Time series Design

a class of quasi-experiments in which the performance of a single group of subjects is measured both before and after some experimental manipulation

time series design

a research design that involves measurements made over some period of time

time series design

A sequence of data points measured typically at successive points in time apaced at uniform time intervals

Time series design

A within subjects design in which the performance of a single group of subjects is measured both before and after some experimental treatment

Time Series design

A within subjects desing in which the performance of a single group of subjects is measured both before and after some experimental treatment

Time Series Design

within subjects design inwhich the performance of a single root of subjects is measured both before and after some experimental treatment

Time-series designs

A within subjects design which examines the DV over extended periods of time both before and after the IV is implemented.

Time Series Design

a design in which there is more than one IV and evrey IV level is present at all levels of the other IV's

Time Series Design

research designs in which a single group of subjects is tested on the DV prior to some IV manipulation and also treated on the DV afterwards

single group, pretest posttest design

a time series design in which only one measurement of the DV is made prior to IV manipulation and only one measurement is made afterwards

Single group, pretest-posttest design

a design in which a single group of participants takes a pretest, then receives some treatment, and finally takes a posttest;

problem=lack of comparison group

Single Group Pretest-Posttest Design

A time series design in which only one measurment of the DV is made prior to the IV manipulation and only one measurement is made after words

Single group, pretest-posttest design

A time series design in which only one measurement of the dependent variable is made before treatment and only one measurement of the dependent variable is made after treatment

Single Group, Pretest-posttest design

Time series designs in which only one test of DV is given before and after the treatment

Single Group, Pretest-Posttest design

a time series design in which only one test of the DV is made before treatment and only one test is made after

***problems: Practice effects, Placebo effects, regression to the mean are all left uncontrolled

***problems: Practice effects, Placebo effects, regression to the mean are all left uncontrolled

Problems with single group pretest posttest designs

regression to the mean, practice effects, placebo effects

Interrupted Time Series Design

a time series design in which more than one measurement of the DV is made prior to IV manipulation and more than one measurement is made afterwards

Interrupted time series design

a quasi-experimental study that measures people repeatedly on a dependent variable before, during, and after the "interruption" caused by some event

Interrupted Time Series Design

A time series design in which more than one measurment of the DV is made prior to IV manipulation and more than one measurmentis made after

Interrupted time series design

A time series design in which several measurements of the dependent variable are taken prior to independent variable manipulation and several measurements are made afterwards

Interrupted Time Series design

a time series design in which several measurements of the DV are taken prior to IV manipulation and several measurements are made afterwards

Problems: placebo effects, confounds

Interrupted Time Series Design

More than one measurement of DV is made before IV manipulation and more than one measurement is made afterward

Advantages and Disadvantages of interrupted time series designs

advantage: controls for regression to mean and possibly for practice effects

disadvantage: placebo effects and possible confounds with IV manipulation can still explain results

advantages and disadvantages of interrupted time series designs

advantage: controls for regression to mean and possible for practice effects

disadvantage: placebo effects and possible confounds with IV manipulation can still explain results

Non-equivalent before-after design:

Quasi Experiment, but not time series, a design in which a pretest and post test (before and after IV manipulation) are given to two non-randomly assigned groups of subjects. The difference between the pre-test and posttest serves as the DV

Non-equivalent Before After Design

a quasi experiment in which a pretest and a posttest are given to two non randomly assigned groups of subjects. The difference between pre-test and post test scores serves as the DV

Non-Equivalent Before After Design

A design in which a pretest (before IV manipulation) and a post test (after IV manipulation) are given to two non randomly assigned groups of subjects. The differences between the pretest and post test serves as the DV

Non equivalent before after design

A design in which a pre-test (before independent variable manipulation) and a post-test (after independent variable manipulation) are given to two non-randomly assigned groups of subjects. The difference between the pre-test and the post-test serves as the dependent variable

non-equivalent Before-after design

a design in which a pretest (before IV manipulation) and a post test (after IV manipulation) are given to two non-randomly assigned groups of subjects

The difference between the pretest and post test serves as the DV

Problem: floor and ceiling effects

non-equivalent Before-after design

a quasi-experiment design in which a pretest (before IV manipulation) and a post test (after IV manipulation) are given to two non-randomly assigned groups of subjects

The difference between the pretest and post test serves as the DV

Problem: floor and ceiling effects

Non-equivalent before after design

a design in which a pretest (before) IV manipulation and a post test (after IV manipulation) are given to non-randomly assigned groups of subjects. The difference between the post test and pretest scores serves as the DV.

Problem with non equivalent before after designs

an essential assumption of this design is that all of the items on the DV must be equally difficult (that is almost never the case)

Ex Post Facto Design

a design in which the researcher uses archival data to study an event that occurred in the past

ex post facto design

a between-subjects design with at least 2 groups of participants that uses a subject variable or that creates nonequivalent groups

ex post facto design

a non experimental research technique in which preexisting groups are compared on some dependent variable

advantage and disadvantage of participant observation

allows access to groups who are not observable otherwise (advantage), the participation of the researcher may alter the groups behavior because the researcher could influence the group (disadvantage)

Positive aspects of naturalistic Observation (6)

1. often allows data collection impossible to obtain otherwise

2. high in ex. valid.

3. can be used to verify ex. valid. of experiment

4. relatively uncomplicated

5. excellent starting point for generating research ideas testable by other methods

6. studies behavior as it unfolds over time

negative aspects of naturalistic observation (5)

1. practical problems with data gathering

*** diff. to observe/record @ same time

*** can be very time consuming

2. observing can change behavior (don't let them know or habituate)

3. observor bias can influence results

4. likely unethical sometimes

5. cannot be used to determine causation

Correlation Coefficient

a statistic used to express the direction and strength of linear relationships between 2 variables measured on an interval or ratio scale

Correlation coefficient

a statistic used to indicate the direction and the strength of the linear relationship between two variables measured on an interval or ratio scale

Correlation Coefficient

Possible values of a correlation coefficient

+1.00 to -1.00

Possible values of a correlation coefficient

statistics showing the strength of the relationship. (-1.00 to +1.00)

values of a correlation coefficient

varies from +1 (perfect positive relationship) to -1 (meaning a perfect negative relationship

possible values for correlation coefficient

+1 = perfect positive relationship

-1 = perfect negative relationship

Possible values of a Correlation Coefficient

-1 to +1

Positive Relationship

as one variable increases, then the other variable also tends to increase

Positive relationship

Positive Relationship

occurs when as one variable increases the other variable also tends to increase

Negative Relationship

As one variable increases, the other variable tends to decrease

negative relationship

as one variable increases, the other variable tends to decrease. (the variables move in opposite directions)

Negative relationship

As one variable increases, the other variable tends to decrease. or vice versa

Negative Relationship

occurs when as one variable increases the other variable tends to decrease

Proportion of variance accounted for:

the proportion of the variability in one variable that can be predicted from knowing the values of another

Proportion of Variance Accounted For

The proportion of the variability in one variable that can be predicted for knowing the values of another

-Proportion of Variance Accounted For = r^2

-Percentage of variance accounted For = 100% x r^2

Proportion of variance accounted for

r2

Proportion of variance accounted for

The proportion of the total variance in one variable that can be predicted from knowing the values of the other variable

Proportion of Variance Accounted for

r^2

proportion of variance accounted for

r^2

the proportion of the variance in one variable that can be predicted by knowing the values of the other variable

Proportion of Variance Accounted For: Formula:

r squared

Proportion of variance accounted for (know formula)

the proportion of the variance in one variable that can be predicted from knowing the values of another, not the amount that is caused.

Formula: r2 (r squared)

Formula for proportion of variance accounted for

R^2

Proportion of variance accounted for formula

r2 = pvaf

Proportion of variance accounted for Formula

100% * r^2

% of Variance Accounted For:

100% x r squared

Correlation cannot be used to establish causation because: (2)

1. it is sometimes unclear which variable is cause and which is effect (Directionality Problem)

2. A third, unmeasured variable may be responsible for the relationship (3rd variable problem)

Correlation cannot be used to establish causation because:

1. it is sometimes unclear which variable is cause and which is effect

2. A third, unmeasured variable may be responsible for the relationship

Mathematical Limitations of Correlations (3)

1. Poor at capturing NonLinear Relationships

2. Greatly affected by extreme scores (Alaska population example)

3. Can be LOWERED by Range Restriction

(occurs when there is a floor/ceiling effect on one of the variables in a correlation)

Mathematical limitations of correlations

1. poor at capturing non-linear relationships

-2 variables can be strongly related in a non linear way and produce a low correlation

2. It is greatly affected by extreme scores (outliers)

3. Can be lowered by range restriction

-occurs when there is a floor or ceiling effect on one of the variables in a correlation

Mathematical Limitations of Correlations

1. Poor at capturing non-linear relationships

2. Greatly affected by extreme scores

3. Ca be lowered by range restriction

mathematical limitations of correlations

-poor at capturing non-linear relationships

-greatly affected by extreme scores

-can be lowered by range restriction

Mathematical limitations of correlations (3)

1. Poor at capturing non-linear relationships

-Two variables can have a very strong non-linear relationship and produce a low correlation

2. Greatly affected by extreme scores

- Think of pop. Alaska example

3. Can be lowered by RANGE RESTRICTION

- Occurs when there is a floor or ceiling effect on one of the variables in a correlation

Solution to: Poor at capturing non linear relationships (mathematical limitation of correlation)

always graph data to look for non linear relationships

Solution to: greatly affected by extreme scores (mathematical limitation of correlation)

report correlation with all data included and also with extreme points removed

Solution to: Lowered by range restriction (mathematical limitation of correlation)

if possible, choose variables that do no exhibit floor/ceiling effects

Correlational Research best used in 2 instances

1. to solve practical problems requiring prediction

2. ELIMINATE variables as cause of phenomenon

Single Subject Designs (n=1 designs)

research designs based on testing one subject

Types of Single Subject Designs

1. Case Study

2. Experimental Single Subject Designs

Case Study

Description of behavior or abilities of single (usually exceptional) individual

Case Study Best Used:

to generate hypothesis testable by other methods

case study

the comprehensive description of an individual that focuses on the assessment or description of abnormal behavior or its treatment

Case Study

Is a description of the behavior or abilities of a single (usually exceptional) individual

-Cast studies are best used to generate hypothesis testable by other methods

case study

a description of the behavior or abilities of a single individual

case study best used

to generate hypotheses testable by other methods

Case Study

a description of the behavior or abilities of a single casually exceptional individual

Best used to generate hypotheses testable by other methods

Experimental Single Subject Designs

designs in which the effect of an IV on a DV is studied using one subject (YOU CANNOT DO STATS ANALYSIS ON THESE)

include: reversal, multiple baseline, and multiple element designs

single-subject experimental designs

used to study the behavior change that an individual or group exhibits as a result of some intervention or treatment.

Experimental Single Subject Design

Design in which the effect of the IV and the DV is studied using one subject

-Cannot statically test on a single subject

experimental single subject design

a research design based on testing 1 subject

experimental single subjects designs

designs in which the effect of an IV on a DV is studied using one subject (YOU CANNOT DO STATS ANALYSIS ON THESE) include: reversal, multiple baseline, and multielement

Experimental single subject designs

Designs where the effect of an independent variable on a dependent variable is examined using one subject

Experimental Single Subject Design

designs wher ethe affect of the IV on the DV is examined using one subject

Experimental single subject design

designs where the effect of the IV on the DV is examined using one subject

-cannot perform statistical tests

Experimental single subject design

designs in which the effect on the IV on DV is determined using one subject

Reversal Designs (ABAB)

a subjects behavior is recorded prior to treatment, then a treatment is introduced and behavior is measured again. Then treatment is withdrawn to see if behavior reverses to pre-treatment levels, then treatment is reintroduced

LOOK FOR: one treatment for one behavior

Problems with Reversal Designs

Likely Placebo Effects, Some treatments likely persist after being withdrawn, good likelihood of Type I error

Multiple Baseline Design

a) several different behaviors of one subject are monitored

b) a treatment is applied to only one of the behaviors

c) the researcher determines if the treated behavior changed relative to others

d) process is repeated for each of the other behaviors

LOOK FOR: only design where you measure multiple behaviors and one treatment for each

Multiple baseline design

Measure multiple behaviors and treat each behavior one at a time

Problems with Multiple Baseline Designs

in order for logic of design to work, the measured behaviors must be completely independent

Multi Element Design

Each time period, a different treatment is chosen at random and the behavior being treated is measured. Each treatment is given to the subjects multiple times and the treatment with the best average effects on the behavior is deemed the best treatment

LOOK FOR: several treatments for one behavior

Multi-element Design

Each time period a different treatment is chosen at random and the behavior being treated is measured. Each treatment is given to the subject multiple times and the treatment with the best average effect on the behavior is deemed the best treatment

multielemental design

several different treatments for one behavior are being tested. During each time period, one treatment is selected randomly and applied to the subject, and the subject's behavior is measured. This process is repeated several times for each of the treatments. The treatment with the best mean effects is deemed the most effective.

Problems with Multi Element Designs

Placebo Effects and only works with treatments that dont have carry over effects

problems with multi element designs

placebo effects and only works with treatments that don't have carry over effects

Range Restriction

the range on a variable is restricted, causing correlations with other variables to be artificially low

Range Restriction

Occurs when there is a floor or ceiling effect on one of the variables in the correlation

range restriction

(occurs when there is a floor/ceiling effect on one of the variables in a correlation)

Can lower correlations

Range Restriction

lack of variability in participants' response to a survey question or DV. includes floor and ceiling effects

Range Restriction

occurs whenthere is a floor or ceiling effect on one of the variable in a correlation

Range restriction

occurs when there is a flop or ceiling effect on one of the variable in a correlation

Reasons why correlation cannot be used to determine causation

1. It is sometimes unclear which variable is cause and which is effect (directionality problem)

2.A 3rd variable may be responsible for the relationship (Third variable problem)

Reasons why correlation cannot be used to determine causation (2)

1. Sometimes unclear which variable is cause and which is effect (DIRECTIONALITY PROB)

2. A third unmeasured variable may be responsible for any observed correlation (3RD VARIABLE PROB)

Problems with all single subject experimental designs

1.Likely chance of type I and type II errors

2.Placebo effect

3.Experimenter bias may influence results

4.Generalization difficulties

Problems with All Single Subject Experimental Design

1. Likely to cause Type I and II error

2. Placebo Effect

3. Experimenter bias

4. Generalization Difficulties

4 Problems with Single Subject Experimental Designs

Likely Type I and II errors

Likely experimentor bias

Likely placebo effects

Generalization Difficult

Problems with all single subject experimental designs

- Likely chance of Type I and Type II errors
- Likely placebo effect
- Experimenter bias may be responsible for the results
- Generalization difficulties

Problems with all Single Subject Experimental Designs

Problems with all single subject experimental designs

1. likely chance of Type 1 and 2 errors

2. placebo effects are likely

3. experimenter bias

4. generalization difficulty

Problems with All Single Subject Experimental Designs

1. Likely chance of Type I and Type II errors

2. Placebo effects are likely

3.Experimenter bias

4. Generalization difficulties

When to use correlation

to work out if 2 continuous variables are associated

interval and ratio scales

WHEN TO USE: Correlation

Determining the direction and strength of the linear relationship between two variables

Correlation

relationship between

Rosenthal (Experimenter Expectancy) Effect

Type of demand characteristic where the research give unintentional cues that change the subject's behavior.

experimenter expectancy effect (rosenthal)

a demand characteristic that occurs when subjects change thier behavior due to unintentional cues from the researcher

Experimenter Expectancy/ Rosenthal Effect

Demand characteristic that occurs when subjects change their behavior due to unintentional cues from the researcher

Experimenter expectancy effect (rosenthal)

a demand characteristic that occurs when subjects change their behavior because of their expectation that change should occur.

ExPost Facto Design

A design in which the researcher uses archival data to study an event occurs in the past

Reversal design

A type of single-subbject design that involves repeated alternatinos between a baseline period and a treatment period.

reversal design

a subjects behavior is recorded prior to treatment then a treatment is introduced and behavior is measured again then the treatment is withdrawn to see if the behavior reverses to pre-treatment level and the the treatment is reintroduced

multielement design

An experimental design in which two or more conditions (one of which may be a no-tx control condition) are presented in rapidly alternating succession (e.g. on alternating sessions or days) independent of the level of responding; differences in responding between or among conditions are attributed to the effects of the conditions *( also called concurrent schedule design, alternating treatment design, multiple schedule design).*

Multielement design

Each time period, a different treatment is chosen at random and the behavior being treated is measured. Each treatment is given to the subjects multiple times and the treatment with the best average effects on the behavior is deemed the best treatment

LOOK FOR: several treatments for one behavior

Multielement design

Several different treatments for one behavior are being tested. During each time period, one treatment is selected randomly and applied to the subject and the subject’s behavior is measured. This process is repeated several times for each of the treatments. The treatment with the best mean effects is deemed the most effective

Reversal Design (ABAB Design)

A subjects behavior is recorded prior to treatment, then the treatment is introduced and behavior is measured again. Then the treatment is withdrawn to see if the behavior reverses to pre-treatment levels, and then treatment is re-introduced

When interpreting an ANOVA Summary Table, the MS value for an effect is found by:

dividing the SS value from the same line by the df value from the same line

When interpreting an ANOVA Summary Table, the degrees of freedom for F are...

F(same line df value, above total df value)

When interpreting an ANOVA Summary Table, the F ratio is calculated by:

dividing the (MS value from the same line) by (MS value from the line above the total)

When interpreting an ANOVA Summary Table, you can determine the type of ANOVA by the number of lines present:

3: Between Subjects one Way Anova

4: Within Subjects One Way Anova

5: Between Subjects Factorial Anova

***Includes the Total Line

Reasons Why Correlation Cannot Infer Causation

1. It is unclean which variable is caused and which is effected (directional problem)

2. A third unmeasured variability may be responsible for the relationship (The Third Variable Problem)

Mstreat

measures the variablity in the dv due to the iv and error

MStreat

the variance among the Sum of the IV. the more effect the IV is having on the DV the larger MStreat is.

-it measures variability in the DV due to the IV and Error

-it measures variability in the DV due to the IV and Error

Msw

measures the variability in the dv due only to error

F

measures how much greater mstreat is then msw and whether the IV is affecting the dv

(t(x))^2=F(1,x)

Mathematical relationship between the t and F distributions

F

measures whether MStreat is reliably greater than MSwithin and thus whether the IV is affecting the DV

F

measures how much greater MStreat is to MS within and thus wether the IV is affecting the DV

F

Measures how much greater MStreat is to MSwithin and thus wether the IV is affecting the DV

correlation co-efficient

a statistic Used to express the direction and strength of the linear relationship b/w 2 variables measured on an interval or ratio scale

Correlation co-efficient

Statistic used to indicate strength and direction (positive or negative) of the linear relationship between two variables (measured on an interval or ratio scale)

Correlation co-efficient

statistic used to indicated strength and direction (+ or -) of a linear relationship between two variables

-interval or ratio scales

Correlation Co-efficient

a statistic used to measure the direction (positive or negative)

correlation co-efficient

measures the direction and strength of the linear relationship of two variables measured on an interval or ratio scale

problems with all single subject experiment designs

-likely type I and type II errors

-placebo effects

-experimenter bias

-generalization difficulties

when to use a correlation test

used to determine the relationship b/w variable measured on an interval or ration scale

hen to use a within subjects variance test

used to compare the variances for within subjects experiments w/ one independent variable that has exactly 2 levels

interupted time series design

a time series where the dependent variable is interrupted by the manipulation of the independent variable

pretest post test design

a single group of participants is measured on the dependent variable

High Stat Power is desirable because...

it prevents type II errors

Increasing Alpha...

increases power. but increasing alpha/power also increases probability of making a type I error. Increasing alpha/power, decreases beta or type II errors though

3 ways to increase stat power

1. choose IV levels that will max effect size

2. try to lower variability in data

3. collect more data or increase sample size

4 types of scales and properties each has

nominal (identity)

ordinal (identity and mag)

interval (identity, mag, eq. int.)

ratio (identity, mag, eq. int., and abs zero)

why is high statistical power desirable?

it prevents type II errors

MStreat measures what? (in an ANOVA)

measure of the variability in DV due to the IV and error

MSw measures what? (in an ANOVA)

the variability in the DV due to ONLY error

F measures what? (in an ANOVA)

how much greater MStreat is than MSw and thus whether the IV is affecting the DV

What does F measure in an ANOVA?

