Re: st: Peer variable coefficient estimate nonsense

[William Buchanan <william@williambuchanan.net>]

Jian,

There is some theoretical support for your idea to some degree.  I would recommend looking at the work of Vygotsky who is largely responsible for advancing theory regarding the social context and mechanisms at work in learning and development.  

With regards to the statement from Clyde, your assumption would be correct if the students all took a test scored on a single scale.  However, the majority of educational assessments are not vertically aligned and are not intended to support cross-grade inferences.  For example, a scaled score of 40 for a fifth grade student and a scale score of 50 for a fourth grade student may mean that both student scores are exactly at the mean for their respective grades.

In addition to the other feedback you've received I would also consider controlling for semester with regards to the assessment tool itself. If you are using a benchmark assessment tool this would make sense (since the assessment is designed to measure growth over time), but if you are using a summative assessment then the data is likely to be measuring skills that students have not yet learned.  There are a few papers on peer effects available through the NBER website, but the evidence seems to be mixed at best.

Regards,
Billy 

Sent from my iPhone

On Mar 2, 2012, at 20:26, Jian Zhang <jian32@gmail.com> wrote:

> Hi Clyde Schechter,
> 
> thanks for your comments. as you guessed, the peer group is the
> student's class and the score is his/her math score.  In my regression
> analysis I include grade and time semester dummies, i.e., I have
> controlled for grade and semesters.  Thus this means what i am looking
> at is students who are in the same semester and grade: I am comparing
> students in the same grade and same semester.  Then the question is
> really: given a student, 1 year older of his peers matters more than 1
> year older of his own age to his math achievement? For such a student,
> it seems that his own age matters more than his peer average ages to
> math achievement. Of course, there is no theory/empirical evidence
> supporting this.  it is just out of my bold intuition...
> 
> Jian
> 
> On Sat, Mar 3, 2012 at 1:52 AM, Clyde B Schechter
> <clyde.schechter@einstein.yu.edu> wrote:
>> Leaving aside the technical issues brought up by Nick Cox of how correlated variables might partition variance among themselves in a regression analysis, substantively, I wonder why Jian Zhang thinks it is nonsensical for peer age to be a stronger predictor of student score than the student's own age.  It isn't explained what kind of scores and peer groups these are, but if the "peer" group in question is the student's class, it wouldn't surprise me at all that mean peer group age, which is then a very strong proxy for grade level, would be a stronger predictor of, say, math achievement, than the child's individual age when the two are used together.  In fact, it _would_ surprise me if the opposite were true.  After all, we would expect 5th graders to have higher math scores than 4th graders, but there is no reason to think that an older 4th grader would outperform a younger 5th grader.
>> 
>> More generally, there are many other instances where a group attribute is a stronger predictor of an individual outcome than the analogous attribute of the individual.
>> 
>> Clyde Schechter
>> Dept. of Family & Social Medicine
>> Albert Einstein College of Medicine
>> Bronx, NY, USA
>> 
>> 
>> 
>> *
>> *   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/