How to utilise sample weights in Stata? - QUICK QUESTIONS

[Jonathon River]

Hi Everybody,


I am running a regression analysis using HLFS on Stata. I have been told that I need to use sample weights to have reliable estimates. The HLFS data has a "factor weight" column for each observation. My questions are as follows:


1) Does it mean that HLFS data already provide me with the sample weights? So, I don't need to calculate it?

2) If the answer is yes to (1), how do I use this on Stata?

I am writing a command as below, but I am not quite sure if I am weighting twice. [pweight=weights] --> The bold represents the factor weight column on HLFS data.

oaxaca LnWage var1 var2 var3 var4 var5 [pweight=weights], by(Gender) pooled

3) If answer to (1) is no, then how can I calculate the sample weights, as I don't know the total population of a sector?



Any suggestion would be much appreciated!

Thank you very much for your time and effort.

[William Lisowski]

One suggestion is that Statalist members come from a broad range of disciplines, so you should avoid jargon from your particular field, and you should not assume that their background is similar to yours. There are lots of members experienced with Stata's weights but unfamiliar with "HFLS" and reluctant to offer advice without more details about the weights provided.

I was interested in addressing your question, and when I search the web for HFLS, the first page of results includes

Hebrew Free Loan Society (HFLS) | Interest-free Loans for ...
HFLS | Primary Human Fibroblast Like Synoviocytes | Cell ...
Human Fibroblast-Like Synoviocytes: HFLS, adult | Sigma ...
Household Food and Livelihood Security Project (HFLS ...
Hangzhou Foreign Languages School - Wikipedia
HFLS-RA | Primary Human Fibroblast Like Synoviocytes ...
Synoviocytes - Primary Cells | Sigma-Aldrich
HFLS (Hampshire Family Legal Solutions)
Meet HFLS Borrowers - Jewish Communal Fund

None of these seem likely to be a source of data.

Perhaps you could provide a link to the source of your data, and its documentation.

[Jonathon River]

Apologies, that is a very fair point Mr. William.

HLFS (Household Labour Force Surveys)

Here I attached the structure of the micro HLFS data.

Just to add, the data set I have created with dummy variables. I run the regression, but without sample weights.

Many thanks again!

[William Lisowski]

There are Household Labor Force Surveys collected in many countries. A look at the PDF you attached suggests that your data was collected in Turkey. After much searching on the web, at

https://data.tuik.gov.tr/Kategori/Ge...-and-Wages-108

we can learn that

In the study, while reaching the final weight, the design weights have been calculated depending on the selection criteria; have been controlled for external distribution and corrected for non-responses. In weightning, age group, gender, NUTS Level 2, urban-rural and household size are based on as external control.

so for your questions

1) Yes, like many surveys, yours has provided you with appropriate weights for your use.
2) I do not understand why you think you might be weighting twice. The output of the command help weight tells you that

[pweight=Faktor]

is how you specify that the variable Faktor is to be used as a sampling weight, which is appropriate given the description quoted above.

[Jonathon River]

Mr. William, I would like to thank you so much for your time and effort. It's very thoughtful of you to search the Web not once, but quite long time for my questions.

The reason I was thinking "twice" scenario, I thought the command helps you calculate the weights, as I already have the weights I wasn't convinced with the command I have written.

You now made it very clear!

Thank you so much again!