Right, but you can use -adjust- to easily 1) compare means across
groups (e.g. foreign) and 2) evaluate the expected values at specific
values of your independent variables.
On Wed, Jan 14, 2009 at 1:46 PM, I M <[email protected]> wrote:
> Hello,
>
> Thank you Alan.
>
> By using the 'adjust' command, I have verified that the mean is taken (i.e., by also using the sum, d command, the mean is the same).
>
> ---begin example---
>
> . adjust
>
> -----------------------------------------------------------------------------------------------------------------
> Dependent variable: price Command: qreg
> Variables left as is: weight, length, foreign
> -----------------------------------------------------------------------------------------------------------------
>
> ----------------------
> All | xb
> ----------+-----------
> | 5473.59
> ----------------------
> Key: xb = Linear Prediction
>
> . predict test, xb
>
> . sum test, d
>
> Linear prediction
> -------------------------------------------------------------
> Percentiles Smallest
> 1% 948.5573 948.5573
> 5% 1959.794 1361.076
> 10% 3384.565 1920.458 Obs 74
> 25% 4723 1959.794 Sum of Wgt. 74
>
> 50% 5435.645 Mean 5473.588
> Largest Std. Dev. 1813.017
> 75% 6385.588 8804.352
> 90% 7638.702 9254.924 Variance 3287029
> 95% 8804.352 9423.244 Skewness -.1285146
> 99% 9771.52 9771.52 Kurtosis 3.297427
>
> ---end example---
>
>
> Note the mean is 5473.59, the same in the adjust and sum commands.
>
> But I also wanted to verify that even though the commmand is qreg (for quantile regression), I should use the mean to report the statistics, and not the 50th percentile?
>
> Thank you!
>
> ----------------------------------------
>> Date: Wed, 14 Jan 2009 13:39:10 -0500
>> Subject: Re: st: How to interpret and summarize adjusted medians
>> From: [email protected]
>> To: [email protected]
>>
>> You might want to look at the help file for -adjust-. Here are some examples:
>>
>> sysuse auto
>> qreg price weight length foreign
>> adjust
>> adjust, by(foreign)
>> adjust, by(rep78)
>> adjust, by(rep78 foreign)
>> adjust weight length, by(foreign)
>>
>> Best,
>> Alan
>>
>>
>> On Wed, Jan 14, 2009 at 1:30 PM, I M wrote:
>>> Hello,
>>>
>>> I have a multivariate model (quantile regression, qreg) with a set of covariates.
>>>
>>> I have run the quantile regression model, then generated predicted values (values adjusted according to the covariates in my model).
>>>
>>> Essentially, these predicted values are the predicted medians.
>>>
>>> How do I summarize these predicted medians? i.e. if I want to report the predicted median by gender, how would I do this?
>>>
>>>
>>>
>>> ---begin example---
>>> /*quantile regression model*/
>>> qreg [outcome] [predictor variable 1] [predictor variable 2, i.e. gender]
>>>
>>> /*generate predicted medians based on estimates*/
>>> predict [predicted medians], xb
>>>
>>> /*summarize the predicted values*/
>>> sum [predicted medians], detail
>>>
>>> ---end example---
>>>
>>>
>>>
>>> The sum, detail command would give me the percentiles and the mean (and some other statistics).
>>>
>>> If I want to summarize the predicted medians, would I state the mean from the "sum" command, or would I state the median from the "sum" command?
>>>
>>> In other words, do I want to report the mean of the 'predicted median', or the median of the 'predicted median'?
>>>
>>> Thank you immensely!
>>>
>>> *
>>> * For searches and help try:
>>> * http://www.stata.com/help.cgi?search
>>> * http://www.stata.com/support/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/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/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/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
