Re: st: interval data as independent variable

[Steve Samuels <sjsamuels@gmail.com>]

Tobias:


You have only one good choice: convert the interval variable to a continuous one by scoring  x  = interval midpoint value. Then you can fit regressions models, e.g. by using -fp- to control for "x". There should be little loss in precision compared to having the original, ungrouped, measurements. 

 -intreg- has a strong assumption of a normal distribution for the random part of the regression.  The contributed command -intcens- by Jamie Griffin (from SSC) will fit a variety of  distributions to positive interval censored data. To use the command, subtract off the minimum value from your dependent variable from each observation, so that the dependent variable starts at zero.

I am curious about the measurement process that produced such unusual intervals. 

Steve
sjsamuels@gmail.com


On Jan 26, 2014, at 9:31 PM, Tobias Wille <redacted> wrote:

Hello,

I conducted an experiment in my seminar experimental economics and I have a problem with intreg. The data is as follows:

Independent variable: interval data - 15 intervals - no gaps, not of equal size - but should be no problem for intreg. Intervals are as follows: 

1-1,02
1,02-1,04
1,04-1,06
1,06-1,08
1,08-1,1
1,1-1,14
1,14-1,18
1,18-1,22
1,22-1,26
1,26-1,32
1,32-1,38
1,38-1,44
1,44-1,52
1,52-1,6
1,6-1,7

Independent variables:
1 dummy variable
now my problem:
1 dependent variable which is of the same nature as the independent one(interval data, same kind of intervals)

How can I do this regression with an independent interval variable?

Thank you in advance!!

Tobi
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/