NEWS16 March 2016
All MRS websites use cookies to help us improve our services. Any data collected is anonymised. If you continue using this site without accepting cookies you may experience some performance issues. Read about our cookies here.
All MRS websites use cookies to help us improve our services. Any data collected is anonymised. If you continue using this site without accepting cookies you may experience some performance issues. Read about our cookies here.
Insight & Strategy
Columnists
Impact magazine is a quarterly publication for MRS members. You can access Impact content on this website.
NEWS16 March 2016
Asia Pacific Europe Latin America Leisure & Arts Middle East and Africa News North America Public Sector Trends UK
UK – Internet panel sex data scores just two-out-of-five stars for its veracity and reliability according to a ratings system in a new book by Professor David Spiegelhalter, professor of risk at Cambridge University.
Spiegelhalter was speaking at the MRS’s annual conference, Impact 2016, in an insightful and humorous presentation that examined sex and stats.
The first challenge of studying people’s sexual behaviour is defining sex itself, he told delegates. "How can we ask questions of peoples’ behaviour that they might be embarrassed about telling us? One of the problems with sex surveys is that people understand different things by sex."
The issue of defining what sex is came to prominence in the late 1990s when US president Bill Clinton and White House aide Monica Lewinsky were embroiled in a scandal over her performing oral sex on him.
"What was interesting was his definition of sex," Spiegelhalter said. "He claims that he didn’t have sexual relations with her, while she claims that she did have sexual relations with him." There was a vote in the Senate and it found in his favour. Oral sex was not sex.
Spiegelhalter went on to explain how he has devised a star rating system that he expounds on in his recently published book on sex and statistics, Sex By Numbers.
"One of things I do in the book is to crudely give star ratings to data," he said. "If you Google sex stats most of what comes up is nonsensical."
Spiegelhalter awards four stars to surveys "where you can believe the numbers and don’t put margins of error on them". Such statistics include that in 1973, one in 20 16-year-old girls got pregnant
Three-star stats are for what Spiegelhalter considers "reasonably accurate numbers, say up to around 20% up or down".
Internet panels fall into the two-star category for Spiegelhalter, a rating that marks data that "can be used as a very rough ballpark figure, but with details that are unreliable".
But the most derisory rating applies to the specious findings unearthed in magazine surveys and their ilk, which come bottom of Spiegelhalter’s system, scoring just one star.
Spiegelhalter is donating his fee for speaking to charity Karuna, which is dedicating to empowering women in South Asia.
Newsletter
Sign up for the latest news and opinion.
You will be asked to create an account which also gives you free access to premium Impact content.
Survey questions asking participants to categorise themselves by race are still essential to provide a metric for u… https://t.co/yMOW4vVrcw
ESRC opens environmental research fund https://t.co/PrdQd3uuWq #mrx #marketresearch
Voxco acquired by private equity company https://t.co/0SvakRlTaO #mrx #marketresearch
The world's leading job site for research and insight
Hasson Associates
Quantitative Research Manager, brand strategy
£40000–44000
Affinity Water
Customer Insight Executive
£33–34,500
Hasson Associates
Qual Research Manager, Pharma Experience
£35000–40000
Featured company
Town/Country: London
Tel: +44 (0)20 7490 7888
Kudos Research are leading providers of premium quality UK and International Telephone Data-Collection. Specialising in hard to reach B2B and Consumer audiences, we achieve excellent response rates and provide robust, actionable, verbatim-rich data. Methodologies include CATI, . . .
Related Articles
The post-demographic consumerism trend means segments such age are often outdated, from @trendwatching #TrendSemLON
0 Comments