0
$\begingroup$

I am studying the cold war interventions of the United States and Soviet Union. To that end I plan to create a dataframe with the following format:

War_id | Year | country | opponent    | us_int | su_int | other vars about country
--------------------------------------------------------------------------------
  1        1946   FakeNation   rebels A      0       1
  1        1947   FakeNation   rebels A      0       1
  1        1948   FakeNation   rebels A      0       0     
  2        1946   AlsoFake     rebels B      1       0
  2        1947   AlsoFake     rebels B      1       0
  2        1948   AlsoFake     rebels B      1       0

So for each individual nation there are rows for each year from 1946 until 1991. Both us_int and su_int are dummy variables that denote an intervention from either the US of the SU in that country in that year.

However, when I started merging datasets I got a different result. What I had not accounted for was that some countries fight multiple conflicts at once, or fight multiple opponents at the same time. As such, rows where the year and country variables are duplicates.

War_id | Year | country | opponent    | us_int | su_int | other vars about country
--------------------------------------------------------------------------------
  1         1946   FakeNation   rebels A      0       1
  1         1947   FakeNation   rebels A      0       1
  1         1947   FakeNation   rebels C      0       1
  3         1947   FakeNation   rebels D      0       1
  2         1948   FakeNation   rebels A      0       0     
  2         1946   AlsoFake     rebels B      1       0
  2         1947   AlsoFake     rebels B      1       0
  2         1948   AlsoFake     rebels B      1       0

(I know I could merge all year and country rows together with a 'other_wars_present' dummy variables, but then I lose a lot of conflict specific information that I would like to use.)

So my question is a follows:

I was planning to run logistic regressions to see which other vars about country influence US or SU interventions, but now that there are 'duplicate rows' for some years, will these rows distort / dilute the results? (As some countries are now over represented, but its somewhat artificial as the rows are all from the same year)

$\endgroup$
  • 2
    $\begingroup$ It definitely does not invalidate your data - this is incontrovertible fact! It doesn't even completely invalidate your analysis. It does mean that there is variance in your data that is explainable by the fact that the same country is involved. This can be accounted for in your model. A common approach would be to use a mixed model, with a random intercept for country (and possibly also for year). $\endgroup$ – mkt May 24 '18 at 11:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.