becky w.

goodness of fit

-deviation of data from prediction

multinomial test

-count observations in every category

-convert to z-score

-h0 predicts z should be near 0

Advertisement

observed frequencies

-frequencies actually obtained in a data set

expected frequencies

-frequencies that you would predict

marginal frequency

-total count for each category, ignoring levels of another variabel

chi square statistic

-null hypothesis: variable are independent

-x^2= sum of(f.obs-f.exp)/f.exp

parametric statistics

-most common type of inferential statistics

-mathematically fully described

-r,t,F

-normality

-homogeneity of variance

-linear relationship

nonparametric statistic

-alternative to parametric statistics

-naive approach: far fewer assumptions about data

-wider variety of situations

-not as powerful as parametric statistics

spearman correlation

-alternative to pearson correlation

-convert data on each variable to ranks

-compute Pearson correlation from ranks

mann-whitney test

-alternative to independent-samples t-test

-combine groups and rank-order all scores

-if groups differ, high ranks should be mostly in one group and low ranks in the other

-test stat(U) to its sampling distribution

-ordinal data, non-normal pop and small sample sizes

wilcoxon test

-alternative to single- or paired samples t-test

-subtract m0 from all scores

-rank order the absolute values

-sum ranks separately for positive an negative dif scor

-ordinal data, non- normal pop

Advertisement

kruskal-wallis test

-alternative to simple anova

-extends mann-whitney

-rank all scores

-test statistic H essentially measures variance of sums of ranks

-ordinal data and non normal

friedman test

-alternative to repeated measures ANOVA

-look at each subject separately and rank order score

-for each measurement, sum ranks from all subjects

frequentist stat

Frequentist statistics works by calculating a test statistic that measures how much the data deviate from the null hypothesis, working out the sampling distribution for the test statistic when the null hypothesis is true, and

setting a rule for retaining or rejecting the null hypothesis that controls the Type I error rate.

bayesian stats

-assume we know the sampling distribution for the test statistic according to both hypotheses

likelihood

-probability of getting an answer to reject the null

-prior probability

-how probable it is before you see any of the data

posterior probability

-how probable it is after you have seen it

baye's rule

-mathematical formula that calculates posterior prob from prior prob and likelihood

"The semester I found StudyBlue, I went from a 2.8 to a 3.8, and graduated with honors!"

Jennifer Colorado School of Mines© 2015 StudyBlue Inc. All rights reserved.