Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Report of weighted cell counts or observations in a survey report


From   Steven Samuels <sjhsamuels@earthlink.net>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Report of weighted cell counts or observations in a survey report
Date   Wed, 13 Aug 2008 09:29:40 -0400

Ángel. I agree that displaying the raw counts is important. I don't know a short cut to doing this.

Some suggestions:

1. Create separate tables of cell counts (no percentages)-these take little space. You might put them in an appendix.

2. Use a table editor in your word processor to create a table skeleton with a small number of cells. Put one Stata table into each cell, shrinking the text as needed. This would allow you, for example, to put the table of raw counts adjacent to the weighted tables.

3. My personal approach: I use -collapse-, -egen-, returned results, merges, -sorts-, and -format- commands to create data sets of statistics, then -outsheet- to text files. I manipulate these in a text editor using GREP and drop the results into rtf table skeletons. The final polishes I do in Nisus Writer's table editor, but you can do the same in Word. In this way I've created tables with over 20 columns, with multiple sections and subheadings. I use Andale Mono with font size of 7 or 8 points; this is very readable. The process takes a lot of time to set up, but after the initial work, it is easy to make modifications. For example, changing the order of super- and sub-columns is easy: it's a matter or changing the -sort- and - outsheet- order of variables.

-Steve


On Aug 13, 2008, at 5:08 AM, Ángel Rodríguez Laso wrote:


Dear Statalisters,

I'm producing a general report (lots of tables and graphs) on the
results of a survey where weigts were used. Appart from percentages
and confidence intervals, I would like to add sample sizes, but I'm in
doubt about reporting weighted cell counts or number of observations
(unweighted). In principle, I'm for reporting weighted cell counts
(many readers would dislike than frequencies and sample sizes do not
match), but number of observations may be very relevant when there are
very few 'measured' individuals in some cells. Any thoughts about
that?

In any case, is it possible to make Stata suppres frequencies in
svy:table when ,obs are asked? (I've found that the option nofreq for
tabulate twoway is not allowed in svy:tab).

Many thanks,

Ángel

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index