Analysis of Categorical Data with the R Package confreq

Round 1

Reviewer 1 Report

The authors provide a thorough tutorial on Configural Frequency Analysis and log-linear modeling. The R code is straightforward and works without error. The descriptions of the data and functions are documented enough to replicate and understand what is doing what.

I'll be sharing this R package with my colleagues who are interested in these analyses.

Well done!

Author Response

Thank you!

Reviewer 2 Report

Please see my comments in the attached document. 

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

Thank you for the positive feedback and for the numerous suggestions. 

All notes on typos and notes on commas / periods after Formulas were corrected as noted -- Thank you very much for pointing out to these.

To provide a point-by-point response I copied your review comments from the PDF here. My Responses / Notes are in red 
 
"Factor analysis does not necessarily assume linear relationships."

Response: Thank you for this hint - we agree that there are at least variants of a factor analysis without this prerequisite and have therefore taken out the word linear here. In essence, this sentence is only a contrast to show that the variable-oriented approach (as the name says) focuses on the associations between variables, whereas the person-centered approach focuses on inter individual differences from characteristic configurations. 
 

"What if some patterns don't occur?"

Response: Thank you for pointing out this ambiguity that was still present in the text. In fact, it is important to point out that in the tabulated data all possible configurations are always listed first -- if some of the configurations are not observed in the empirical data, the corresponding configurations get the absolute frequency zero. To clarify this we have added a footnote: 'this implies that in some cases the observed frequencies also might be zero for patterns which are not observed in the empirical data'.
 

''Put comments before the syntax."

Response: Indeed, some longer syntax comments after the actual R commands made some R examples difficult to read. this is now fixed. 
 

"Is this a fixed-with font? The part where chi square values df etc. reported looks a bit messy. " (Comment on the presentation of the R output)

Response: All R example syntax and the corresponding displays of R outputs are now set in fixed-width font -- so the displayed outputs correspond exactly to what a user gets when executing the syntax example.

 
"Can you elaborate on what these options do?" (Comment on the options for the 'type' argument in the summary function)

Response: Thank you very much for this advice. We see that there was some confusion about the options for the 'type' argument. In order to clearly assign the options to the different test statistics, we have adjusted the previous sentence (line 356, page 9 in the revised manuscript) accordingly. In addition, we have referred to a textbook on CFA in which the different test statistics are explained in detail. 
 

"I would appreciate a bit more contextual information based on the results being represented in the paper. In the context of the Lienert-LSD data, how can one interpret the findings based on the CFA analysis? Some brief guidance to readers would be great!"

Response: Thank you for pointing out a missing (comparative) interpretation of results from different CFA models. Even if, in particular, the findings in the result objects res3, res 4, res5 in r snippet 8 are already interpreted above at least briefly comparatively, we agree that something can be added here. The interpretation of the findings on structural extreme cells (quasi independence model) seems to be important here. We have therefore added a short section from text line 554 (page 12 in the revised manuscript) that explains the findings. 

"What is the advantage of confreg over other categorical data analysis packages in R?"

"Are there any other packages that could run similar analyses in R or other software programs?"

"Readers need to understand why they would need to use confreq instead of another package in R (also why CFA instead of other categorical data analysis methods). "

Response:  Thank you very much for these suggestions - indeed, these are important points for the summary and conclusion - in turn, we have added a few sentences to this section that address these points (between text line 828 and 841, page 18 in the revised manuscript).