Psych 395 Cooper Fall 2009 Sept8.docx

Psych 395: 9/8/2009 Correlation = biggest way to study in psych Helps make predictions *Data from class survey shown* Correlation = assign numbers (even if numbers have no value -like M/F) Correlation Coefficient: tells us direction + strength Positive corr./negative corr. = direction Positive = variables move in same direction as _____ increases, _______ increases As _____ decreases, ______ decreases Ex: the more pets she has, the happier she gets Ex: the less pets she has, the less happy she is Negative = variables move in opposite directions As _____ decreases, ______ increases As ______ increases, ______ decreases Ex: the more siblings someone has, the less happy they are Ex: the more happy they are, the less siblings they have -1 <==> 0 <==> +1 Strength = absolute value of correlation Closer to |1| = stronger (-1 or +1) Closer to 0 the weaker + = positive correlation - = negative correlation Question: which of the following 2 correlations is stronger? # showers per week + gpa = .09 # miles walked per week + gpa = -.27 Which is stronger? The miles + gpa? *remember, positive & negative signs don?t matter when figuring whats a stronger correlation? its which is closest to 1!* Correlations show if 2 variables are related Use data & plug into scatter plot 1 var = Y; other var = X Ppl stats = dots Best fit line: Linear relationship (+ or -) will hold throughout scale Perfect corr. = dots on line More ppl who fall off line less & less able to make predictions Third variable can alter relationship Not just 2 affecting each other We = ?barney generation? Research: kids with great self-esteem do well in school Changed parenting Some did poorly but ok b/c good self-esteem? so not correlation but third variable: directionality issue Correlations = might cause & effect, but don?t know which = drivers seat Ex: bring up grades good self esteem vs. barney generation Pellegra: correlation + causation lesson? Poor sanitary conditions Pellagra contagious + must be quarantined 1 didn?t believe: took fluids + scabs of sick, baked, ate prove not contagious so must be third variable? low protein diet When corr. Sounds like it makes sense, the more likely mistakes will be made Multiple regression/correlation: try think of all possible third variables when studying/making correlation Why scatter plot? B/c Pearson Correlation Product only works for linear relationships & the scatter plot will reveal if linear or not? Curved = not linear = doesn?t work?