Dear Statalist,
I have experimental data on how individuals set up and maintain financial portfolios. They are endowed with an initial capital and have the choice between four asset classes.
Now, my task is to analyze which portfolios are set up.
Initially, I analyzed decisions across the asset classes, like in a mlogit model, i.e., newportfolio in asset class 1 = 1 if the new allocation in that asset class is below the original in that asset class, etc.
However, this approach yielded the problem that I still don't know what individuals overall did --> this is just an individual asset class analysis ( I can say 25% of the individuals performed such and such action). Further, I don't know which asset class combinations are accurate (this asset class reveals nothing about another asset class).
Since I have 250 individuals, would it make sense to group the new portfolio on an individual level, or are summary statistics the way to go, i.e., the mean, min, max, stdev - like in https://www.statalist.org/forums/for...stics-by-group over the different treatments, participant age, etc.
I would be most grateful for any help or advice!
Thank you in advance!
I have experimental data on how individuals set up and maintain financial portfolios. They are endowed with an initial capital and have the choice between four asset classes.
Now, my task is to analyze which portfolios are set up.
Initially, I analyzed decisions across the asset classes, like in a mlogit model, i.e., newportfolio in asset class 1 = 1 if the new allocation in that asset class is below the original in that asset class, etc.
However, this approach yielded the problem that I still don't know what individuals overall did --> this is just an individual asset class analysis ( I can say 25% of the individuals performed such and such action). Further, I don't know which asset class combinations are accurate (this asset class reveals nothing about another asset class).
Since I have 250 individuals, would it make sense to group the new portfolio on an individual level, or are summary statistics the way to go, i.e., the mean, min, max, stdev - like in https://www.statalist.org/forums/for...stics-by-group over the different treatments, participant age, etc.
I would be most grateful for any help or advice!
Thank you in advance!
