Idil Ugurluoglu if I understand your question correctly, you need to have a variable that is coded one way (e.g., 1) if a participant saw the partner first and coded another way (e.g., 2) if a participant saw the stranger first.

Then that variable becomes the independent variable, depending on whether participants belonged to Group 1 or Group 2.

I hope I've understood your question / problem, however. Please feel free to get back.

thank you so much for your answer.

I coded one variable for sequence of seeing priming figure: firstly partner, then stranger=1, firstly stranger then partner=2. I measure affect after seeing partner then in another trial after seeing stranger and visa versa for the other condition. But at the end of a day, I have to enter one dependent variable for after partner and one dependent variable for after stranger on SPSS. I think there is a problem when I assume that both group saw partner first then stranger and create an independent variable for sequence. I hope I could explain my problem. Thank you for your kind interest. Robert Trevethan

Hello Idil,

Hmm, I can't quite picture the situation. Let me try this: I think you have done the right thing for your independent variable, which is coded either 1 or 2. Then, in your SPSS data file, you should have just a single dependent variable, and in that variable you enter what score each participant got on that variable. Then, if, say, you want to do t-tests, you compare Group 1 versus Group 2 on the (single) dependent variable.

Does that make sense?

Again, pls feel free to get back.

Perhaps you have more than one dependent variable. Might you be able to tell us what your dependent variable(s) is/are. That could make the picture clearer. But I think you have only one dependent variable, namely affect, right?

Idil Ugurluoglu I can't help but point out a split plot design would have solved this question as well as provided a more sure answer to your research question. As a practice, I think it is always better to consider a standard experimental design(see

https://www.google.com/search?q=Montgomory+Design+and+Analysis+of+Experiments&rlz=1C1CHBF_enUS874US874&oq=Montgomory+Design+and+Analysis+of+Experiments&aqs=chrome..69i57.42037j0j1&sourceid=chrome&ie=UTF-8 ) and depart from it only if you REALLY can't do it. My old canard about this is Proper Prior Planning Prevents Poor Performance. Which I think was the advice given to your previous question( https://www.researchgate.net/post/Should_I_involve_counterbalancing_as_an_independent_variable_to_my_design ). That said, I believe that you have adopted the quite reasonable answer provided by @David Morse there. Best wishes, David Booth

Hello Idil,

If sequence were randomly assigned, I'd use it (sequence) as a blocking factor in anova (or as another IV in a regression model; be sure to include all relevant interactions though). Worst case scenario is the loss of a couple of error df (if sequence effect is nil). Best scenario is reducing what otherwise could contribute to an inflated error estimate, making for a more sensitive test. If alternated, then split-plot would likely be better, as suggested by David Eugene Booth .

Good luck with your work.

Hello again,

I may not have been clear in the prior answer, so I'll try one more time!

1. The sequence should be included as an IV in your design.

2. Assignment to sequence should, ideally, be random.

3. The basic analysis design would be (in addition to whatever other factors/IVs you incorporated in your research): one-between subjects factor (sequence: A-B vs B-A) and one-within subjects factor (score on stimulus A, score on stimulus B). From the early days of Fisher and agricultural research, this design was originally referred to as a "split-plot" design.

4. If there is either a main effect for sequence or a sequence by measure interaction, then the order of presentation mattered for your sample.

Best of luck with your study.