A measure of whether MStreat is reliably greater than MSwithin and thus whether the IV effects the DV

possible values of a correlation co-efficient

-1.00 to +1.00

Possible values of a correlation co-efficient

-1, +1

Possible values of a correlation co-efficient

-1 to +1

Percent of variance accounted for

100% * r squared

Parametric Statistical Tests

require assumptions about the distribution of the DV (e.g. normality) in order to obtain correct p-values

Parametric Statistical Tests

Require assumptions about the distribution of the DV (Normality) in order to obtain correct p value

Parametric statistical tests

Require assumptions about the distribution of the dependent variable (eg, normality) in order to obtain correct p-values

Parametric statistical tests

require assumptions about the distribution of the DV in order to obtain correct p-values

Nonparametric statistical tests and advantages/disadvantages

a class of tests which do not assume the DV is normally distributed - do assume symmetry however

ad: less restrictive assumptions, often easier to calculate

dis: tend to be overly conservative (more type II errors), no nonparametric tests exist for some research designs (e.g. factorials)

when to use a correlational statistical test

used to determine the relationship between variables measured on an interval or ratio scale

when to use the test of independent frequencies (chi-squared test)

to compare whether the frequencies of different responses are reliably different when each observation comes from a different subject

Advantages of non-parametric statistics (2)

less restrictive assumptions

often easier to calculate

Advantages of Non-Parametric Statistics

-Less restrictive assumptions

-Often easier to calculate that parametric tests

-Often easier to calculate that parametric tests

disadvantages of non-parametric statistics (2)

tend to be overly conservative (more type II errors)

no non-parametric tests exist for some research designs (e.g. Factorial)

Disadvantages of non-parametric statistics 2

1. tend to be overly conservative (more type II errors)

2. no non-parametric test exist for some research design

Disadvantages of Non-Parametric Statistics

-Tend to be overly conservative

-No non-parametric exist for some research designs (ex. factorial designs)

-No non-parametric exist for some research designs (ex. factorial designs)

Disadvantages of Non-Parametric Statistics

1. Tend to be overly conservative

2. No non-parametric tests exist for some research designs (factorial designs)

factors increasing type I errors

regression to the mean

confounds
Factors Increasing Type Two Errors

-Nuisance Variables

-Floor and Ceiling Effects

-Narrow Range of the IV

factors that can increase type 2 errors

1. nuisance variables

2. floor & ceiling effects

3. narrow range of the independent variable

Advantages to Non Parametric Statistical Tests

1. Less restrictive assumptions

2. Often easier to calculate

Disadvantages of Non Parametric Tests

1. Tend to be overly conservative

2. No non parametric test exists for some research design (factorial)

mathematical relationship between the t and F distributions

at any alpha level: (t(x))^{2}= F(1,x)

Example: t(9) = 2.262

(t(9))^{2}=5.12, F(1,9)=5.12

Mathematical Relationship between the t and F distributions

at any alpha level:

(t(x))^{2}=F(1,x)

(t(x))

Survey Research

research in which data is collected by directly asking subjects questions

Survey research

research method where data is collected by directly asking subjects questions

Convenience Sampling

sampling individuals who are readily available without regard to their characteristics

Convenience sampling

A nonprobability sampling method involving selection of individuals on the basis of their availability and willingness to respond; that is, because they are easy to get

Convenience sampling

Sampling individuals who are readily available without regard for their characteristics

Father of Modern Survey Research

George H. Gallup

Quota Sampling

type of sampling in which people are selected on the basis of pre-specified characteristics so that the sample will have the same distribution of characteristics as the population

Quota sampling

A type of sampling in which people are selected on the basis of prespecified characteristics so that the total sample will have the same distribution of characteristics as the population

Quota sampling

a type of sampling in which people are selected on the basis of pre-specified characteristics so that the total sample will have the same distribution of characteristics as the population

Quota sampling

sampling in which people are selected on the basis of pre-specified characteristics so that the sample will have the same distribution of characteristics as the population

Probability Sampling

method of sampling that involves a random selection component

Types of Probability Sampling

Simple Random Sampling

Stratified Random Sampling

Multistage Sampling

Simple Random Sampling

every member of a population is listed and members are then randomly selected for questioning (expensive)

Simple Random Sampling

Every member of the population is listed and members are then randomly selected for questioning.

Stratified Random Sampling

total population is divided into demographic groups and then members of each group are randomly selected for questioning (most expensive)

Stratified Random Sampling

Total population is divided into demographic groups and then members of each group are randomly selected for sampling.

Stratified Random Sampling

The total populaiton is divided into demographic groups, and then members of each group are selected for questioning

Stratified random sampling

total population is divided into demographic groups and then members of each group are randomly selected for questioning

Stratified random sample

The total population is divided into demographic groups and then members of each group are randomly selected

Multistage Sampling

in a first random selection stage, natural groups (e.g. zip codes, countries, colleges) are selected. Then in a 2nd stage, people in each of the groups that were selected in stage one are randomly selected for questioning

Multistage sampling

In a first random selection stage, natural groups (e.g., zip codes, counties, colleges) are selected, then in a second random selection stage, members from each selected group from the first stage are randomly selected for questioning

Convenience Sampling --> Quota Sampling --> Probability Sampling

As one goes down the list, each sampling technique becomes more accurate and more expensive.

When is it okay to use convenience sampling or quota sampling?

when the population to which you would like to generalize your results is relatively homogeneous

Factors to consider when evaluating polls/surveys (8)

1. bias in choosing sample

2. lack of full disclosure when reporting results

3. lying by poll respondents

4. non respondents influencing sample

5. characteristics of poll taker

6. type of question asked

7. wording of question

8. question order

open questions

questions in which answers are not provided - allow people to more accurately convey what they think, however, open questions can constrain respondents by not legitimizing answers the researcher intended to include

**best used when there are a large number of legitimate responses

closed questions

questions in which answers are provided - easier to score but may bias responsdents to give answer the researcher wants - best used when there are only a small number of possible responses

Response Acquiescence Effect

tendency of people to say to yes to questions they have thought little about

Response Acquiescence Effect

The tendency of subjects to respond "yes" to questions they have thought little about.

Response Acquiescence Effect

People tend to respond yes to questions they have thought little about

information you must know in order to evaluate a survey properly

how many non-respondents there were and how they were handled, how the questions were exactly worded, what the characteristics of the poll takers were, what other questions were asked

best use of surveys

to compare changes in responses over time or across different demographic groups

Best use of surveys

to compare changes over time or across different demographic groups

Floor effects

The task is too hard or otherwise biased toward producing low scores

Floor Effects

When the values of the DV are so low that they are unlikely to be affected by the IV

Floor effects

occur when the values of the DV are so low that they are unlikely to be affected by the IV

Ceiling Effects

performance is too high; at the top of the scale

Ceiling Effects

When the values of the DV are so high that they are unlikely to be affected by the IV

Ceiling Effects

occur when the values of the DV are so high that they are unlikely to be affected by the IV

Properties of Scales

- Identity
- Magnitude
- Equal Unit Size
- Absolute Zero

Factors to consider when evaluating polls

-How many non-respondents were there and how were they handled?

