Her distinguished career path has seen her posted to the UK Data Archive, UNESCO and the African Centre for Statistics at the UN. She has also taken academic teaching roles at the University of Essex and most recently, her current position as the Head of the School of Social Science & Public Policy at King’s College London. As a former President of the Royal Statistical Society, Denise was good enough to take time out to talk to StatsLife on what she believes are the major issues facing statistics today.
From your experience at home and abroad, how would you explain the importance of official statistics in helping to run a country and creating evidence based policy?
They’re absolutely fundamental; you can’t manage if you don’t measure. Without them you can’t develop proposals to what policies you need, to implement the policies and to evaluate them afterwards. So right throughout the process you need high quality statistics.
Policy is always going to be messy; it’s always going to using other sorts of material. We elect governments because we believe in their ideology, so we are sometimes conflicted because ideology isn’t always evidence based. There will be situations where governments need to act and can’t wait for data to be collected. Sometimes they have to confident enough to make decisions with less than perfect data.
Having said that I think there is a lot of data which is underutilised. For seven years I worked in the UK Data Archive, so I am a great proponent for using data that has been already been expensively collected. Which is what is exciting about the census review. There is no rationale for going and collecting information from people that we already know.
How do you think the traditional role of the survey is changing in modern times?
The population is increasingly protective of their own time and also concerned about privacy. Ironically we are less worried about some aspects of privacy in how we put so much information up on the web, but people are more wary about their privacy in relation to the government now.
When you used to get an 80% response rate you were reasonably confident in the representativeness, if it’s down to 50%, are the other 50% completely different? For example, we know young people are less likely to respond so we need to adjust for that.
How do you think technology is affecting our ability to make sense of ever increasing data sets?
Computing power has increased dramatically. That enables so much more analysis, but it also causes problems because the whole of statistical philosophy was predicated on the fact that you had a hypothesis about a relationship and you tested whether it was true or not.
If you are just throwing computer power at a data set to find relationships, you will find them. But some of them may just be chance, so distinguishing the chance ones from the meaningful ones becomes problematic.
From your experience when you were President of the International Statistical Institute, how does the aim to advance statistics differ from country to country?
In broad terms they are similar. However, when you come to the specifics, there are countries that don’t have a good official statistics system; there are countries that don’t have good statistics established at universities. Then there are countries where the government is manipulating the data so the general public doesn’t have access to high quality information. In many countries there is no statistical society, so they look to the ISI to provide this. So the relationship of the ISI and the UK is completely different from the relationship between the ISI and Mali for example.
Do you think the UK has a unique heritage in the statisticians that Britain has produced past and present?
Our statistical heritage is very rich and we do today have some very strong statisticians. In statistics over the last 100 years, our heritage has been remarkable, but if you look further back then France is a very important example. So I don’t think we can say we are the only country with such a strong statistical heritage. Also, if you look at who are the important statisticians now, the United States is producing a lot of them. However if you look at their origin, a lot of them are coming from India and China. So to some extent it’s about how high quality universities can attract top staff.
What are your thoughts on the new GCSE curriculum proposals to move statistics away from maths and more into the scientific subjects?
I’m ambivalent about it; I want statistics to still be closely linked with maths because I think strong mathematical skills are important. Having said that, if you are going to excite and interest students in the subject it must be in the applications areas. Therefore, I think it’s wonderful that statistics is embedded in geography and other social sciences. I think the critical thing is how we are helping and supporting teachers in how they teach statistics.