Hi, Prof. King,

I just tried Amelia for two days. Everything is new to me.

I have a longitudinal data sets with 283 children across four time points.

Since the longitudinal structure of the data, I tried to impute the missing values using the time series cross-sectional function of Amelia.

Here is first four cases in my data input:

famid age safsib dpeer risk npar

1 13 5 8 0 0
1 15 8 15 1 7
1 17 . . . .
1 19 10 20 1 10
2 13 2 15 0 0
2 15 . . . .
2 17 . . . .
2 19 . . . .
3 13 10 8 0 0
3 15 10 5 0 0
3 17 8 10 2 3
3 19 . . . .
4 13 . . . .
4 15 10 14 1 5
4 17 4 20 2 8
4 19 . . . .

Famid represents each child and age indicating the time of accessment (2 years apart).

And I sorted the data by famid and age.

I want to impute four variables: safsib, dpeer, risk, and npar.

My global setting is:

Amts: 2

Amcs: 1

Amusets: 0

Amusecs: 0

Amlagvs: 3 4 5 6

Amtstep: 2

I can read the data into the program, but when I finished all the global settings and pushed "tab" key, the program pop up a warning saying:

The data is not sorted by time within each cross section.

The data need not be sorted by cross section, but within each

subset of data corresponding to one element of the cross section,

the data must be sorted by time. That is, the next observation of any cross section, must be the next observation in time.

As you can see, my data have already been sorted by time within each cross section.

So, I don't know why I got the warning message.

Could you provide any suggestions?

Thanks a lot!

Yi-fu

=============================
Yi-fu Chen, Ph.D.
Research Statistician
Center for Family Research
University of Georgia
1095 College Station Rd.
Athens GA 30605