chi square tests in SPSS and Stata

[Lana Chikhungu]

Hello users

I am undertaking some chi square tests in SPSS and Stata. In one of the analysis the results are statisticaly signficant in SPSS but not in Stata. I have applied weights to both the SPSS and Stata analysis. What could be the reason for the difference. Your help will be greatly appreciated.

Kind regards

Lana

[Nick Cox]

Please show full results from both. Otherwise the only guesses are very weak speculations such as

1. At least one program is wrong.

2. You didn't really do the same thing.

3. Defaults for complications may differ.

If this is any kind of survey analysis you may need to give much detail.

[William Lisowski]

Welcome to Statalist, Lana.

I think Nick's third option is the most likely - I expect the interpretation of weights is different between your SPSS analysis and your Stata analysis. The following link may help you review your SPSS and Stata code to understand why the results differ.

https://stats.idre.ucla.edu/other/mu...-spss-support/

[Lana Chikhungu]

Thank you Nick and William. I have attached the outputs from SPSS and STATA. Having read the link that you shared with me William I think the weights applied in SPSS may be frequency weights while those in STATA are probability weights. I am using the same weight variable. In SPSS I simply choose weight cases by Sampleweight while in STATA I use the following command svyset [pweight=Sampleweight]. I need to learn how to do probability weights in SPSS, otherwise I can just stick with STATA results.

Thanks again for your for replies.

Lana