How do I expand period data into data by year?

[Kyle Cook]

Hello, relatively new user here, using stata for my university class and have a base level proficiency at it, and have self taught myself a good amount of what I have needed so far from this forum.

I am trying to merge two datasets together for a regression. The problem is that one of the datasets is paneled by country/year while the other one is country/period. These periods are 5 years each. What I am trying to do is duplicate each period 5 times, and then either rename the period variable or add a year variable so that each of those duplicates now represents one year from the period. For example, for period one, which is 1960-1964, I want to duplicate the period one observation 5 times, and rename so that there is one for 1960, 1961,etc. This way I can merge it smoothly with my other dataset. I have done some research and found the 'expand' command, but I could find no way to have each duplicate have a different name, making it difficult for me to give them each a unique year in an efficient way.

Any help would be appreciated, and apologies if this is a question that has already been answered, did my best to search but may have not used the right key words.

[William Lisowski]

Instead of duplicating the country/period data into 5 yearly copies, in your country/year dataset calculate a new "period" variable that is - following your description - 1 for years 1960-1964, 2 for 1965-1969, etc.

So something like

use country_year_data
generate period = 1+floor((year-1960)/5)
merge m:1 country period using country_period_data

where the floor() function rounds down a fraction to an integer.

[Kyle Cook]

That makes total sense, I definitely got too focused on solving the problem in a much more complicated way. This was incredibly helpful, really grateful for the time you took and the response!

[Nick Cox]

@WIlliam Lisowski's theme is expanded upon in https://www.stata-journal.com/articl...article=dm0095


Here's another way to do it:

. clear

. set obs 10
Number of observations (_N) was 0, now 10.

. gen year = 1959 + _n

. gen period = ceil((year - 1959) / 5)

. l

     +---------------+
     | year   period |
     |---------------|
  1. | 1960        1 |
  2. | 1961        1 |
  3. | 1962        1 |
  4. | 1963        1 |
  5. | 1964        1 |
     |---------------|
  6. | 1965        2 |
  7. | 1966        2 |
  8. | 1967        2 |
  9. | 1968        2 |
 10. | 1969        2 |
     +---------------+