How to Read Predata’s Digital Momentum Scores for the French Election
What are the signals?
We build signals inside the Predata system for each candidate based on a curated selection of web pages across Wikipedia, Twitter and YouTube. Each signal, very broadly, represents/captures the official campaign message for that candidate. The selection of sources includes the Wikipedia pages for the candidate and the candidate’s party/political movement, the Twitter accounts for the candidate and the candidate’s main advisors/spokespeople, and all the material on the candidate’s official YouTube channel. As activity around a bundle of sources (i.e. a candidate signal) increases, so does the candidate’s score. “Activity” should be taken to include page views as well as engagement (i.e. replies, retweets, edits, comments, etc.). We then aggregate all the candidate signals into a master signal for overall conversation online about the election; this signal also includes neutral/background material on the election and the French political system generally.
What is a digital campaign score?
The digital “scores” for each candidate (i.e. the daily values that form the basis for the chart above) are daily, rolling 28-day correlations between the candidate signal and the master signal for overall conversation about the election. A score of 65% means the candidate signal has a correlation of 0.65 to the signal for overall online conversation about the election on that day. A higher score for candidate A vs. candidate B means, generally speaking, that candidate A is tracking the overall conversation online about the election more closely than candidate B.
What is digital momentum?
However, having a higher score really only becomes important when the gap is significant; where many candidates are clustered together, the score/correlation level/percentage (all the same thing) matters less than the shape of the candidate’s recent signal history (i.e. are they gaining or losing momentum?). “Digital momentum” is really a derivative of the score; it’s the rate of change in the score over a given period. From the chart above, it’s clear that Le Pen and Hamon have gained momentum over the last week, while Fillon and Macron have been largely stagnant. Mélenchon was also stagnant until a day or two ago, when he started gaining momentum.
Ultimately, when evaluating how individuals are “performing” in the digital campaign, you should think about shape/trend/momentum first, then consider the value of the score. Comparing the shape of the digital momentum scores to the polls and to benchmark securities for event risk associated with the election (e.g. the French-German 10-year spread) is also a useful comparative exercise. ⏪
For more information, contact Aaron Timms, Predata’s Director of Research, at firstname.lastname@example.org.