-Exactly how was the question worded?

-What were the characteristics of the poll takers.

Factors to Consider when Evaluation Polls

1.) Bias in choosing the sample

2.) Lack of full disclosure when reporting results

3.) Lying by pole respondents

4.) Non-respondents influencing the sample

5.) Characteristics of the poll taker

6.) Type of question asked: open or closed

7.) Wording of the question

8.) Question order

2.) Lack of full disclosure when reporting results

3.) Lying by pole respondents

4.) Non-respondents influencing the sample

5.) Characteristics of the poll taker

6.) Type of question asked: open or closed

7.) Wording of the question

8.) Question order

factors to consider when evaluating a poll

1. Question order

2. wording of the question

3. type of question asked (open vs. closed)

Ethical Principles when Conducting Psychological Research (4)

-Benefits > Risks

-Withdraw Anytime

-Minimize Discomfort

-Informed Consent

Ethical Principles When Conducting Psychological Research

- Subjects should agree to participate

- Subjects should not be coerced into participation

- Subject data should be confidential

- The gain in science should outweigh any harm done to the subjects

- Subjects should be fully informed about the purpose of the experiemnt before they decide to participate

Ethical Principles when Conducting Psychological Research

1.) Ideally, subjects should agree to participate

2.) Ideally, subjects should not be coerced into participation

3.) Ideally, the subject should be fully informed about the purpose of the experiment when deciding whether to participate

4.) Ideally, an experiment involving deception, subjects should be fully informed about he true nature of the experiment at its conclusion

5.) The gains to science from an experiment should outweigh any harm done

6.) Ideally, subjects should be warned of any potential harmful effects

7.) Ideally, subject data should be kept confidential

8.) All reports of the data must be accurate

What is one ethical principle when conducting psychological research?

Informed consent

Ethical Principles when conducting psychological research

1.Subjects should agree to participate

2. all reports of data must be accurate

3. subject data should be kept confidential

4.the gains for science must outweigh any harm done to the subjects

margin of error

means that there is a 95% chance that if the survey were conducted on the entire population in exactly the same way it was conducted on the sample, the results would be within the margin of error

***NEVER should be calculated for a sample that was not randomly selected

Margin Of Error

It means that there is a 95% chance that if the survey conducted on the entire population in exactly the same way it was conducted on the sample, the results would be within the margin of error.

margin of error

the range of outcomes we expect for a population, given the data revealed by a sample drawn from that population

Margin of Error

There is a 95% chance that if the survey were conducted on the entire population, exactly as it had been conducted on the sample, the results would be within the margin of error

Margin of error

the 95% chance that a survey can be reproduced on the entire popluation

Nuremberg Code of Ethics

any subject must voluntarily agree to participate in research, the consenting subjects much be free from deceit or coercion, all potential hazards must be made known to the subjects beforehand

APA code of ethics

differs from Nuremberg in that the APA code allows certain forms of deceit/coercion to be applied to research subjects

Ethical principles when conducting psych. research

1. subjects should agree to participate

2. subjects should not be coerced into participation

3. subject should be fully informed about the purpose of the experiment when deciding to participate , 4. subjects should be informed about deception when applicable, 5.gains to science should outweigh harm to subjects, 6. subjects should be warned about harmful effects, 7. data should be confidential, 8. all reports of data should be accurate

correlation between IV levels (within sub designs only)

larger correlation gives more power

solution to: poor at capturing non-linear relationships

always graph the data

solution to: greatly affected by extreme scores

report correlation with all data included and also with extreme points removed

solution to: lowered by range restriction

if possible, choose variables that do not exhibit floor/ceiling effects

single subject designs

research designs based on testing one subject

problems with ethical principles to consider

virtually every instance when people participate in research involves some sort of coercion, some experiments require deception, and because of the placebo effects some warnings may do harm

Title Page

contains: title (15 words or fewer), authors name(s) and affiliation(s)

Title Page

- 15 words or fewer in length

- Authors names

- Authors affiliations

Title Page

Contents: 15 words or less

Title page

Contains title, author's name(s), author's affiliation

Title page

always page one, obtains authors and their affiliation

Abstract

always alone on page 2 - brief (150 words or less) summary of article

should tell problem under investigation, subjects, experimental method(s), the findings, and the conclusions

Abstract

-Always alone on page 2 of the manuscript

-Is a brief (less than 150 words) summary of the article

-Should tell 1) the problem under investigation, 2) The subjects, 3) Experimental method, 4) Findings, 5) Conclusion

Abstract

Always alone on p.2 of the manuscript. The abstract is a brief (less than 150 words) summary of the artible

Abstract

One paragraph summary of article; includes problem under investigation, subjects, research method, findings, conclusions

Abstract

Always on page 2 - one paragraph summary that includes the problem under investigation, the subjects, the research method, findings, conclusions

Abstract

always on page 2

summary of article

introduction

always begins on page 3 of manuscript

purpose: introduce problem, review previous literature, and state purpose/rationale for your research

Introduction

-Always begins on page 3 of the manuscript (longest section)

-Purpose; 1) Introduce the problem, 2) Review the previous literature, 3) End of the intro, shate the purpose and retionable for the research

Introduction

Always begin on page 3. Purpose:

a. introduce the problem

b. review the previous literature

c. state the purpose and rationale for your research.

Introduction

Introduce the problem, review previous literature (literature review), state purpose and rationale

Introduction

always on page 3 - introduce the problem, review previous literature, state the purpose and rationale of the reported research

Introduction

always on page 3

reviews previous research

method

recipe for conducting the reported research

usually divided into 3 subsections

a. participants (# of and compensation received)

b.apparatus (how each piece of equip. was used)

c. procedure (explains the design aspects of the research

***the iv, dv, and levels of iv, decribe counterbalancing, and anything else necessary to replicate

Method

-A recipe for conducting the reported research

-Three subsections; 1) Participants, 2) Apparatus, 3) Procedure

Method

A recipe for conducting the reported research. Divided into 3 subsections. *participants reports the # of subjects and the composition they received. *Apparatus: describes how each piece of equipment was used. *procedure describes the design aspects of the research. The DV and the levels of the IV, describe counter balancing, describe the subjects tasks, describes anything else necessary to replicate.

