Repeated measures

[Ansuyah Magan]

Hi
I have longitudinal data. I used "xtreg variable i.age" in Stata 14.0. I get regression coefficients for each age. Does the regression output reflect the changes for every year from a single baseline time point (the time of 1st measurement) or does it reflect the year on year changes? I would like the coefficients for the year on year changes.

[Carlo Lazzaro]

Ansuyah:
I'm not clear with the way you -xtset- your data, first, nor you report whether a -hausman- test for performed.
Yiour query seems to focus on the different flavours of within and between variance.
Could you please provide an excerpt/example of your data via -dataex- (type -search dataex- from within Stata to intall it).

[Joseph Coveney]

It reflects the changes from the baseline. For year-on-year changes, use

contrast ar.age, nowald

after fitting the model.

[Ansuyah Magan]

Thank you for your feedback Joseph and Carlo
Below is a sample of my data
I have successfully done R anova for "bone" against "age" and obtained contrasts.

apv is a biological age scale that was used for each individual.
ypv was calculated from apv as a continuous variable. For repeated measures a categorical variable ypvcat was generated. Stata does not handle negative numbers for repeated measures, so ypvage was generated. For some "ypvage" there are duplicates which I do not want to drop. This is related to the way in which the categories were defined.
Because of the duplicates I could not do an R anova.
I did the following:
xtset data
xtreg bone i.ypvage, cluster(Id)
contrast ar.ypvage works only if I use i.ypvage
The regression coefficients I obtained for each ypvage I presume are for each year against the baseline year as opposed to year on year. The data has been analysed longitudinally using "ypv" to give me an overall slope.
I would like to compare the annual changes in the growth of the bone against ypvage.

id  age apv ypv  ypvcat ypvage bone
104 12 12.65   .27  0  7  9.53
104 13 12.65  1.28  1  8  9.91
104 14 12.65  2.32  2  9 10.66
104 15 12.65  3.25  3 10 10.71
105 10 12.83 -2.71 -2  5  8.23
105 11 12.83 -1.45 -1  6  8.74
105 12 12.83  -.33  0  7  9.34
105 13 12.83   .65  0  7  9.66
105 15 12.83  2.31  2  9 10.13
105 16 12.83  3.29  3 10  9.91

[Joseph Coveney]

I would like to compare the annual changes in the growth of the bone against ypvage.

anova bone id ypvage
contrast ar.ypvage

or

xtset id
xtreg bone i.ypvage, fe
contrast ar.ypvage

[Ansuyah Magan]

My output after xtreg bone i.ypage, fe

Does the coefficient for ypvage 4 represent the slope between data ypvage 1-4 or between ypvage 3-4?

ypvage	coeff	SE
2	0.08	0.12
3	0.27	0.12
4	0.45	0.11
5	0.70	0.11

[Joseph Coveney]

Does the coefficient for ypvage 4 represent the slope between data ypvage 1-4 or between ypvage 3-4?

Well, you could

table ypage, contents(mean bone)

and answer your own question by inspection, but it might be more profitable in the long run to

help factor variables

because Stata's factor variables have some flexibility that might come in handy in the future, and it would be good to get to know them better.

[Carlo Lazzaro]

Ansuyah:
- 0.45 represents the difference between 4 and the reference category (1, in your case);
- the difference between 4 and 3 is: (0.45-0.27)=0.18.

PS: crossed in the cyberspace with Joseph's more helpful reply.

[Ansuyah Magan]

Thank you Joseph and Carlo for your assistance.
Kind regards
Ansuyah