Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Pooling Data
Nick Cox <email@example.com>
Re: st: Pooling Data
Fri, 15 Apr 2011 10:24:18 +0100
If your variables are totally comparable, you have no problem and it
is difficult to know why you are asking the question. But we don't
know more about the situation than you do.
If your variables are not comparable, you do any calculations you like
but it is difficult to know what the results will mean.
Logically, it is on a level with you evaluating someone you know and
saying "he got 42%" and me evaluating someone I know and saying "she
got 56%". We have numbers but unless we have some way of comparing the
measurement scales what is there to say?
On Fri, Apr 15, 2011 at 2:06 AM, Dennis Kramer <firstname.lastname@example.org> wrote:
> given that the X and Y are both achievement on the same subject-- math
> -- (just calculated differently) could I not apply a pooled
> regression coefficient function after estimating the OLS regression
> models for each district?
> On Thu, Apr 14, 2011 at 7:52 PM, Nick Cox <email@example.com> wrote:
>> I agree with Joerg. The question sems to be: I have measures for
>> variable y in area A and x in area B, and how can I compare them? I
>> can't see that you do anything without a cross-calibration between y
>> and x.
>> On Fri, Apr 15, 2011 at 12:48 AM, Joerg Luedicke
>> <firstname.lastname@example.org> wrote:
>>> On Thu, Apr 14, 2011 at 7:23 PM, Clive Nicholas
>>> <email@example.com> wrote:
>>>> Dennis Kramer replied:
>>>>> I am attempting to compare the effectiveness of teachers in two
>>>>> districts (based on a common evaluation tool) where the student test
>>>>> scores outcomes are calculated differently. the first district
>>>>> calculates the scores on a cumulative gain index standardized across a
>>>>> number of factors and the other district uses a value-added index.
>>>> I think I'll do my best impersonation of a David Greenberg posting and
>>>> say that, to me, your problem appears to cry out for a multilevel
>>>> modelling solution, perhaps via -gllamm-.
>>> I do not know how a multilevel model would be of help here, besides
>>> there only being 2 districts, which is not enough anyway. But first
>>> and foremost, this seems to be a measurement problem. I do not know
>>> anything about these two indices but if they are picking up different
>>> stuff and one index cannot be translated into the other, I guess then
>>> the data from the two districts cannot be combined.
* For searches and help try: