Data Bytes - Tweeting about exams

Data Bytes - Tweeting about exams

At what time of day were the most exam-related tweets sent this summer? New research from our Data and Analytics team reveals the trend behind this phenomena.

In his recent blog on the 2016 summer exam series, Cambridge Assessment Group Chief Executive Simon Lebus drew attention to the increasing use of social media platforms by exam candidates. Whereas previously candidates would have conducted their "exam postmortems" in private conversation, these discussions are now often shared on Twitter, Facebook, and other public online forums.

This summer we undertook to measure this phenomenon by using the Twitter Streaming API to record exam-related tweets in real time. We will be undertaking further research and analysis on the data we have collected but, in this Data Byte, we present a high-level summary of data covering the Summer 2016 exam session.

View PDF

What does the chart show?

We collected a total of 6.44 million tweets during the period 14 May to 30 June (excluding the 'half-term' week beginning 30 May when no exams took place). The chart shows the median number of tweets per minute, grouped by weekday. We have used the median rather than the mean because it gives us a picture of a 'typical' day; the effect of days with extremely high numbers of tweets is minimised.

To aid interpretation, we have highlighted the time periods when exams typically take place. The published starting time of all examinations is either 9:00 am or 1:30 pm, although centres have flexibility to vary these start times by up to half an hour. The durations of exams vary; we have highlighted the most common lengths of 1 hour after the start time (dark shading) and 1 hour 30 minutes (light shading).

We would expect a certain level of noise in the data because there is exam-related discussion on Twitter all year round. In addition, some of our search terms may pick up tweets that are not directly related to exams. The thin grey line shows the typical volume of tweets that we collected during July after the end of the exam session. Because the volumes we collected 'in session' are markedly above this baseline, we can have confidence that we are observing genuine trends.

Why is the chart interesting?

The chart reveals that exam-related Twitter activity follows a distinct pattern. Message volumes gradually increase in the period leading up to an examination, for example in the evenings and the hours immediately before the scheduled start time. There is much less activity during exam periods, but message volumes increase after the exam is finished. This reinforces the 'public postmortem' idea described by Simon Lebus.

No examinations are held at the weekends and there is a correspondingly lower number of tweets from Friday evenings onwards. However, message activity does pick up on Sunday as candidates turn their minds to the following day's exams.

See more from our Data Bytes series

Research Matters

Research Matters 32 promo image

Research Matters is our free biannual publication which allows us to share our assessment research, in a range of fields, with the wider assessment community.