- Joined
- Jan 14, 2010
- Messages
- 149
- Reaction score
- 0
So I am translating a measure from English to Spanish. As part of the translation procedure (after a series of back translations and revisions etc), my PI wants me to give *both* measures to a sample fluent in both English and Spanish (i.e., bilingual). This will be a counterbalanced within-subjects design. Thus, some individuals will receive the English version first and others the Spanish version. Initially we discussed just looking at the correlation between the two versions (which would hopefully be nice and high if the translations are pretty close).
However, recently he mentioned looking at order analytically... And of course, here is where I got stuck conceptualizing the analyses in my head 🙂.
I'm used to looking at order effects in experimental studies where you're essentially comparing means [e.g., You have two conditions A and B that all participants get. Thus some get A-B (group 1) and others B-A (group 2). Your DV is depressive sxs let's say]. You then end up doing a repeated measures ANOVA, with condition x group (within x between), to see if there were any order effects (i.e., you'd probe a significant interaction). [Side note: although I have seen people out there, once they counterbalance, just lumping things together...].
Ok. So I get that conceptually.
But in my case... how would I handle order effects? The same way? or...?
A peer told me that I would do a correlation/regression in group A (i.e., english version scores predicting spanish version scores) and another one in group B (i.e., spanish version scores predicting english version scores). And then see if those two correlations are significantly different from each other.
Does this sound right to others on here who may be more familiar with this?
Or... would I just test means as in the example given above with two conditions... I guess the interaction being language x group.. in ANOVA?
Ugh. Sigh. I think I have confused myself.
(My PI is out of the country at the moment... thus, hard to get in touch with to clarify how he wanted to look at order effects. So *any* input is appreciated - as I feel the answer is just looking me in the face).
ps - I will be testing the validity/reliability of the Spanish measure after the translation is finished in multiple samples using other statistical techniques... and the funny thing is... I understand those analyses (which seem more complicated) more easily than this! *facepalm*
However, recently he mentioned looking at order analytically... And of course, here is where I got stuck conceptualizing the analyses in my head 🙂.
I'm used to looking at order effects in experimental studies where you're essentially comparing means [e.g., You have two conditions A and B that all participants get. Thus some get A-B (group 1) and others B-A (group 2). Your DV is depressive sxs let's say]. You then end up doing a repeated measures ANOVA, with condition x group (within x between), to see if there were any order effects (i.e., you'd probe a significant interaction). [Side note: although I have seen people out there, once they counterbalance, just lumping things together...].
Ok. So I get that conceptually.
But in my case... how would I handle order effects? The same way? or...?
A peer told me that I would do a correlation/regression in group A (i.e., english version scores predicting spanish version scores) and another one in group B (i.e., spanish version scores predicting english version scores). And then see if those two correlations are significantly different from each other.
Does this sound right to others on here who may be more familiar with this?
Or... would I just test means as in the example given above with two conditions... I guess the interaction being language x group.. in ANOVA?
Ugh. Sigh. I think I have confused myself.
(My PI is out of the country at the moment... thus, hard to get in touch with to clarify how he wanted to look at order effects. So *any* input is appreciated - as I feel the answer is just looking me in the face).
ps - I will be testing the validity/reliability of the Spanish measure after the translation is finished in multiple samples using other statistical techniques... and the funny thing is... I understand those analyses (which seem more complicated) more easily than this! *facepalm*
