3
$\begingroup$

I have a case-control study in which the cases are firms with health insurance and the controls are firms with no health insurance. I am studying the factors affecting enrolment in health insurance and was therefore using a logistic regression, which includes several covariates on firm characteristics that were measured in a survey. I have randomly sampled the firms from a database that includes two strata: insured and uninsured firms. I selected 65 from each group. However, within the group I also sampled from four strata that correspond to industry. I am therefore wondering if I need to use conditional logistic regression, as opposed to unconditional logistic regression. However, I was under the impression that conditional logistic regression was for matched case-control studies or panel studies. In other feedback I've received I've been told that because I sampled on the outcome, I need to use the conditional model. Could someone please help me figure out which m? Any references would also be much appreciated. Thank you.

$\endgroup$
4
  • $\begingroup$ Could you clarify how you sample from the four industry strata? Did your sampling scheme specify that you picked a certain number from each stratum? Also do you have data at the employee level or only at the company level? $\endgroup$
    – onestop
    Feb 14, 2011 at 7:57
  • $\begingroup$ I first divided each stratum (insured and uninsured) by industry (4 groups). Within these industry groups I further divided the sample into 4 groups that correspond to the size of the company. This was to ensure that I collected similar numbers within each industry and that i didn't up with only small firms. I then numbered all observations and used quota sampling with a random number generator to select firms from each of the 32 groups. I used sampling weights in the analysis to ensure that each observation was weighted by the inverse of the probability of being selected. $\endgroup$
    – Sarah
    Feb 14, 2011 at 15:58
  • $\begingroup$ No, I don't look at any employee level data within the sample. I realize that my sampling is probably not the best but I was mainly interested in getting some descriptive results. However, I'd also like to be able to use the logistic regression analysis to understand the factors affecting whether or not a company enrolls in health insurance. Thank you! $\endgroup$
    – Sarah
    Feb 14, 2011 at 15:58
  • $\begingroup$ Clarification: I wrote that I used quota sampling but it was in fact stratified sampling. $\endgroup$
    – Sarah
    Feb 14, 2011 at 19:32

1 Answer 1

4
$\begingroup$

I don't agree that you sampled on the outcome, since you sampled on company and enrollment is your outcome. You may want to deal with the company as a random effect and the other features as fixed effects. So I am suggesting yet a third alternative: generalized mixed models.

After clarification: If the outcome is company enrollment rather than employee enrollment, then it is an ordinary case-control study for which unconditional logistic regression should be the standard approach. Conditional logistic regression is not necessary unless there were further conditions on the sampling regarding other company features.

Further clarification: If you were using R, then the package to identify and install would be not surprisingly: "sampling" by Thomas Lumley. It provides for the appropriate incorporation of the two-way sampling strategy you have outlined in the design phase prior to estimation with the svyglm() function. Stata also has a set of survey functions and I imagine they can also be used with the general linear modeling functions it provides. SAS didn't have such facilities in the past so the SUDAAN program was needed as an added (expensive) purchase, but I have a vague memory that this may have changed with its latest releases. (I don't know about SPSS with regard to sampling support for GLM models.)

$\endgroup$
4
  • $\begingroup$ Thank you. I did sample on the outcome. I edited my question above to be more clear. The database included two different strata: enrolled companies and unenrolled companies. I chose 65 from each stratum. In this case, what would you suggest I use for the model? Thank you! $\endgroup$
    – Sarah
    Feb 14, 2011 at 1:27
  • $\begingroup$ I am still hoping someone can answer my question, based on further details in response to comments from "onestop", above. Thanks! $\endgroup$
    – Sarah
    Feb 16, 2011 at 18:41
  • $\begingroup$ I'm no expert on analysis of survey data so I'm not certain, but I tend to agree with DWin that conditional logistic regression is not necessary here. Ordinarly logistic regression with survey weights should be ok. I do have a question though: If you already have all the data available in a database, why sample at all? Why not use all the firms in the database? Or are you then going on to collect additional data on the sampled firms? $\endgroup$
    – onestop
    Feb 16, 2011 at 22:18
  • $\begingroup$ Yes, I only have a list of the firms and after selecting them I'm going out to collect data on them. Thank you for your comments. $\endgroup$
    – Sarah
    Mar 9, 2011 at 15:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.