Re: st: Problem tracking students

[Eric Booth <ebooth@ppri.tamu.edu>]

<>

It doesn't look like there is enough information here for us to know where your extra 255 students came from.  

A couple of thoughts:

> All of these 1,381 cases are verified in the data to have missing
> values in year 3. Can someone please explain why my year 3 flag is
> identifying more than 1,126 students?

1.  If you've verified that all 1,381 have missing values in year 3, then why would you expect a different count (1,126 students) from your code? Where did this expected 1,126 count come from?  Maybe this 1,126 count comes from some other report or paper; if so, you'll need to figure out what students that figure is excluding that you aren't..we can't guess why you are getting a different count than is reported elsewhere)

2. You might try reshaping this data to long, it tends to make this type of data easy to work with.
3. You can list your gr* variables for some of these 255 cases and see if you see anything suspicious.
4. Check for duplicates (e.g., duplicates report id)
5. Check to see how your code accounts for students in your cohort that dropout in a one year and then come back later.  
So, could students drop out in your first year (02-03) and then come back in later years or could students drop out in 04-04 and reenter in 04-05(?) 
6. It probably wouldn't affect this issue, but be sure to check how your code accounts for the students who repeat or skip grades across these school years.
7. Double check that all the students in your data are students in 6th grade in 02-03 (do you need to account for whether it is their first attempt at 6th grade in 02-03 when selecting this group of students?)  and check to be sure that the cohort1 variable isn't selecting (or filtering) unwanted students.

If none of these help, we'd probably need more information and a good example of your data to offer more advice.

- Eric

__
Eric A. Booth
Public Policy Research Institute
Texas A&M University
ebooth@ppri.tamu.edu
Office: +979.845.6754



On Mar 16, 2011, at 4:41 PM, S. D'Souza wrote:

> Hi everyone,
> 
> I am working with a student level dataset that contains longitudinal
> data in wide format (case by variables).  I am interested in 6th grade
> students in 2002-03 only and the dataset has been limited to these
> students. The total sample is 21,731 students.
> 
> The problem: I am interested in tracking students who drop out of the
> sample over the next two years, but am having difficulty doing so. As
> you can see in the table below, in 2003-04 (year 2), the sample drops
> by 1,922 for a total of 19,809 students:
> 
> 
> Grade level |
> 
>     2003-04 |
> 
>   180th day |      Freq.     Percent        Cum.
> 
> ------------+-----------------------------------
> 
>           6 |        886        4.08        4.08
> 
>           7 |     18,833       86.66       90.74
> 
>           8 |         80        0.37       91.11
> 
>           9 |         10        0.05       91.16
> 
>           . |      1,922        8.84      100.00
> 
> ------------+-----------------------------------
> 
>       Total |     21,731      100.00
> 
> 
> 
> In the following year, 2004-05 (year 3), the sample drops by 1,126 for
> a total of 18,683 students:
> 
> 
> Grade level |
> 
>     2004-05 |
> 
>   180th day |      Freq.     Percent        Cum.
> 
> ------------+-----------------------------------
> 
>           6 |         14        0.06        0.06
> 
>           7 |      1,336        6.15        6.21
> 
>           8 |     17,207       79.18       85.39
> 
>           9 |        115        0.53       85.92
> 
>          10 |         10        0.05       85.97
> 
>          11 |          1        0.00       85.97
> 
>           . |      3,048       14.03      100.00
> 
> ------------+-----------------------------------
> 
>       Total |     21,731      100.00
> 
> 
> I would like to create flags for students who drop out in years 2 and
> 3 respectively. I am able to create a flag for the 1,922 students who
> drop out in year 2 without any problem. However, the flag for year 3
> shows that 1,381 students are being flagged instead of 1,126 students.
> All of these 1,381 cases are verified in the data to have missing
> values in year 3. Can someone please explain why my year 3 flag is
> identifying more than 1,126 students?
> 
> Here is my code where “gr05_180” provides grade level in 2004-05 and
> “gr04_180” provides grade level in 2003-04:
> 
>  gen gry3m=.
> 
> replace gry3m=1 if gr05_180==. & gr04_180!=. & cohort1==1
> 
> 
> . tab gry3m
> 
> 
> 
>       gry3m |      Freq.     Percent        Cum.
> 
> ------------+-----------------------------------
> 
>           1 |      1,381      100.00      100.00
> 
> ------------+-----------------------------------
> 
>       Total |      1,381      100.00
> 
> 
>  Thanks,
> 
> Stephanie
> 
> *
> *   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/