Psych 2300 Chapter 9 Notes: Multivariate Designs
Psych 2300 Chapter 9 Notes: Multivariate Designs Psych 2300
Popular in Research Methods In Psychology
Popular in Psychlogy
This 5 page Class Notes was uploaded by MadsSwart on Friday July 8, 2016. The Class Notes belongs to Psych 2300 at Ohio State University taught by Emanuele Rizzi in Summer 2016. Since its upload, it has received 67 views. For similar materials see Research Methods In Psychology in Psychlogy at Ohio State University.
Reviews for Psych 2300 Chapter 9 Notes: Multivariate Designs
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 07/08/16
Chapter 9: Multivariate Designs - Means many variables - Study more than 2 variables in a single study - Enhance understanding of bivariate associations Longitudinal design - Measures same variables and same sample at several time points - Use to assess variable relationships across time - Use to see variable stability across time - 3 types of correlations assessed o Cross sectional: two variables measured at the same point in time [bivariate correlations] o Autocorrelations; testing variable against itself measured at two time points. o Cross-lag correlations; early measure of one variable against later measure of the other variable Moderation Language - Violence on TV vs. violence on TV with the parents - As longitudinal design; cross-sectional correlation o TV violence 3 grade & aggression 3 grade : r=.21 th th o TV violence 13 grade & aggression 13 grade : r=.05 - Auto correlational o TV violence 3 & TV violence 13 grade : r=.05 o Aggression 3 grade and aggression 13 grade : r=.35 - Cross-lag correlation rd th o TV violence 3rdrade & aggression 13 gradth: r=.31 o Aggression 3 grade & TV violence 13 grade : r=.01 Tells us that violence came first Aggression doesn’t lead to them watch violence Watching violence leads to aggression Interpreting Longitudinal Designs - 3 possible outcomes o A at time 1 b at time 2 o B at time 1 X a at time 2 A predicts b Temporal precedence is established o Reverse of previous outcome o A predicts B and B predicts A Cyclical Cycle - 3 possible outcomes o TV violence predicts aggression o Aggression predicts TV violence preference o Both predict each other; cyclic relationship - Temporal precedence is dealt with BUT third variable problem still exists Multiple Regression rd - Dealing with 3 variables - Multivariate regression - Test several variables’ relationship with a/an outcome variable - Implies that other variables have an effect in the outcomes we are seeing - NOT AN EXPERIMENT - Have criterion; you think is effected by something else; dependent variable - Predictor(s) – as many as you want; things that effect the criterion; independent variable - R for each predictor with criterion Multivariate designs - Test correlation between two variables while ‘controlling for’ other variables o Is a related to b when c and d are held constant o Is a related to be for all levels of variable c and d - Statistically: what proportion of variability in DV is accounted for by each predictor? - Does correlation between a & b hold for all subgroups of a particular type? Multiple Regression Model - Beta (β) o Similar measure to r [-1-1] o Standardized value; allows to compare apples to oranges as ‘fruits’ o Strength and direction of relationship WHEN you are controlling for the predictors held at a CONSTANT value - b value o unstandardized slope o keeps measurement in original measured units o more meaningful to interpret behavior problems decrease by x points for every minute of recess children have o This beta means that IV is/is not associated with DV such that as values of IV increase/decrease values of DV increase/decrease when controlling for predictors IV, IV, IV, and IV… Predicting DV: Violence Beta (N=348) (sig) Interpreting Regression Models Peer’s deviance .56** The only reason French rap predicts Violent media .16** Lyrics’ Importance .09* violence is because the kids who listen American Rap -.05 to rap music are just more into violent (ns) media in general. I bet they watch French Rap .11** more violent TV and stuff like that. Hip hop/soul -.07 (ns) Gangsta/Hardcore rap .08 (ns) Words implying multiple regression o Controlled for o Taking into account o Correcting for o Adjusting for o Keeping x constant Tells you it’s both an association claim and that there are multiple variables - Multiple Regression ≠ Causation o You don’t know what relevant variables you have not accounted for o Experiments actually control variables - Patterns & Parsimony o A good scientific theory offers the simplest explanation - Pattern of Results: Smoking o How do we know smoking causes cancer Create a simple theory with a causal mechanism Use multiple regression and longitudinal studies to test all possible factors Mediation - Finding a causal path o Not the method you use that determines the kind of study you have, but its what you measure and what you manipulate o WHY/HOW a relationship exists between two variables Mediation IS a causal statement they imply covariance, temporal precedence, and internal validity o specify a mechanism for relationship tested using multivariate analysis o the path the causal variable has to take to affect the criterion/dependent variable implies that there is NO direct relationship between predictor and criterion (IV and DV) R=.6 Psychical Activity R=-. 3 Availability of R=-.1 Behavior recess 5 Problems - Happiness and longevity; mediation effect o Can 1 million women be wrong about happiness and health? I was not the method [multiple regression] that was wrong, but rather the INTERPRETATION that was incorrect They assumed happiness played no role on behavior, when it did. What they thought they had Happiness Pos. Report r Longevi Pos. Happiness r Report Longevi Other variables like R=0 What they actually had Pos. Other variables like r Pos. R= r Happiness 0 Longevi Report Mediation, Moderation, or Third Variable Problem rd Mediation Moderation 3 Variable Problem Why are 2 variables When, for whom, or under 2 variables are correlated, related? what conditions are 2 because they are both linked to variables related? a 3rd variable. A is related to B because… A is related to B for one typeA is related to B but only of C but not for the other because… type of C A leads to C which leads to C type 1 : A is related to B C and A are related & C and B B C type 2: A is not related to are related B Viewing violent TV Viewing violent TV & A=viewing violent tv desensitized to violence aggressive behavior have B=aggressive behavior aggressive behavior; r=.35 in boys & r=.1 in girls C=having lenient parents viewing violent TV aggressive behavior MAIN POINTS - Multivariate designs can help us address weaknesses in bivariate associations o Gain further insight on relationship of variables of interest - Longitudinal designs help deal with temporal precedence o Cross-sectional, auto correctional, cross-lag - Patterns of correlations suggest nature of variable relationships - Multiple regression deals with 3 variable problems o Do relationships between variables still exist after taking into account other variables? - Compare effects of different variables using beta’s - Mediation is a utilizing multiple regression to assess a ‘causal pathway’ between variables o How does variable a cause variable b Exercise - Studies have shown that maternal responsiveness (how quickly a mother responds to her infant’s crisis) is associated with the fussiness of the baby. Specifically, more responsive mothers have less fussy babies. But which comes first? The responsiveness of the mother or the fussiness of the babies? o Design a longitudinal correlational study to test this question o Map out the different pairs of correlations you might fight Cross-sectional correlations Autocorrelations Cross-lag correlations o Which pattern of cross-lag correlations: Would suggest that responsiveness causes less fussiness? Would suggest that infant fussiness causes less responsiveness? Would suggest that these two variables are in a mutually reinforcing relationship?
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'