Final project for survey design investigating attitudes towards medicare for all based on Twitter data.
Using tweets referencing a policy of medicare-for-all, we want to discover the attitudes and sentiments Twitter users have for the policy.
Real time Twitter data comes from https://www.vicinitas.io with tags: medicare for all, single-payer system, and universal health care. Data was collected starting on October 27th up until November 3rd 2020. We chose these dates because these were the days leading up to the 2020 election. We figured this is when Twitter users would be discussing policies on the platform. From our data collection, we collected a sample of about 30,000 tweets. However, after removing the duplicates, which came up due to retweets, we were left with a sample of about 10,000 tweets.
Because we are relying on Twitter data, our survey does not have a questionaire. Instead, we are taking our data from the textual analysis done to the tweets. We are looking at what users' general attitudes are towards the policy.
Sentiment scores calculated using Bing index, producing a positve and negative sentiment score for each tweet. A total sentiment rating was then calculated and used for further analysis. See RMD file.
Though we have already gone ahead with a more in-depth sentiment analysis, here is a preliminary study we conducted by manually going through a set of tweets and assigning them as positive, negative, or neutral:
The search of keywords on twitter regarding Medicare for all resulted to over 2000 tweets that have focused on the topic. From the 2000 identified tweets, we pulled out 753 usable tweets for the analysis as some of the tweets were in foreign languages making it difficult for one to assess them and understand whether they are neutral, positive or negative on Medicare for all programs. The results from the 750 respondents who took part in the study are shown in the figure below
Positive 518 69%
Neutral 139 18%
Negative 96 13%
Total 753 100%