Hello everyone,
A small query.
I am conducting a factor analysis (covariance matrix based rather than
correlation matrix based)......I have noticed that the communalities of
some variables are greater than 1. For example:
Var Factor1 Factor2 Factor3 Factor4 Communality
1 1.063 0.106 0.242 -0.057 1.202
2 1.047 0.690 0.078 -0.029 1.580
3 0.977 -0.186 0.150 0.116 1.026
4 0.799 -0.420 0.122 -0.156 0.853
5 0.753 -0.268 0.300 0.085 0.735
6 0.731 0.141 0.289 -0.343 0.754
7 0.484 -0.113 0.436 -0.024 0.437
8 -0.181 0.941 0.004 0.226 0.968
9 0.139 -0.155 0.897 0.011 0.849
10 0.259 0.362 0.718 -0.004 0.713
11 0.028 -0.232 -0.025 -1.119 1.307
5.2725 1.8965 1.7852 1.4697 10.4239 Variance
0.380 0.137 0.129 0.106 0.752 % Var
Say we extract 4 factors, the communalities are defined as 'the
proportion of variance of each variable explained by the 4 factor model'
so say the communality of a variable was 0.657, this would mean that
65.7% of the variance in this variable was accounted for by the 4 factor
model......so in my case what interpretation should be made when the
communality is greater than 1?
Many thanks,
Kim.
|