It was standing room only at the ‘Margins of Error’ event on 14 May 2013, a major policy debate co-hosted by the Royal Statistical Society with pollsters Ipsos Mori and King’s College London.
The event, which examined public understanding of statistics in an era of big data, is the first of a series of public lectures discussing perceptions of statistics. This lecture marked the launch of new research conducted by Ipsos MORI on behalf of the RSS and King’s College London on the public’s understanding of statistics and their perceptions of the role it plays in public policy.
‘Margins of Error’ was chaired by RSS president John Pullinger, who introduced proceedings by talking about the aims behind the RSS getstats campaign. He introduced the panel of speakers, comprising Ipsos MORI MD Bobby Duffy, UK Statistics Authority chair Andrew Dilnot and head of the School of Social Science and Public Policy at King’s College London (and former RSS president) Denise Lievesley.
Bobby Duffy kicked-off proceedings by presenting the findings of new research which identified a lack of public confidence in politicians using official statistics accurately when talking about their policies – only seven per cent felt they did so. He also, however, revealed a rise of public trust in science and also a marked increase in civil servants.
Public attitudes to numbers were mixed: while 92% of respondents correctly deduced that 50 is 25% of 200, only a quarter correctly calculated that the probability of getting two heads in two tosses of a coin is 25%. The research also suggested that the public appeared not to value understanding of numbers very highly. When asked which would make them most proud of their child, 55% said ‘being very good at reading and writing’, as opposed to just 13% who opted for ‘being very good at numbers’.
Andrew Dilnot focused on the importance of data to challenge public misconceptions. He talked about policy makers being too focused on small quarterly changes in GDP, which has actually risen fivefold since 1948. Another misconception, Dilnot pointed out, was so-called ‘soaring teenage pregnancies’ which statistics show have actually been dropping in recent years. He demonstrated the importance of conveying the level of uncertainty with figures, using recent norovirus figures as an example. He also talked of the importance of looking at the ‘bigger picture’ rather than ‘getting wound up about small, short term trends’.
Denise Lievesley gave the statistician’s perspective, and spoke of the difficulties often faced by them when presenting data. She cited a number of ‘tensions’ faced by statisticians in their work, such ‘relevance vs autonomy’, ‘trust vs scepticism’, ‘measurement vs quality’ and ‘pragmatism vs purism’. She declared that statisticians should aid interpretation, adding: ‘numbers don’t speak for themselves’.
A lively discussion followed, which covered the lack of statistical training on university journalism courses, supporting statisticians around the world and the challenges of big data, which prompted Andrew Dilnot to conclude: ‘Big data does not necessarily equate to more intelligence’. John Pullinger ended the debate on a brightnote, pointing out that it ‘has the potential to do so and it is up to all of us to make it happen’.