3005PSY Survey Design & Analysis Final Exam Notes & Assignment Analyses
1
Table of Contents
❖ Bivariate Regression………2
❖ Multiple Regression………3
❖ Regression Diagnostics……….5
❖ Assignment Analysis Steps & Syntax………..7
❖ Single Case Design (N = 1)………..9
❖ Chi-Square (χ2)………11
❖ Longitudinal Designs………14
❖ Master Syntax File……….18
3005PSY Survey Design & Analysis Final Exam Notes & Assignment Analyses
2
Sample Notes
Multiple Regression:
• Syntax:
o *Multivariate Regression + Residiuals Normality Check*.
o regression var= variable1 variable2 variable3 variable4 variable5 o /statistics= defaults zpp
o /dep= dependentvariable o /enter
o /residuals = histogram(zresid) id(ID) o /scatterplot (*zresid, *zpred).
• Sometimes referred to as ordinary least squares regression (OLS)
• Extends the least squares procedure to estimate 𝑏0 and 𝑏1 and 𝑏2 to provide prediction of Y when considering all variables jointly
• Least squares estimates give us the lowest possible value of e when e is squared and summed over all data points
• B-weights:
o Numbers we multiply each X to make a composite X o These are now partial slopes
o Each b tells us how much change in predicted Y there will be for a change of 1 in that X when all other X’s are held constant
o Slopes with interrelationships of other predictors accounted for
• Multiple regression formula:
o Y’ = 𝑏0 + 𝑏1𝑋1+ 𝑏2𝑋2 + e
o E.g. Y’ = 36.13 + .053GRE-Q + 12.157Attendance + error o Still using B-weights from SPSS
o 𝑏0 = Constant
o 𝑏1𝑋1 = Variable 1 B-Weight o 𝑏2𝑋2 = Variable 2 B-Weight o In Coefficients section in SPSS
• For every point that someone scores on X, their predicted score on Y will increase by the slope when holding constant the other variables
• 𝑅2 = The amount of variance accounted for by all predictor variables o Not the sum of the two
o Called a joint quantity
o Also, an indicator of model fit o Found in “Model Summary” in SPSS
• Steps in interpreting multiple regression:
o Overall F ratio: The significance of F in the ANOVA section of output
o Multiple R & 𝑅2: 𝑅2 = proportion of variance in criterion variable (Y) accounted for by the set of predictor variables
o Bs, βs, and the test of Bs:
▪ Unstandardised vs. Standardised weights
▪ Strength/significance of each assessed using t-tests of B-weights in Coefficients section of SPSS
o Zero-order (r), part (pr), and partial correlations (sr):
▪ Zero-order (r) correspond to Pearson’s product-moment correlation values
▪ Part (sr) shows semipartial correlations between each predictor & criterion