The 2015 British election outcome will be remembered as the time when the polling industry got the outcome so badly wrong. The run up to Election Day also highlighted that the political and academic commentary that many media broadcasters relied on largely accepted the data that pollsters provided as a given. This over-reliance resulted in a ridged and one-dimensional view of voter behaviour. There was little room for alternative interpretations. Why was this the case?
I want to answer ‘the why’ by providing three insights.
- The data produced and analysed by pollsters was presented as ‘fact’. This made it more difficult to challenge because a ‘fact’ is naturally framed as a given. Taking from social anthropology, the pollsters in the media took on a similar role as religious leaders do in ritual. These rituals work to formalise the congregation and prevent rebellion or difference. Pollsters had power because the data they produced, analysis and presented symbolised knowledge of a sacred order that is rarely challenged.
- Polls and pollsters are embedded within the cultural ecosystem of elections. Their visibility on the media has become institutionalised. The media rely on pollsters to give easy-to-consume data that is used to build political analysis and discussion. In the run up to Election Day the pollsters took on the role of the objective voice which gave the perception of an impartial and balanced view. This framed how audiences consumed and perceived the data. In other words, if the pollsters give their opinion the audience should accept it.
- The nature of the data gathered by pollsters lacked the ability to understand the deep cultural trends that was shaping voter thinking. The data was too ridged and was unable to tap into and deconstruct voter behaviour from holistic sociological and anthropological point of view. Furthermore, the data was driven by the need for quick and cost effective insights rather than in-depth and informative insights.
Nine months on from the 2015 elections the post-mortem into ‘why’ there was such a wide spread failure in predicting the outcome of the elections has been published.Professor Sturgis , who has led the independent review for the British Polling Council, makes the point that “there needs to be a shift in emphasis away from quantity and towards quality”. Furthermore, he suggests that “the emerging upshot is that the companies are going to have to be more imaginative and proactive in making contact with – and giving additional weight to – those sorts of respondents that they failed to reach in adequate numbers in 2015”. In other words, the current approach is too thin and the methodology does not capture the right type of data.
How can data be gathered and used in more creative and informative ways by the media?
I want to propose a new approach to understanding voter behaviour that is based on cultural analysis. Cultural analysis seeks to understand how everyday culture shapes opinion, behaviours and values amongst individuals and communities. It focuses on deconstructing the deeper cultural stories and drivers that influence behaviour and thinking. Importantly, cultural analysis does not aim to produce ‘facts’ but instead works with different streams of data to produce intelligent and informed interpretations of, in this case, voter behaviour and thinking.
A cultural analysis model would combine two research approaches, each holding the same weight of importance. The first would be a quantitative approach that would be based on a more robust model of the data that pollsters provided in the 2015 UK General Election run up. This would bring to light important macro trends into voter thinking. The second approach would be qualitative, using culturally sophisticated approaches such as ethnography (including mobile ethnography), focus groups (which are conversation led), social media discourse analysis and interviews over the period of the election campaign (and ideally before).
The outcome would lead to media accounts based on an informed interpretation, where political analysts, academics and journalists discuss, challenge and build on the cultural insights and trends.
However, as the quote below from the editorial from the Guardian (19th January 2016) shows, there is still some way to go before we can move from a ‘fact’ based data model to an interpretative based data model. “For people who care about politics, flawed polls remain better than total ignorance”. Accepting something that is flawed is surely a form of ignorance as well.