Using Wordscore in STATA to assess trends in press releases

[Sigurd Fosli]

I'm new to the forum, but I was told I might get some help here.

I'm doing an MA in political science and I'm working on a project using the Wordscore tools in STATA to see if a government adopts talking points and arguments found in lobby groups press releases. In my case, I'm looking at the UK gov during COVID to see if the treasury adopted proposals, policy, or arguments from the Federation of Small Businesses and the Confederation of British Industry. However, I don't think I'm doing it right. See example below

First example: tBudgetmar2020 and tBudgetmar2021 are budget speeches in Parliament in March 2020 and March 2021. tCBIPR and tFSBPR are press releases issued by the CBI and FSB, and I'm looking for the term "aid".

Second example uses the same data, but this time I'm looking for the term "economic".

My first problem is that the score is exactly the same for two different terms...

My second problem is that I don't quite understand the transformed score. Does the difference between tBudgetmar2020 and tBudgetmar2021 (0.9151 vs 2.9072) mean that the term I've searched for appear twice as often in 2021 as in 2020?

My data can be found here: https://www.dropbox.com/s/yoi8w6c02z...mple1.dta?dl=0

I'd be grateful for any help

Example 1:

Example 2

[Leonardo Guizzetti]

You have cross-posted this to Reddit. Please note our policy on cross-posting is that we ask you to tell us that you did (and where). This is done to prevent those from potentially helping you when your problem may have already been solved.

[Sigurd Fosli]

I'll be sure to update the post if or when a solution has been found, no matter if the solution is posted here or on Reddit.