calculating elasticity using dataset that do not contain weight

[Andy]

Dear Professor Bierlaire,

Hope this email finds you well.

I have a dataset that does not contain weight.

Is it possible to estimate elasticity using such data?

Should I assume all observations to have an equal weight (=1)?

Thank you for your help.

[Michel Bierlaire]

There are two things to keep in mind:

1) Whenever you use sample enumeration to calculate *aggregate* indicators (market shares, elasticities, willingness to pay, etc.), your sample *must* have the same statistical properties as the population. This is why weights are needed. And I have never been exposed to a sample that did not need to be re-weighted. And, by the way, this is also why you should *never* use stated preference data, generated using any experimental design, to perform sample enumeration. I don’t know if your data is SP or RP. But SP data are useless in this context.

2) At the disaggregate level, calculating the elasticities does not require any weight.

Therefore, if you are interested in disaggregate elasticities, don’t worry about the weights. If you are interested in aggregate elasticities, make sure you calculate the weights first. Do *not* assume they are equal to 1.

See https://transp-or.epfl.ch/documents/technicalReports/Bier18a.pdf

> --
> You received this message because you are subscribed to the Google Groups "Biogeme" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to biogeme+u...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/biogeme/3ca0ca8a-9c37-4199-b2a9-1f338f1a694bn%40googlegroups.com.

Michel Bierlaire
Transport and Mobility Laboratory
School of Architecture, Civil and Environmental Engineering
EPFL - Ecole Polytechnique Fédérale de Lausanne
http://transp-or.epfl.ch
http://people.epfl.ch/michel.bierlaire