We analyzed the data of two years of field experiments by analyzing each year separately, but the editor said that “It is not clear why years were analyzed separately. Combining years and analyzing data with mixed models would generate more robust conclusions.”
Can I know, is the combined-over-year analysis of variance is possible in this case even we do not have the same results in the two experiment years? And how we can do this combined analysis using SPSS?
Just write your ANOVA model as a regression and add a 0,1 dummy variable for year. The dummy regression coefficient is the test for a difference between the two years. See the attached screenshot for details on doing this. Best wishes, David Booth
In this case, the suggestion “mixed models would generate more robust conclusions” is correct.
What does it mean the statement “even we do not have the same results in the two experiment years”? If we do not have the same results (but same variable measured), it is why these models apply!
Hello Abdellatif Boutagayout. I have several questions, some of which were also raised by Roberto Molteni.
Thank you for clarifying.
Edit: Re point 3 above, please show us command syntax for the two analyses you did.
Roberto Molteni
Dear Professors, thank you for your response and consideration.
the statement “even we do not have the same results in the two experiment years” means that repeated measures of ANOVA indicated that there was a significant difference between the data of the experiment years.
Bruce Weaver
Dear Professors, thank you for your response and consideration.
As responses to the questions :
Thank you again.
Thanks for responses to questions, Abdellatif Boutagayout. In one response, you said that you have "repeated measures" over time. In other words, the results from the two years are not from two independent samples. Rather, the DV is measured twice for each independent unit of observation. Is that right? In other words, if Time was the only explanatory variable, you would have have a paired t-test rather than an unpaired (independent groups) t-test. Right?
If that is correct, your two main options in SPSS for analyzing repeated measures are GLM REPEATED MEASURES (with Time as a within-Ss factor) and MIXED (as suggested by the editor). The UCLA website has some nice pages you can review for more info:
- https://stats.idre.ucla.edu/spss/seminars/repeated-measures/
- https://stats.idre.ucla.edu/spss/seminars/spss-mixed-command/
Suppose that A is one of your focal explanatory variables. To ask if the effect of A changes over time, you would need to include the Time x A interaction in your model. HTH.
- Badges
- Science topic