Do i need a two level model? PreviousNext
Mplus Discussion > Structural Equation Modeling >
 victor china posted on Saturday, July 21, 2007 - 8:45 am
i have a data set that is based on behavioral observation, i have around 1800 single interactions between individuals that are taken from a total of 130 observations (each observation consists of 10-20 single interactions, also in each observation i had conditions that were somewhat different).

i did a path analysis model in which i added the different conditions of each observation as an editional variable for each single interaction, (is it methodologicaly appropriate?)

do you think that introducing the different conditions of each observation whould be better through a two-level model?

also i must add that the amount of interactions is not constant for each observation.
 Linda K. Muthen posted on Saturday, July 21, 2007 - 12:20 pm
I am not clear on the structure of your data. How large is your sample?
 victor china posted on Saturday, July 21, 2007 - 1:45 pm
sample size is 1800 single interactions.
the data collection is done by a continuous sesion that lasts for 15 minutes in which between 10-20 signle interactions occure.

Each session has its own characteristics that can be described by a single variable (lets call it N).

I am interested to test how the effect of N effects each signle interaction.

So the question is can i simply add this variable to my path analysis as one of the variables that i use to describe each single interaction or do i need to test it through a two level model.
thanks in advance.

 Bengt O. Muthen posted on Saturday, July 21, 2007 - 5:14 pm
Is a "single interaction" an interaction between two people? The first thing to settle is what the independent sample unit is. If you consider interactions between two individuals, the independent sample unit is the pair and the 10-20 interactions can be seen as multivariate variables observed for the pairs. That leads to a single-level analysis. But I may not understand your situation. See also the Dagne et al article on our web site on dyadic data analysis.
 victor china posted on Saturday, July 21, 2007 - 9:36 pm
thanks very much, the "single interaction" is a behavioral session between a cleaner fish and its client fish :-).. it is indeed the independent sample unit.

i think that a single level analysis is appropriate here also.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message