Method

A recipe for conduction the reported research - three sections: participants, apparatus, precedure

Method

participants

apparatus

procedure

results

reports data collected in research and the results of the stats tests

contains graphs/tables showing means of the DV at different IV levels, also includes obtained values and p-values for the stats tests

Results

-Reported the data collected in the research and the results of the statistical test

- Contains graphs or tables showing the means of the DV at different IV levels, also includes the obtained values and p-values for the statistical tests

Results

reports the data collected in the research and the results of the statistical tests. Contains graphs or tables showing the means of the DV at different IV levels

Results

Reports the data collected in the research; includes graphs or tables showing data and results of statistical tests

Results

Reports the data collected in the research - include graphs and reports of statistical tests

Results

reports the data

stats analysis

discussion section

relates data from the research to the research question posed in the intro

includes: theoretical implications of data, potential limitations (optional), and ideas for future research (optional)

in press

the status of an article that has been accepted for publication but which has not yet appeared in the journal

in press

status of an article that has been accepted for a journal but has not been published yet.

galley proof

type set copy of the article that is sent to the author for final approval

galley proofs

a copy of the article that is sent to the author for final aproval

Mulitstage Sampling

IN a first random selection stage, natural groups (e.g., zip codes counties, colleges) are selected, then in a second stage people in each of the groups selected in stage 1 are randomly selected for questioning

Factors to Consider when Evlauating Surveys

1. Bias in choosing the sample

2. Lack of full disclosure when reporting the results

3. Lying by poll respondents

4. Non-respondence influencing the sample

5. Characteristics of the full taker

6. Type of question asked (open or closed)

7. Wording of the question

8. Question order

Discussion

-Relates the data from the research to the research question posed in the introduction

-Includes; 1) Theoretical implications of the data, 2) Discuss potential limitations, 3) Ideas for future research

Discussion

Relates the data from the research to the research question posed tin the introduction. Include: *theoretical implications. *discuss potential limit * can also discuss ideas for future research.

Discussion

Relates the results of the research to the research question posed in the Introduction; includes discussion of theoretical implications of results

Discussion

Related results of the research to the research question posed in the introduction

Discussion

interprets the results

Difference between Nuremberg code and the APA code

The APA code allows certain forms of deceit or coercion to be applied to research subjects.

How the Nuremberg Code and APA Code differ

Allows certain forms of deceit or coercion

4 Ethical problems when conducting psychological research.

1. Ideally subjects should agree to participate in the research.

2. Ideally subjects should not be coerced into participation

3. Ideally subjects should be fully informed about the purpose of the experiment when deciding whether to participate

4. Ideally experiments involving deceptions subjects should be fully informed about the nature of deception after the experiments.

In comparing Mean and Median, if the Mean and Median are = what kind of distribution is this?

Normal Distribution

Positively Skewed Distribution / Mean > Median

A Negatively Skewed Distribution Mean < Median

Between Mean and Median, what provides more consistency?

Mean provides more consistency.

Between Mean and Median, which has an easier calculation?

Mean has an easier calculation

Mathematical relationship between t and F

At any alpha:

(t(x))^{2} = F(1,x)

How does the Nuremberg code of ethics differ from the APA code?

APA code allows for deception and coercion to be applied to potential research objects

Falsification

There must be at least some hypothetical facts that the theory would not be able to explain

Falsification

For a theory to be useful, it must be able to generate predictions - some hypothetical facts that the theory is unable to explain

Falsification

for an explanation to be useful, it must be able to generate predictions. As such there must be at least some hypothetical facts that would prove the theory false.

It must always be possible to prove a scientific theory false.

It must always be possible to prove a scientific theory false.

Modes Ponens

1)If p is true, then q is true 2)p is true

3)Therefore, q is true

Valid form of argument

Not useful to science

3)Therefore, q is true

Valid form of argument

Not useful to science

Modes Tollens

1)If p is true, then q is true 2)q is not true

3)Therefore, p is not true

Both valid and useful

3)Therefore, p is not true

Both valid and useful

Independent variable (IV)

The variable in an experiment for which the researcher chooses values

Independent Variable (IV)

The variable for which the researcher chooses the values

Dependent variable (DV)

The variable the researcher measures to enter mine the effects of the independent variable

Dependent variable (DV)

the variable the researcher measures to determine the affects of the IV

Dependent Variable (DV)

The variable the researcher measures to determine the effects of the IV.

Quasi-experiemtnd

A research technique in which the researcher manipulates the IV, but which fails to have random assignment

Reasons Why an Experiment Cannot be Done

1)Difficult/Impossible to manipulate IV

2)Unethical to manipulate IV

3)Subjects cannot be randomly assigned

Reasons why an Experiment Sometimes Cannot be done

1. Difficult or impossible to investigate the IV

2. Unethical to manipulate the IV

3. Subject cannot be randomly assigned

Reasons why an experiment sometimes cannot be done:

1. Difficult or impossible to manipulate the independent variable.

2. Unethical to manipulate the independent variable.

3. Subjects cannot be randomly assigned.

Mode

Most often occurring number in a data set

Potential problems with within subjects design

1. Practice effects

2. Sensitization effects

3. Carry-over effects

Areas to compare Mean and Median

-Effects of Extreme Scores

-Consistency

-Ease of Calculaton

Comparison of Mean and Median: Effects of Extreme Scores

The mean is far more affected by extreme scores than the median. (The median is often a more useful descriptive statistic in skewed distributions.)

Comparison of mean and median: Effects of extreme scores

Mean is more affected by extreme scores

Comparison of Mean and Median: Consistency

The mean provides more consistent results across repeated samples than the median.

Comparison of mean and median: Consistency

Mean provides more consistent results across repeated samples

Comparison of Mean and Median: Ease of Calculation

Mean is usually easier and faster to calculate.

Ease of calculation comparison of mean and median

Mean is usually easier and faster than median

Comparison of mean and median: Ease of Calculation

Mean is usually easier & faster

Standardized Scores (z-scores)

Allow comparisons to be made between scores measured on different scales by placing all scores on a common scale

Standardized Scores (z scores)

Allow comparison to be made between scores measured on different scales by placing all scores on a common scale

mean= 0 standard deviation= 1

Normal Scale of Measurement

Identity only

Ordinal Scale of Measurement

Identity and Magnitude

Interval Scale of Measurement

Identity, magnitude, and equal intervals

Ratio Scale of Measurement

Identity, magnitude, equal intervals, and absolute zero

Methods of Counterbalancing Practice Effects

-Use all treatment orders

-Latin Square

Why is Statistical Power desirable?

It lowers Type Two Errors

Narrow Range of the Independent Variable

Levels of the IV are so similar that their effects cannot be distinguished

Narrow range of the independent variable

occurs when levels of the IV are so similar that their effects on the DV cannot be distinguished

Experimental Logic Breaking down with Confounds

We don't know if the IV or confound is causing the changes

Factors which can increase Type II errors

1. Nuisance Variables

2. Floor and ceiling effects

3. Narrow Range of the IV

research technique in which the researcher simply observes and describes behavior

Naturalistic Observation

research technique in which the research determines the relationship between variables without manipulating the variables

correlational approach

variable in an experiment for which the researcher chooses values

independent variable

variable the researcher measures to determine the effects of the IV

dependent variable

research design in which each subject is assigned to one level of the IV

between subjects design

research study that satisfies the following criteria: random assignment & researcher manipulates the IV

experiment

finding an effect of the IV on the DV, when in reality no such effect exists (false positive)

Type I Error

failing to find n effect of the IV on the DV when in reality an effect does exist (false negative)

