I want to calculate a mediation path model with my data using lavaan. This looks like the following:
CLT.model <- " Factual.knowledge ~ a*Condition + b* Quali.Cooperation + c* Cognitive.Activation + d*Previous knowledge. Conceptual.knowledge ~ e*condition + f*quali.cooperation + g*cognitive.activation + h*prior knowledge Quali.cooperation ~ i*condition + j*attitude.to.cooperation Cognitive.activation ~ k*condition Cognitive.activation ~~ Quali.Cooperation Conceptual knowledge ~~ Factual knowledge" clt.fit.boot <- sem(CLT.model, data = data.clt, missing="fiml", se="bootstrap", bootstrap=5000) parameterEstimates(clt.fit.boot, ci=TRUE, level = 0.95, boot.ci.type = "perc")
However, my problem is that my data is partially dependent on each other. The people have worked together in groups of two. I would like to control this dependency, but I don't know exactly how this works. Our method consultant suggested using block bootstrapping, but was unwilling to help me with the programming. I have been able to program some of it myself, but I can't get the index variable to work properly (examples on the internet are always based on simpler models "lm" and time series). Does anyone have experience with Block bootstrapping and could help me with the programming?
Or: Is there another way to control the dyadic dependency in my model?
I hope someone can help me!