This data-driven story seeks to explore the relationship between electability and Twitter engagement as Indonesia approaches the 2024 presidential election.
We found that the average tweet per day and reaction per tweet can help explain a politician’s electability in the poll for the presidential election. Politicians who were more active tended to be more popular in the poll.
We also found that most politicians had not tweeted more between March and July 2022.
We collected the data from Twitter and leading polling agency Indikator Politik. We collected the tweet data through Twitter’s API using R.
The story has gained 428 views as of Jan. 19, 2023
We used R for the data collection, analysis and visualization. To collect the tweet data, we used the rtweet package to request tweet data through Twitter’s API.
We also perform a linear regression in R to predict electability based on Twitter activity, after controlling the engagement or reaction.
Then we use the ggplot2 package to visualize the analysis result.
Context about the project:
This is arguably a novelty data-driven story. While there are many media running a story on presidential election polls, the relationship between electability and social media engagement is seldom covered, especially with a quantitative approach
What can other journalists learn from this project?
I hope this story can inspire other (Indonesian) data journalists to pick up a programming skill and basic statistics or data analytics. These skills allow us to pursue novelty stories and look at big issues like the presidential election from a fresh perspective by wrangling and analyzing large datasets