Blog post: Project overview
This data science project aspires to inform all persons entitled to vote about the possible influence of mass media on their voting behavior.
With this information, all those eligible for voting may reflect on different media that they consume during the pre-electoral and voting periods, so that they will be less likely swayed by candidate media coverage. Hopefully the information we communicate serves to increase voter independence.
To estimate the media coverage of presidential candidates, we study data on both the frequency and positivity/negativity of candidate mentions in mass media.
These properties answer two very important questions:
– Who is discussed?
– How are they discussed?
The true weight of our findings will of course be determined when the election results come in. However, we have compared candidate scores created using the two previously mentioned properties (total media mentions and percentage of positive mentions) with the results of a recent Yle presidential poll, and we found a significant relation between them.