[Nick Cox]

Dianne:

The forum software lets you create tables, but in my view they are often difficult to read. Notice how Sarah and Svend and Sergiy are displaying code and results as Code. See Section 12 in the FAQ Advice for how to do this. It takes about 1 minute to learn.

[Svend Juul]

Dianne,
In posting #8 you treat the NHANES data as if SXQ639 was asked to females and SXQ627 to males. That is not the case; the same persons have been asked about the number of contacts with females (SXQ639) and with males (SXQ627).

In posting #10 you tell us that the original data were not grouped; you created the >=2 groups yourself. But if you want to add the number of contacts with males and females, you should do this before grouping the original variables. You may want to present grouped data, but you should not group them before performing intermediate calculations.

In posting #10 you also tell us that the code for missing is 99999 in SXQ639 and 77777 in SXQ627. That is not the case according til the NHANES documentation. 77777 is the code for refused and 99999 the code for don't know in both variables. To change these to Stata's missing values, you could (because nobody reported more than 1000 contacts):

recode SXQ639 SXQ627 (1000/max=.)

Best,
Svend

[Dianne Devine]

Thank you for your hep everyone