I am one of Martha's supervisors for this PhD programme and we would be really grateful if someone could help with suggesting an approach to power calculations probably using R for what will need to be a 4-way (2 x 2 x 2 x 2) loglinear analysis.

Power calculations are not within my expertise when complex.

One of the dichotomous variables effectively functions as both a factor and a DV (juror verdict: guilty vs. not guilty) in a loglinear model. The outcomes on this variable will clearly vary, unlike the other factors as Martha can ensure approximately equal allocation to those groups (three of the four factors).

Martha needs sufficient power for interaction effects, not just the main effects.

We were looking at expanding on chi-squared analyses, but it is the varying 4th factor that makes prediction complex.

We know loads of participants are required. We think (hope) this is manageable.

Thanks in advance.

Josh

Martha Babbs Did you ever get an answer to this? I have exactly the same question.