We used automatic text analysis based on previous British Election Study surveys to automatically code the open ended most important issue responses for wave 1 in the variable miilabel. A second variable miilabel certainty includes the probability that the statistical model assigned to the chosen category. We suggest not relying on MII categories coded with a low level of certainty.
We will release a blogpost describing the text coding process and including the R code used that will allow anyone to choose a different approach for the coding if they prefer.
This automatic coding will be updated with manual coding of the text after the BES team decide on a new coding schema for the most important issue question.