Pairwise Comparison of Means

[Co Ar]

Dear all,
I did a pairwise comparison of means using the following command:
pwmean Urban_Peri , over(Foodgrp12 Wealthindex2 ) pveffects cimeans cformat(%9.3f) pformat(%5.3f) sformat(%8.3f)

I got the following results:

Urban and Peri_Urban Dwellers
Food Groups#Wealth Index	Contrast	Std. Err	t	P>t	[95% Conf. Interval
(1 1) vs (1 0)	-0.33153	0.097091	-3.41	0.001	-0.52253	-0.14054
(4 1) vs (1 0)	-0.17915	0.103159	-1.74	0.083	-0.38208	0.023782
(5 1) vs (1 0)	-0.4531	0.128189	-3.53	0	-0.70527	-0.20093
(6 1) vs (1 0)	-0.31941	0.150245	-2.13	0.034	-0.61497	-0.02385
(9 1) vs (1 0)	-0.42042	0.162605	-2.59	0.01	-0.74029	-0.10055
(12 1) vs) ( (10 1)	-0.26486	0.133916	-1.98	0.049	-0.5283	-0.00143
(2 0) vs (1 1)	0.466667	0.142141	3.28	0.001	0.187051	0.746283
(4 0) vs (1 1)	0.196396	0.097091	2.02	0.044	0.005402	0.387391
(6 0) vs (1 1)	0.266667	0.152951	1.74	0.082	-0.03422	0.567549
(8 0) vs (1 1)	0.466667	0.206239	2.26	0.024	0.060957	0.872376
(9 0) vs (1 1)	0.466667	0.159752	2.92	0.004	0.152406	0.780928

My problem is that I can't interpret what, for example,(1 1) vs (1 0) mean. Can somebody help please?

[Rich Goldstein]

those are the cells from your two "over" variables; you would find the output easier to read, I think, if you added value labels to the values of your "over" variables; the following might help:

sysuse auto
pwmeans price, over(foreign rep78) pveffects

[Co Ar]

The value labels do exist, but stata cant use them on this table. I have tried and can even send part of the data. My problem was interpreting those significant results.
The (1 1) or (9 0) are what I don't know how to read. Can I say that those eating food group 9, and live in urban areas are poor (wealth index was coded 0-poor 1-high income)

[William Lisowski]

If your problem is how to interpret your results, then beyond the information available from help pwmean you may find the Remarks and Examples section of the pwmean documentation in the Stata Base Reference Manual PDF included in the Stata installation (since version 11) and accessible from within Stata - for example, through Stata's Help menu.

If your problem is understanding what comparison is indicated by notation like "(1 1) vs (1 0)", then that example is comparing the mean of Urban_Peri for observations with Foodgrp12=1 and Wealthindex2=1 with the mean for observations with Foodgrp12=1 and Wealthindex2=0. You don't tell us how your variables are coded, so I can't comment on the interpretation. In particular, Urban_Peri sounds like an indicator variable, and if so, someone else will have to comment on the applicability of these comparisons to indicator variables.

As Rich suggests, use of value labels for Foodgrp12 and Wealthindex2 would make "(1 1) vs (1 0)" easier to read. Your response that "The value labels do exist, but stata cant use them on this table. I have tried and can even send part of the data." is uninformative. My experimentation shows that pwmean has reasonably robust handling value labels. In general, as well as for this specific issue, Statalist readers can better help you if we know what commands you have tried and what Stata told you to indicate that there was a problem. Please review the Statalist FAQ linked to from the top of the page, as well as from the Advice on Posting link on the page you used to create your post. See especially sections 9-12 on how to best pose your question. It's particularly helpful to copy commands and output from your Stata Results window and paste them into your Statalist post using CODE delimiters, as described in section 12 of the FAQ. And, as FAQ 12.1 advises,

Say exactly what you typed and exactly what Stata typed (or did) in response. N.B. exactly!
...
Never say just that something "doesn't work" or "didn't work", but explain precisely in what sense you didn't get what you wanted.

[Co Ar]

Ok. Let me see how to go about it. Thanks for your good advice