Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: OLS assumptions not met: transformation, gls, or glm as solutions?

From   Maarten Buis <>
Subject   Re: st: OLS assumptions not met: transformation, gls, or glm as solutions?
Date   Mon, 17 Dec 2012 12:32:27 +0100

On Mon, Dec 17, 2012 at 11:43 AM, Laura R. wrote:
> I estimated an OLS model with the number of minutes (1-1440) spent on
> an activity on a day as dependent variable. At first sight, the model
> works fine. I receive some interesting results which are robust across
> model specifications. I would like to keep it as it is, but:
> - The regression diagnostics shows that the error terms are not
> normally distributed, but right skewed.
> - In addition, there is heteroskedasticity.

I would not worry too much about heteroskedasticity or the
distribution of the errors. I would worry about the linearity

One possibility would be to dividing your dependent variable by 1440,
such that you have proportion of the day spent on an activity. Than
you can use a fractional logit, that is, use -glm- with the
-family(binomial)-, -link(logit)- and -vce(robust)- options (*).

Hope this helps,

(*) There is a program called -fraclogit- available on SSC. However,
I'd recommend using -glm- as the programming of -fraclogit- is
somewhat problematic. A list of problems has been sent to the author.

Maarten L. Buis
Reichpietschufer 50
10785 Berlin
*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index