Bookmark and Share

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, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: SVY medians and Elixhauser

From   Steve Samuels <>
Subject   Re: st: SVY medians and Elixhauser
Date   Sun, 14 Apr 2013 12:00:55 -0400

A rule of thumb quoted on page 143 of a UN document (below) is to round
the standard error to two significant digits and then report that many
digits in the main estimate. The Oxford Journals have this guideline:

"Numbers in the articles and tables should be reported with no more
precision than they merit. Careful thought, not computer packages or the
need to align tables, should govern how many significant digits are
reported. Remember that significant digits are not the same thing as the
total number of digits reported. Do not report more significant digits
than the standard errors suggest."

Thus you are justified in reporting that the median is $37,200 with
standard error $1,700. You would round the CIs in the same way.


Designing Household Survey Samples: Practical Guidelines
December, 2008
Series: Studies in Methods (Ser. F), No.98
Publisher: United Nations, Department of Economic and Social Affairs

On Apr 12, 2013, at 3:25 AM, Mike Butterfield wrote:

Thanks Stas, I think I got it to work!  So it looks like the median
cost of a person with an infection is $37192, which I get from the
second table.  Could you clue me in to what's being described in the
first table though?

. epctile totchg, p(50) over(infection) svy speclab
(running mean on estimation sample)

Percentile estimation
            |             Linearized
     totchg |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
      p50_0 |      23533        626    37.59   0.000     22306.06    24759.94
      p50_1 |      37192       1691    21.99   0.000      33877.7     40506.3

*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index