Boolean factor analysis
Hi.
I am investigating how different organizations work to further responsible practices in innovation and research. All my variables are binary e.g. they have a code of conduct (1) or not (0), they engage in science education actiities (1) or not (0) or they plan citizen science initiatives (1) or not (0). There are 21 of such variables. I have a dataset with 217 organization across 16 countries and five subtypes (university, company.. etc.). Now my ultimate goal is to make an (exploratory) cluster analysis to see whether patterns exists in the data (e.g. that some type of organization uses specific mechanisms to further responsible practice). Now I found moderate correlations between some of the variables which may be a problem in the cluster analysis, so I decided to run a factor analysis, investigating whether some of these variables are measuring the same underlying concept - which is probable. As I understand, a simple Principal Component Analysis is not appropriate when dealing solemnly with binary variables and that I instead should seek out a Boolean factor analysis. Now the questions is - which one and how? Does anyone have good suggestions for how to do this in Stata and preferably also with information on interpretation of the output. I would VERY much also appreciate suggestions for literature on the area and comparisons of different methods; book chapters as well as articles!
Thank you in advance.
I am investigating how different organizations work to further responsible practices in innovation and research. All my variables are binary e.g. they have a code of conduct (1) or not (0), they engage in science education actiities (1) or not (0) or they plan citizen science initiatives (1) or not (0). There are 21 of such variables. I have a dataset with 217 organization across 16 countries and five subtypes (university, company.. etc.). Now my ultimate goal is to make an (exploratory) cluster analysis to see whether patterns exists in the data (e.g. that some type of organization uses specific mechanisms to further responsible practice). Now I found moderate correlations between some of the variables which may be a problem in the cluster analysis, so I decided to run a factor analysis, investigating whether some of these variables are measuring the same underlying concept - which is probable. As I understand, a simple Principal Component Analysis is not appropriate when dealing solemnly with binary variables and that I instead should seek out a Boolean factor analysis. Now the questions is - which one and how? Does anyone have good suggestions for how to do this in Stata and preferably also with information on interpretation of the output. I would VERY much also appreciate suggestions for literature on the area and comparisons of different methods; book chapters as well as articles!
Thank you in advance.
Comment