Filling in missing years and interpolating categorical values corresponding to that newly created years.

[Philip Samji]

Hi All,
I am currently undergoing a huge crisis in my dataset.
I am using a World Value Survey data to work upon the countries of Bangladesh, India and Pakistan to measure the impact of globalization over trust amongst people and confidence over the governmental institutions of the people across these 3 countries.
So, my problem is this:
1. The dataset is a timeseries dataset but the years covered in the survey are 1996, 2002 and 2018 for the country of Bangladesh and the same for India and Pakistan but with a different years. I want to create the years which are missing in this dataset.

So, my real query is how to create this missing years and even if I do it, how many observations should be there.

2. Can we interpolate a categorical variable and if this categorical variable is interpolated in the missing areas, how to maintain the values as the same categories that is preceding and succeeding that area of missing values, rather than the interpolated values changing to continuous values.

Can someone help me with this?

[Nick Cox]

I don't know if you are going to call this helpful.

By the way, the problem is strictly one of absent values, not missing values!

If those are the years available, then bulking up to include all intervening years amounts to guessing at the great majority of your data. You can't defensibly add information that way. I think interpolation can be defended to fill small holes when you are confident you are understand the data generating process, but this dataset is mostly very big holes, and over those years there have been all kinds of social, economic and political ups and downs in those countries, just like almost anywhere else.

I advise against anything except working with the data as they are.