Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Exploratory factor analysis using a mix of categorical and continuous variables


From   Urmi Bhattacharya <[email protected]>
To   [email protected]
Subject   Re: st: Exploratory factor analysis using a mix of categorical and continuous variables
Date   Fri, 3 Feb 2012 17:33:25 +0530

Hi Nick,

Thank you for sharing your thoughts on this. It was most helpful. You
are right in the sense that I want to use factor analysis as an
exploratory tool. The objective is to see if the 16 variables I am
using to proxy  what I call "school quality" are measuring the same
aspect of school quality. In that case I can use factor analysis to
see if the job can be done by a fewer number of variables.

Best

Urmi Bhattacharya

On Fri, Feb 3, 2012 at 3:22 PM, Nick Cox <[email protected]> wrote:
> I don't see that this can be easily classified as correct or incorrect.
>
> Some people recommend strongly against such a mix, and some people
> would argue that the pragmatic defence is whether it provides
> interesting or useful results.
>
> As you yourself have labelled the exercise "exploratory" the acid test
> is surely what do you learn from the data by examining the results.
>
> This is a cross-disciplinary list and we can only report our own
> perspectives. In what is nominally my own discipline, geography,
> factor analysis in the sense of an exploratory exercise throwing all
> the data into one pot, stirring and seeing what you got, was probably
> the most popular technique of all in the late 1960s and early 1970s. I
> met several people whose one statistical idea was to read everything
> into SPSS and do a factor analysis. I even met some people who did not
> know that there were simpler statistical techniques. In geography this
> fashion faded rapidly as too many people did not understand what they
> were doing or found no useful new results. However, I am now touching
> on quite different stories.
>
> In terms of your question, my only guess is that from your variable
> names you have a ragbag here and you won't find much interesting or
> useful structure. It is better to decide what are your response or
> outcome variables that you most want to explain or predict and think
> how those might be modelled. The basic problem is not soluble by using
> a slightly different multivariate command. Having a mix of predictor
> types, dummies, categorical and continuous variables, is of course a
> soluble problem.
>
> Nick
>
> On Fri, Feb 3, 2012 at 9:32 AM, Urmi Bhattacharya <[email protected]> wrote:
>
>> I am using exploratory factor analysis to generate factor loadings and
>> the corresponding uniqueness values using 16 variables. I have a mix
>> of dummy variables (taking values 1 or 0), categorical variables
>> (positive integers), and continuous variables. I am using the
>> following command:
>>
>> factor govt_school chais_desk_s schl_toilet_s schl_water_s
>> hrs_electric_s  num_classoutdoor_s num_mixedgrade_s reg_fee_gen_s
>> tuit_fee_ gen_s pupil_teach_s inservice_training_s library_s
>> computer_use_s playgrnd_s  formal_teach_eval_s distance_primaryschool,
>> ipf factor(1).
>>
>> My question is whether this is the correct procedure to use when I
>> have variables that are not continuous? If not, is there a command in
>> Stata that better handles this?
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index