Dear R experts How can you perform a regression analysis in R when the dependent variable is countiuous but bounded, say between 0 and 100? I would be grateful for pointers to R-functions but also for hints to relavant litterature since I have never worked with this problem before. Thanks in advance. Kim Mouridsen. [[alternative HTML version deleted]]
On Thu, 4 Dec 2003, Kim Mouridsen wrote:> Dear R experts > > > > How can you perform a regression analysis in R when the dependent > variable is countiuous but bounded, say between 0 and 100? > > I would be grateful for pointers to R-functions but also for hints to > relavant litterature since I have never worked with this problem before.The usual way is to perform a logistic regression, and nls() has code to fit various logistic regression: glm could also be used with response/100. -- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595
What are your assumptions about the distribution of the dependent
variable and its relationship to independent variables? Have you
considered "glm" with family = quasibinomial?
hope this helps. spencer graves
Kim Mouridsen wrote:
>Dear R experts
>
>
>
>How can you perform a regression analysis in R when the dependent
>variable is countiuous but bounded, say between 0 and 100?
>
>I would be grateful for pointers to R-functions but also for hints to
>relavant litterature since I have never worked with this problem before.
>
>
>
>Thanks in advance.
>
>Kim Mouridsen.
>
>
> [[alternative HTML version deleted]]
>
>______________________________________________
>R-help at stat.math.ethz.ch mailing list
>https://www.stat.math.ethz.ch/mailman/listinfo/r-help
>
>
Kim This would depend a lot on how the DV is distributed within the 0-100 range. It's even conceivable that the usual linear model could be used without much problem (e.g. if the mean is 50 and the sd is small, then the bounds of 0 and 100 will exist more in theory than in practice; of course, you'd still want to do a lot of diagnostics). 0 to 100 makes me think of a percent, there are methods for dealing with percents. More generally, you may want to look at Long, JS (1997). Regression Models for Categorical and Limited Dependent Variables. Sage. which I find useful frequently. HTH Peter Kim Mouridsen wrote:>Dear R experts > > > >How can you perform a regression analysis in R when the dependent >variable is countiuous but bounded, say between 0 and 100? > >I would be grateful for pointers to R-functions but also for hints to >relavant litterature since I have never worked with this problembefore.> > > >Thanks in advance. > >Kim Mouridsen. > > > [[alternative HTML version deleted]] > >______________________________________________ >R-help at stat.math.ethz.ch mailing list >https://www.stat.math.ethz.ch/mailman/listinfo/r-help > >______________________________________________ R-help at stat.math.ethz.ch mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help Peter L. Flom, PhD Assistant Director, Statistics and Data Analysis Core Center for Drug Use and HIV Research National Development and Research Institutes 71 W. 23rd St www.peterflom.com New York, NY 10010 (212) 845-4485 (voice) (917) 438-0894 (fax)
<... the dependent variable is countiuous but bounded, say between 0 and 100. Do you know why the dependent variable is 'bounded'? <... hints to relavant litterature > If the dependent var is a percent John Cornell's book on mixture data might be of interest. Is it truncated in the sense that negative values are impossible? This situation is often dealt with in econometrics literature. This is somewhat out of date but a classic: G. S. Madalla, Limited Dependent and Qualitative Variables in Econometrics, Cambridge University Press, Cambridge, 1983. Is it censored? Maybe search for Tobit analysis depending on what assumptions you are willing to make (also see something like Madalla for this). There is a large literature on censoring. Maybe you just happened to not see anything outside [0,100]? Maybe the boundedness is ignorable:) Bob -----Original Message----- From: Kim Mouridsen [mailto:kimm@pet.auh.dk] Sent: Thursday, December 04, 2003 10:39 AM To: r-help@stat.math.ethz.ch Subject: [R] regression with limited range response Dear R experts How can you perform a regression analysis in R when the dependent variable is countiuous but bounded, say between 0 and 100? I would be grateful for pointers to R-functions but also for hints to relavant litterature since I have never worked with this problem before. Thanks in advance. Kim Mouridsen. [[alternative HTML version deleted]] ______________________________________________ R-help@stat.math.ethz.ch mailing list https://www.stat.math.ethz.ch/mailman/listinfo/r-help [[alternative HTML version deleted]]