Being a Stat Prof I will say yours is right. ! look at your reseach question. Since the same dv is in each all covariates go in at once. Contrasts because you want to compare means(ANOVA model form ) or regression coeffs(regression model form).I am NOT an SPSS guy so I would run this as a regression which is easier to interpret IMO. For full details see:
https://www.google.com/search?q=Montgomery+Design+of+Experiments&rlz=1C1CHBF_enUS874US874&oq=Montgomery+Design+of+Experiments&aqs=chrome..69i57.17978j0j1&sourceid=chrome&ie=UTF-8
Best, D. Booth
Hello Esra,
I noticed that your four questions do not include two of the IVs (candidate status and experimental condition). In light of this, I agree with David Eugene Booth 's suggestion that OLS regression might be a simpler, straightforward solution.
To determine if a given "covariate/IV" does or doesn't help explain differences on the dependent variable (given the others in the model), just look at the test of the four individual regression coefficients (these will be posted as t-tests in spss regression output).
If you elect to run the ancova as described in your query, the relevant tests are for the IVs you mention. The covariates (ed_level, etc.) will each yield a 1 df F-test, and the lone IV that is part of your four questions (time/pressure) will also yield a 1 df F-test.
Good luck with your work.
ANCOVA AND REGRESSION ARE EXACTLY THE SAME STATISTICALLY. I generally choose the easiest model to interpret. If you are testing interactions(as you appear to be) the regression model is usually easier IMO. Never forget to look at the interaction plot. The Montgomery reference gives full info on both approaches. The attached gives SPSS examples. Best, David Booth
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA