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

Re: st: equivalence tests

From   Richard Goldstein <>
Subject   Re: st: equivalence tests
Date   Tue, 10 Jul 2007 09:38:25 -0400

I am again, probably temporarily, on Statalist so ...

As I have said before, there are at least 2 bugs in equim/equip:
(1) the lower limit of the t-based confidence interval is wrong

(2) the results are nonsense when the actual delta is smaller
than the selected delta

Further, as I have said before, I am not interested in these
techniques and will not be fixing the programs.

I was recently contacted, privately, by someone who expressed
interest in fixing up the programs -- this person may or may
not end up doing so and, of course, I know nothing about the
timing for this work.

At present, I recommend that those who want these techniques
use some other software (e.g., equivtest from statistical

Rich Goldstein

Cathy L. Antonakos wrote:
Celine Genty wrote:

How can I do equivalence tests for means and for proportions ?

There is a message on "Equivalence Testing" in the Stata archive, Jan 14 2004, that mentions problems with -equim- or -equip-. I could not resolve the results in the STB Reprint Vol. 3 pp. 107-112, STB-17, sg21 (R. Goldstein) when I applied formulas from a paper I was given by a client (Hwang, IK & Morikawa T. 1999. Design issues in noninferioirty/equivalence trials. Drug Information Journal, vol. 33, pp. 1205-1218).

I wound up using formulas in Excel based on the Hwang & Morikawa paper (and still had trouble getting reasonable results, possibly based on a narrow equivalence interval, or lack of clarification in the paper).

There is also the Bland & Altman method. The reference is Bland JM, Altman DG. 1986. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet i, 307-310.

Cathy Antonakos
*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index