Interpretation of xttab

[Rose Simmons]

Hi,

I am using a panel dataset (N=237 and T=13).
vote is my dependent variable: 1 if the respondent voted in an annual leadership election, and 0 otherwise (so I am using nonlinear methods).

I wanted to interpret this table:

. xttab vote

                  Overall             Between            Within
   vote   |    Freq.  Percent      Freq.  Percent        Percent
----------+-----------------------------------------------------
       No |    1803     59.06       221     93.25          63.45
      Yes |    1250     40.94       193     81.43          50.14
----------+-----------------------------------------------------
    Total |    3053    100.00       414    174.68          57.25
                               (n = 237)

I read the example in the Stata manual (http://www.stata.com/manuals13/xtxttab.pdf) but I do not understand what is meant by "ever" - please could you explain this to me?

I understand the Overall column: 40.94% on average voted.

However, I am unclear of the meaning of the Between column: 81.43% were "ever" voters.

I am also unclear of the meaning of the Within column: 50.14% of those "ever" voters were always voters.

Thank you

[Marcos Almeida]

Your question relates to the core knowledge of panel data.

In panel data, there are within effects (say, blood glucose level changes "within" the same ID, people get divorced and married anew, etc) and between effects (difference in sex "between" IDs, for example, assuming no transexual surgery was done "within" - sorry for the pun - in the follow-up time).

This is a very, very short explanation. To understand panel data, much more reading will be necessary.

But I hope the text below, from the Stata Manual, will become more clear now:

This means that there are women who sometimes have msp 0 and at other times have msp 1. The within percent tells us the fraction of the time a woman has the specified value of msp.

[William Lisowski]

The Overall column tells us that 40.94% of your observations have vote "Yes", 59.06% have vote "No", and together there are 3053 observations of your 237 respondents, so some respondents are missing in some time periods.

The Between column tells us 193/237 = 81.43% of your respondents have vote "Yes" in at least one of his or her observations (ever voted) and 221/237 = 93.25% of your respondents have vote "No" in at least one of his or her observations (ever didn't vote).

The Within column tells us that for your 193 respondents who have vote "Yes" in at least one of his or her observations, 50.14% of their observations are vote "Yes", and for your 221 respondents who have vote "No" in at least one of his or her observations, 63.45% of their observations are vote "No".

You wrote for the Within column "50.14% of those "ever" voters were always voters". However, "always voters" is not correct; rather, for the group of 193 "ever" voters, 50.14% of those observations were vote "Yes".

[Rose Simmons]

Marcos Almeida thank you for the explanation and examples (and pun!), I can now understand the text from the Stata Manual better

William Lisowski thank you for the detailed explanation, I can now notice the difference between observations and respondents, understand which figures have been used to calculate these percentages, and realise the correction of my Within statement, thanks again

[Hovhannes Nahapetyan]

Thank you so much William Lisowski, I finally understood the difference, have not seen any other explanation that would be remotely as clear as your answer!

[Bob Wen]

William Lisowski The clearest explanation of xttab I have ever seen. Thank you!