Type II Error

anything other than IV that can affect DV; factor that can increase Type II errors, found in every experiment, decrease power, but not fatal

nuisance variables

Experiment expectancy effect

a demand characteristic that occurs when subjects change their behavior due to unintentional ques from the researcher

values of DV are so low or so high that they are unlikely to be affected by the IV; factor that can increase Type II Errors

Floor and Ceiling Effects

a nuisance variable that varies non-randomly with the IV

confound

an IV whose levels were chosen randomly from a population of possible values; if reliable then it may be generalized to all levels of population

random factor

an IV whose levels were chosen non-randomly; if reliable may NOT be generalized outside of tested levels

fixed factor

if p is true then q is true

p is ture

therefore, q is true

valid form of argument, but not useful in science because assumes theory is true

Modus ponens

if p is true then q is true

q is not true

therefore, p is not true

valid form of argument and useful to science

used to prove theories false

modus tollens

if p is true then q is true

q is true

therefore, q is true

invalid form of argument; logically impossible to prove a theory is true

affirming the consequent

when manipulating the variables would be difficult or impossible

when manipulating the variables would be unethical (can't do experiments)

when to use the correlational approach

1) difficult or impossible to manipulate the IV

2) unethical to manipulate the IV

3) subjects cannot be randomly assigned

circumstances when the experimental approach cannot be used

allows researcher to infer a CAUSAL relationship between the IV and the DV

advantage of the experimental approach

investigation: can't infer causation

experiment: can infer causation

investigation vs. experiment

ss= Σ (x-x_{bar})^{2}

formula for the SS (sum of squares)^{}

identity, magnitude, equal intervals, and absolute zero

properties of scales of measurements

nominal, ordinal, interval, ratio

types of scales of measurements

* allows use of fewer subjects to obtain the same number of observations

* allows for greater statistical power than between subjects design

advantages of within subjects designs

* practice effects (subject's performance changes as result of experience)

* sensitization effects (realizes what manipulations in study are and causes subject to change behavior)

* carry-over effects (effects of 1 treatment persist when another treatment is introduced)

problems with within subjects designs

assigning subjects to treatment orders in a within subjects design so that across subjects practice effects are balanced

* use ALL possible treatment orders

* use a Latin square

methods of counter-balancing

has as many columns and as many rows as there are IV levels;

columns = treatment order

rows = subjects/groups

treatments placed in each cell of Latin square so appear only once in each row and once in column

Latin Square

Overcoming Rosenthal and Placebo Effects

1. Single blind study

2. Double blind study

* alpha level

* effect size

* variability in the DV

* sample size

* correlation between the IV levels (in within subj designs)

factors determining statistical power

* choose IV levels that will MAX effect size

* try to lower variability in DV

* increase sample size

How to increase statistical power

* nuisance variables

* floor/ceiling effects

* narrow range of the IV

* can't prove a null hypothesis

Factors which can increase Type II errors

* regression to the mean

* confounds

Factors producing Type I errors

Experimenter Expectancy Effect (Rosenthal effect)

demand characteristic that occurs when subjects change their behavior due to unintentional cues from the researcher

* confounds due to subject assignment

* confounds due to manipulation of the IV

types of confounds

* Single-Blind Experiment (overcomes Placebo, not Rosenthal)

* Double-Blind Experiment

(overcomes both Placebo & Rosenthal)

ways to overcome Placebo/Rosenthal effects

it could be the IV or the confound that caused any changes in the DV

How is the logic of experimentation ruined by confounds?

Types of Demand Characteristics

-Experimenter Expectancy/Rosenthal

-Placebo

-Hawthorne

-Placebo

-Hawthorne

Measures the variability in the dependent variable due to the independent variable and error

What MStreat measures

When one variable goes up, so does the other one

Positive Relationship

When one variable goes down, the other variables goes up

Negative Relationship

Measures the variability in the dependent variable due only to error

What MSwithin measures

Measures how much greater MStreat is than MSwithin and thus whether the independent variable is having an effect on the dependent variable

What f measures

A design in which there is more than one independent variable, and every independent variable level is present at all levels of the independent variables

Factorial design

The independent effect of one independent variable on the dependent variable in a factorial design

Main effect

It is sometimes unclear which variable is cause and which is effect; A third variable may be responsible for the relationship

Reasons why correlation cannot be used to determine causation

Multiple t-tests are more work; Multiple t-tests inflate the type I error rate

Why do an ANOVA instead of multiple t-tests

Poor at capturing non-linear relationships, greatly affected by extreme scores, can be lowered by range restriction

Mathematical limitations of correlations

ANOVA MStreat

measures the variability in the DV due to the IV and error

ANOVA MSwithin

measures the variability in the DV due only to error

ANOVA F Ratio

Measures how much greater MStreat is than MSwithin and thus whether the IV is having an effect on the DV

Single Group, pretest, post-test design

A time series design in which only one measurement of the DV is made before treatment and only one measurement of the DV is made after treatment

Problems: Regression to the mean, practice effects, placebo effects, left entirely uncontrolled

Positive aspects of Natural observation

1. often allows data collection that is impossible to obtain otherwise

2. high in external validity

3. can be used to verify the external validity of an experiment

4. relatively uncomplicated

5. excellent starting point for generating hypotheses testable by other research techniques

6. studies behaviors as they unfold over time

Negative aspects of Natural Observation

1. Practical problems with data gathering

2. observing can change the behavior Soln: don't let those being observed know they're being observed

3. obersver bias can greatly affect the results Soln: use a double blind procedure

4. Can be unethical

5. Cannot be used to determine causation

MSwithin

measures the variability in the DV due ONLY to error

Single group, Pretest-Posttest design

Quasi-Experiment

Interrupeted time series design

Quasi-Experiment

Interrupted Times Theory design

Quasi-Experiment

Non equivalent before-after design

Quasi-Experiment

Ex Post Facto Design

Quasi-Experiment

a design in which a researcher uses archival data to study an event that occurred in the past

Types of single subject experimental designs

1. Reversal Design (ABAB) design

Types of Single Subject Experimental Designs

2. Multiple Baseline Design

A. several different behavior of one subject are recorded

B. a treatment is applied to only one of the behaviors

C. the researcher determines if the treated behavior changes relative to the others

D. Process is repeated for each of the other behaviors. If each treatment has its effects only on the targeted behavior, the researcher assumes the treatment was effective.

Types of Single Subject Experimental Design

3. Multielement Design

In the multielement design, several different treatments for one behavior are being tested. During each time period one treatment is selected randomly and applied to the subject and the subjects behavior is measured. This process is repeated several times for each of the tratments. The treatment with the best mean effects is deemed most effective

What does MSwithin measure in an ANOVA?

A measure of the variability in the DV due only to error.

