[Rd] Eigenvalue Bug in "psych" "principal" (PR#14010)

[Duncan Murdoch]

On 10/15/2009 3:15 PM, kyler at smu.edu wrote:
> Full_Name: Kyle Roberts
> Version: 2.9.2
> OS: Windows XP
> Submission from: (NULL) (129.119.244.130)
> 
> 
> In 2.9.2, the Eigenvalues from a principal components analysis using "principal"
> in the "psych" package no longer works. It still works in 2.8 and 2.7, but no
> longer in 2.9.2. For example:

This is not a bug report about R.  You should contact the maintainers of 
the packages involved.

Duncan Murdoch

> 
>> library(MBESS)
>> data(HS.data)
>> library(psych)
> 
>> #### Subset the dataset to include only 8 variables
>> new.data<-subset(HS.data, select=c(deduct, numeric, problemr, arithmet, 
>   paragrap, sentence, wordc, wordm))
> 
>> pc2<-principal(new.data, nfactors=2, rotate="varimax")
>> pc2
> Principal Components Analysis
> Call: principal(r = new.data, nfactors = 2, rotate = "varimax")
>          item  PC1  PC2   h2   u2
> deduct      1      0.68 0.52 0.48
> numeric     2      0.82 0.69 0.31
> problemr    3 0.45 0.54 0.49 0.51
> arithmet    4      0.67 0.54 0.46
> paragrap    5 0.84      0.76 0.24
> sentence    6 0.89      0.83 0.17
> wordc       7 0.76      0.65 0.35
> wordm       8 0.81 0.30 0.75 0.25
> 
>                 PC1  PC2
> SS loadings    3.09 2.13
> Proportion Var 0.39 0.27
> Cumulative Var 0.39 0.65
> 
> Test of the hypothesis that 2 factors are sufficient.
> 
> The degrees of freedom for the model is 13  and the objective function was  0.36
> 
> The number of observations was  301  with Chi Square =  104.87  with prob < 
> 1.9e-16 
> 
> Fit based upon off diagonal values = 0.97
> Measures of factor score adequacy             
>                                               PC1 PC2
> Correlation of scores with factors              1   1
> Multiple R square of scores with factors        1   1
> Minimum correlation of factor score estimates   1   1
>> 
>>   ## Eigenvalues
>>   pc2$values
> NULL
> 
> ####################
> It still works in the principal axis factoring package in "psych", but no longer
> with "principal".
> 
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel