See the attached paper

Dear @Seyyed I wonder what did you mean by controlling the moderating variable? Are you asking whether a control variable (age, gender, etc.) can be used as a moderator? Please shed a little more light on your query so that others can provide a better feedback/suggestion.

Hello Seyyed,

I concur with Imran. It might be helpful to know more about your needs and question. Aside from that, I hope this may help (and I realize you may know this already). I always consider "control" variables as variables that are held constant or whose effects are "removed" in an analysis (e.g., is X associated with Y, while removing the effect that M has on Y). (These are mere phrases and not necessary statistically what is occurring). This could perhaps be done by entering this control variable in the first step of a regression or as a covariate in an ANCOVA. Moderation, while similar, more often is designed to assess how the levels of M (moderator) impact the association of X and Y. I hope this helps.

Dear Imran and Cristopher

I concur with both of you! I really do not know what exactly controlling moderating variables means! Actually, "how did you control moderator variables in your study?" is the question that my Ph.D. application interviewer asked me.

I considered testing administration mode (Paper-based & Computer-based) as the IV to determine its effect on the testing performance of test-takers as the DV by considering the correlation of some moderator variables including computer familiarity, attitudes towards the use of computer in testing, test takers' anxiety towards the use of the computer, age, gender, testing administration mode preference and testing administration mode order or sequesnce with test takers' testing performance. I considered the aforementioned moderator variables to determine how they might affect and moderate the relationship between IV & DV. I examined the correlation of the moderator variables with testing performance to find out whether the effect of IV on DV might be moderated by these external variables. I did not control them! As far as I know, control variables are controlled to eliminate their effects in a study, not moderator variables. Do you agree with me?

I feel the question was not the right question to ask.

It'll be kind of you if you offer me your technical and precious viewpoints

Regards

Imran Anwar

Christopher Modica

Dear Seyyed Morteza Hashemi Toroujeni you wrote:"I examined the correlation of the moderator variables with testing performance to find out whether the effect of IV on DV might be moderated by these external variables." What has this to do with a moderator analysis? There is no need that a moderator has to be correlated with the DV in a zero order correlation.

And for the "control for" question: moderators may also be experimental manipulations, for example you could have had two experimental groups a) with time constraints and b) without time contraints for the task. You you test then if the relationshiop between your IV and DV changes with levels of time constraints, which you would have controlled. Maybe the reviewer thought about such a thing?

Dear Rainer Duesing

Yes, I think you are right. I just got the pint, and maybe the interviewer meant what you explained.

Hello Again Seyyed,

Thanks for clarifying things in your previous post. It feels to me that the question you were asked is perhaps not as precise. As a result, you are left with trying to figure out what was meant, which can be very difficult.

But, I agree with what you wrote. I again believe that moderator and control variables are very similar, yet different. By "controlling" for a variable (in either a correlation design or experimental design) you are really removing it's effects on the DV. Then, you can say the IV is associated with the DV "while controlling for" some other variable(s). In moderation you are examining whether a IV is associated with a DV at different "levels" of a moderator variable (e.g., high, medium, low). I guess in moderation you are in one sense "controlling" for the variable by accounting for it, but it is not "control" in a technical sense.

In my view, in an analysis, one either controls for variable OR examine if that variable moderates, but not both.

I agree with Rainer as well that conditions in an experimental study can also be moderating variables or perhaps control variables. By creating two additional conditions conditions (with time constraints vs without) (beyond the two conditions of your IV you already have - ie paper-pencil vs computer administration) you could "control" for time constraints within the design.

I hope this helps.

Chris

thanks for your clarification of the situation