Using SPSS—ANOVA—with NHIS 02, 02/09
Return to Opening Page of PSYC 4650
ANOVA—For statistical comparison of means of several groups.
Analyze--Compare Means—One Way ANOVA
Dependent (e.g., Adjust)
Factor (e.g., Diagnostic Groups 1,2,3; or Race Recode-White, Black, Other)
Options-Statistics-Descriptive, Means plot--Continue
OK
Post hoc Tests If you find significance with your ANOVAs, you can run a post hoc test to pin point the sources of difference and make multiple comparisons of the significance between all the different means.
Rerun ANOVAs as above but now chose
Post hoc—Tukey—Continue—
OK
TWO WAY (FACTORIAL) ANOVA For statistical comparison of means of two (or more) factors (e.g., diagnosis-ADHD vs. no ADHD; family type-2 parent vs.1), and to determine if there are interactions, (i.e., effects beyond the two factors by themselves but due to the combination of the two factors). E.g., Children without ADHD and children from 2 parent families have better adjustment scores, but children with both no ADHD and 2 parents have even better scores than expected from either factor alone
Analyze—General Linear Model--Univariate
Dependent variables (e.g., Adjusting well)
Fixed Factors (e.g., ADHD vs. No ADHD, Family type)
Plots-Horizontal-ADHD
Separate lines –Family type—Add
Options
Estimated marginal means—move all to—Display means for
Check off—Descriptive Statistics
Continue—OK
REMEMBER
First set missing values.
Use a scale for the dependent variable.
Use categorical variables for fixed factors.
Select cases and use a random subset with equal Ns for the main fixed factor.