Weighting, Generate weights, Mill's ratio

[Chiara Tasselli]

Good morning everyone. I have a sample of workers (25 companies) and I would like to make it representative of the national labor force. I have standardized it with the national database (harmonizing some variables and appending them). Now, how should I proceed? My idea was to generate the dichotomous variable mysample, construct the Mill's ratio with a probit on that, and then use it as a covariate in the analyses only on my sample. Would this be a correct way to weight? Because usually it is apply wek the depvar in the second stage is censored....

Alternatively, can someone point me to literature or some coding for constructing weights?

many thanks in advance for your help and time

[George Ford]

I think you want to a raking procedure.

help ipfweight

[Nick Cox]

Trivial detail: The Mills ratio is so called because it was tabulated by John P. Mills. Surnames like Mills cause difficulty even to people whose first language is English. But Mills ratio, Mills' ratio and Mills's ratio are all defensible renderings, but on the other hand Mill's ratio is wrong.

https://stats.meta.stackexchange.com...onts/2814#2814 includes a selection of common typos and other such details in statistics. Naturally such a list can never be complete and so further suggestions are welcome.

[Andrew Musau]

Related to #3, but slightly off-topic, sometimes how you render the possessive matters. For football fans out there, Newcastle F.C.'s home ground is St. James' Park, not St. James's Park.

[Nick Cox]

I've tried a few times to find out a little more about John P. Mills. All I've found is the 1927 paper talking about the ratio. So, it's safe to say that he's not around to give an opinion.

To pick up Andrew Musau's point. St James Park, St James' Park and St James's Park are all in use somewhere and whoever lives or works at or near any concrete instance has some claim to say which version is customary, preferred or whatever. But St Jame's Park would be wrong unless somewhere or somewhen there was a Saint Jame.

[Chiara Tasselli]

I think you want to a raking procedure.

help ipfweight

George Ford Good morning, thank you very much for your advice; it was extremely valuable! The command worked perfectly. May I now ask how these weights should be applied to my regressions? Looking at my data and the results I found in the weights created and by doing the tabs, it seems that aweight is the most appropriate and also the only one I can apply. Is that correct?

Once again: thank you so much for your help

Nick Cox thanks for the spelling advice

[George Ford]

aw