Natural Oberservation

a research technique in which the researchers simply observe and describe behavior

Explanation for a phenomenon that can be falsified and involves entities that cannot be directly observed

Theory

Tentative statement about a possible relationship between observable variables

Hypothesis

Specific values of the IV that a researcher chooses to use

Levels of the Independent Variable

Random Selection

Once the participants for the experiment have been chosen they all have an equally likely chance to be assigned to each IV level or to each treatment order

Random Assignment

Researcher manipulates the IV but fails to have random assignment

Quasi-Experiment

Distribution that is symmetrical and bell-shaped

Normal

Distribution that has a few extreme high scores

Positively Skewed

Distribution with a few extreme low scores

Negatively Skewed

Effects of extreme scores of the mean and median

Mean is more affected by extreme scores

When a subject's performance of the task changes (for better or worse) as a result of experience with the task

Practice Effects

Subject realizes what the manipulations are and that awareness causes change in behavior

Sensitization Effects

Effects of one treatment persist when another treatment is introduced

Carry-Over Effects

Extent to which your research provides a valid test of the relationship between the IV and the DV

Internal Validity

Probability of making a Type One error

Alpha

Probability of making a Type Two Error

Beta

It is impossible to rove that an IV has no effect of the DV

Can't prove the Null Hypothesis

Tendency for extreme values of a variable to fall closer to the group mean when retested

Regression to the Mean

Extent to which the results of a study can be applied outside the research situation

External Validity

Aspects of a study that indicate to subjects how they are expected to respond

Demand Characteristics

Experimenter Expectancy Effect/ Rosenthal Effect

Demand characteristic that occurs when subjects change their behavior as a result of the expectation that change should occur

Placebo Effect

Demand characteristic that occurs when subjects change their behavior because they know they are being observed

Hawthorne Effect

When the DV is influenced by the IV only because the IV something new (NOT A DEMAND CHARACTERISTIC)

Novelty Effect

Logic of an ANOVA: MStreat

Measures variability in the DV due to the IV and error

Logic of an ANOVA: MSwithin

Measures the variability int he DV due only to error

Logic of an ANOVA: F

Measures how much greater MStreat is that MSwithin and thus whether the IV is having an effect on the DV

When the researcher becomes a member of the group being observed

Participant Observation

Indicates strength and direction of linear relationship between two variables

Correlation Coefficient

When there is a floor or ceiling effect on one of the variables that lowers the correlation

Range Restriction

Description of the behavior or abilities of a single, usually exception, individual

Case Study

Effect of an IV on the DV is examined using 1 subject

Experimental Single Subject Design

Subject's behavior is recorded prior to treatment, then treatment is introduced and behavior is recorded again; then the treatment is withdrawn to see if behavior reverts to pre-treatment levels, then the treatment is reintroduced and behavior is measured again.

Reversal Design (ABAB Design)

Subject's behavior is recorded prior to treatment, then treatment is introduced and behavior is measured again, then the treatment is withdrawn to see if behavior reverts to pre-treatment levels, then the treatment is reintroduced and behavior is measured again *ONE TREATMENT FOR ONE BEHAVIOR*

Reversal Design

Several different treatments for one behavior are being tested; during each time period, one treatment is selected randomly and applied to the subject- the subject's behavior is measured- this process is repeated several times for each of the treatments- the treatment with the best mean effects is deemed most effective *SEVERAL TREATMENTS FOR EACH BEHAVIOR*

Multielement Design

A.) Several different behaviors of one subject are recorded

B.) Treatment is applied to only one behavior

C.) Researcher determines if the treated behavior changes relative to the others

D.) Process is repeated for each of the behaviors

B.) Treatment is applied to only one behavior

C.) Researcher determines if the treated behavior changes relative to the others

D.) Process is repeated for each of the behaviors

Multiple Baseline Design

Several different treatments for one behavior are being tested; during each time period, one treatment is selected randomly and applied to the subject- the subject's behavior is measured and this process is repeated several times for each of the treatments; the treatment with the best mean effects is deemed most effective

Multielement Design

Likely chance of type I and type II errors; placebo effects are likely; experimenter bias; generalization difficulties

Problems with all single subject experimental designs

Class of quasi-experiment in which the performance of a single group of subjects is measured both before and after some experimental manipulation

Time Series Design

Time series design in which only one measurement of the DV is made prior to IV manipulation and only one measurement is made afterwards

Single Group, Pretest/Post-test Design

Time Series Design in which more than one measurement of the DV is made prior to IV manipulation and more than one measurement is made afterwards

Interrupted Time Series Design

Problems with Single Group Pretest Post-test Designs

Regression to the mean; practice effects; placebo effects

Quasi-Experiment, but not time series, a design in which a pretest and post test (before and after IV manipulation) are given to two non-randomly assigned groups of subjects; the difference between the pretest and the post-test serves as the DV

Non-Equivalent Before-After Design

A design in which the researcher uses archival data to study an event that occurred in the past

Ex Post Facto Design

Factors determing statistical power

1. alpha level

2. effect size

3. variability in the DV

4. sample size

5. correlation between IV levels

(in within subjects design only)

Less restrictive assumptions, often easier to calculate

Advantages of non-parametric statistics

Tend to be overly conservative, no non-parametric tests exist for some research designs

Disadvantages of non-parametric statistics

Occurs when the effects of one independent variable on the dependent variable change depending on the level of another independent variable

Interaction

1.) If p is true, q is true

2.) Q is true

3.) Therefore, p is true

*Invalid form of argument because it assumes there could not be a different variable causing q*

Affirming the Consequent

1.) If p is true, then q is true

2.) Q is not true

3.) Therefore, p is not true

*Both valid and useful to science*

Modus Tollens

Different entities receive different scores

Identity

Ordering of the value reflects the ordering of the trait being measured

Magnitude

Difference of 1 on the scale means the same everywhere

Equal Intervals

Score of 0 indicates complete absence of the trait

Absolute Zero

Scale of measurement that only has identity

(EXAMPLE: football jerseys, zip codes, etc.)

Nominal

Scale of measurement with identity and magnitude (EXAMPLE: basketball rankings)

Ordinal

Scale of measurement with identity, magnitude, and equal intervals (EXAMPLE: Celsius degrees)

Interval

Scale of measurement with identity, magnitude, equal intervals, and absolute zero (EXAMPLE: weight in pounds, exam scores, Kelvin temperature)

Ratio

Graph showing number of times each score occurred in a data set

Frequency Distribution

Ways to compare the mean and median

-Effect of extreme scores

-Consistency

-Ease of calculation

Comparison of mean and median: normal distribution

Mean=Median

Comparison of mean and median: Positively skewed distribution

Mean>Median

Comparison of mean and median: negatively skewed distribution

Mean<Median

Bias in choosing the sample, lack of full disclosure when reporting results, lying by poll respondents, non-respondents influencing the sample, characteristics of the poll taker, type of question asked, wording of the question, question order

Factors to consider when evaluating polls (3 of 8)

People tend to respond "yes" to questions they have thought little about

Response acquiescence effect

To compare changes in responses over time or across different demographic groups

Best use of surveys

There is a 95% chance that if the survey were conducted on the entire population exactly as it was conducted on the sample, the results would be within the margin of error

Be able to explain what the margin of error in survey research means

When a subject's performance on the task chances (for better or worse) as a result of experience with the task

Practice Effects