Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: Not sure how to interpret -somersd- results

From   "Roger B. Newson" <>
Subject   Re: st: Not sure how to interpret -somersd- results
Date   Thu, 03 Oct 2013 16:58:38 +0100

Yes. We can say that, for 3-5 years children, the 2006 calories from non-staples may or may not stochastically dominante 1996 calories from non-staples. And the Somers' D for this stochastic dominance is.206, with a 95% CI from -.094 to .472.

I presume that the observations correspond to individual children, sampled independently from each district in each year. I hope I am right.

Best wishes


Roger B Newson BSc MSc DPhil
Lecturer in Medical Statistics
Respiratory Epidemiology and Public Health Group
National Heart and Lung Institute
Imperial College London
Royal Brompton Campus
Room 33, Emmanuel Kaye Building
1B Manresa Road
London SW3 6LR
Tel: +44 (0)20 7352 8121 ext 3381
Fax: +44 (0)20 7351 8322
Web page:
Departmental Web page:

Opinions expressed are those of the author, not of the institution.

On 03/10/2013 16:39, Jain, Monica (HarvestPlus) wrote:
Thanks so much Roger for the clarification on interpretation and also for the suggestion on using asymmetric confidence intervals. I am also looking at the distribution of calories from non-staples  (calnstp) for children 6-23 months and 3-5 years separately,. I run the following command for height z-scores and calories from non-staple for 3-5 years children. For calories from non-staples for 6-23 months, I also run a similar command, which is not shown for brevity sake.

        somersd year haz calnstp , cluster(district) transf(z) tdist

and I get the following results

Somers' D with variable: year
Transformation: Fisher's z
Valid observations: 627
Number of clusters: 4
Degrees of freedom: 3

Symmetric 95% CI for transformed Somers' D
                                    (Std. Err. adjusted for 4 clusters in dist)
                                             |              Jackknife
         year                            |      Coef.       Std. Err.            t       P>|t|     [95% Conf. Interval]
        Haz (3-5 years)        |   .3743973   .049804         7.52      0.005     .2158966     .532898
calnstp	(3-5 years)        |   .3376242  .0966604	3.49      0.040     .0300076    .6452408
calnstp (6-23 months)  |   .2093922    .095348        2.20      0.116    -.0940477     .512832

Asymmetric 95% CI for untransformed Somers' D
                                                 Somers_D           Minimum        Maximum
        Haz (3-5 years)            .35783201          .21260359        .48759325
calnstp	(3-5 years)	.32535471	.02999857	.56845738
calnstp (6-23 months)  	.20638464	-.09377136	.47214883

Since the asymmetric confidence interval for height z-score does not contain zero, as you mentioned I can say that 2006 child height z-scores stochastically dominate 1996 child height z-scores at 5% level of significance.  For 3-5 years children, the 2006 calories from non-staples stochastically dominants 1996 calories from non-staples at 5% level of significance. However, for 6-23 months children this is not true as the confidence interval contains zero. Am I right in my interpretation of the above results?

Can I say anything about statewise dominance from the above results?

Monica Jain
Postdoctoral Fellow
HarvestPlus, IFPRI
2033 K Street, NW
Washington, DC 20006 (USA)
+1 (202) 862-4690

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index