Extrapolating several values

[l cilek]

Hi, I have county-level data (variable "id2") of median age (variable "medage") for three years (variable "y," written as "1" for 2000, "3" for 2008, and "4" for 2012 ). I am trying to extrapolate this data to 2016 ("y" 5) individually by county.
I have my data so that it can be reshaped into wide or long form

I've tried to use this (long form data):

sort id2 y
tsset id2 y
tsappend, add(1)
xi: reg medage i.id2*y
predict medage2016

and this (short form) from which I could find out how to :

by id2: regress medage2012 medage2000 medage2008

In the second example, stata says that there are not enough observations to complete the regression. In the first example, I still can't quite figure out what is happening, but it involves creating several variables and interpolating the results.

I also tried using the ipolate function by creating a missing variable for all values in 2016 and using "by id2:" In this case, stata said there were too many variables...I guess the ipolate function is only for two points of observations?

Any suggestions, or is something written wrong?

Thanks!

[l cilek]

ok, I've figured out that by using the first option in long form:
sort id2 y
tsset id2 y
tsappend, add(1)
xi: reg medage i.id2*y
predict medage2016

the new medage2016 variable has the predicted median age values at all y values (including the missing "2" values for unknown 2004 data).

i have several variables to extrapolate, so I've repeated the last two lines when all of these have been created, i will create a lag variable that will help me separate the predicted 2004 observations from the 2016 observations, then combine the extrapolated 2016 values so that each county has an observation for 2016.