While much has been made of the drop in ratings for this year’s season of Big Brother, not to mention cancellation of the Friday and Sunday episodes, social media activity is actually showing a slight up-tick on last year, based on the generic keywords (e.g. #bbau, @bbau9 etc).
The series of stunts which have seen the return of Tim, and most recently Ben to the Big Brother house are at least causing the social media audience to engage, although whether that’s in a positive manner is less clear:
Year on year then, there are a couple of key differences.
The weekly shape of social media activity last year was extended by the (short-lived) Saturday show, which brought in low ratings and even lower social media activity.
This year’s Sunday show (also, it seems, short-lived) has fared significantly better, with social media volume not dissimilar to the rest of the week. The Monday to Thursday shows have a slightly different shape to last year, largely due to the double evictions and the shifting of nominations (and change from a live to pre-recorded format), while the shortened Friday show had lower volume in both years as you would expect.
Overall though, volume appears to be up year-on-year.
Evictions update
Earlier in the month, we discussed the extent to which Big Brother evictions could be predicted by social media data. Since then, there have been two further evictions, and there remains some unpredictability, and presumably some variation between the social media audience and the voting public.
While The Hypometer has shown a loose correlation between lack of conversation and being evicted in the past, factors such as controversy in the house and housemate fandom on the outside can lead to increased conversation for a housemate, which isn’t necessarily an indication of support.
For the week beginning October 8, the nominated housemates were Lawson, Travis, Priya, Sam, Sandra and Jason with Travis receiving the most votes to save (37%) and Sam sent home with the least (7%):
While social media would have lead us to predict that either Sandra or Jason was going home rather than Sam, Travis was the most talked about after Lawson and consequently saved, as we saw in the past with Skye.
This week’s nominees were Ryan, Sandra, Priya, Jason and David, with Ryan being saved with 38% of the votes and Sandra evicted with 9% and Jason with 16.9%:
Similar to the previous week, Ryan being saved could have been predicted by looking at the most talked about housemates on both Twitter and Facebook – however the data suggested that Sandra was in the running to be saved rather than evicted. Overall, David staying in the house is a clear example of how voting support for a housemate may not be linked to social media conversation.
While social media may not be able to determine which of the lower ranked contestants is most likely to face the axe, it has highlighted those at-risk, which has often differed from the bookmakers perception. It is also proving to be an accurate tool for predicting housemates most likely to be saved.
If this is true and high social media mentions lead to being saved every week, then at this stage you would have to make Skye the favourite to win Big Brother 2014:
Skye has been the most talked about housemate on Twitter and Facebook for the season; overtaking Priya on Facebook in the last week.
With TV ratings falling, but social media audience holding steady, it can perhaps be argued that users of social media are increasingly representative of the viewing public. Perhaps, then, social media is a key way to keep the audience engaged with shows such as Big Brother. Those who participate in such discussions experience a loyalty to the television broadcast that other viewers do not.
The analysis of the data to date also suggests that social media is a good predictor of positive decisions: who will be saved, who will win an ad hoc competition (such as the Lawson and Aisha dance-off we discussed previously) – but less so for negative decisions such as evictions.
Sentiment analysis remains a tricky field, especially on social media where computationally detecting irony in 140 characters remains unsolved, but this may well be the missing element in using social media as a predictor of television audiences.