Learning about statistics

Talking about sex is complicated, and getting to the truth matters.

It’s important to remember that the data in Natsal are what participants have reported, however, for various reasons the truth isn’t always what’s reported.

A great deal of work goes into delivering the survey in a way that helps people answer accurately and honestly, so how the questions are worded is given a lot of thought to ensure that the resulting data are reliable.

So... how many sexual partners? 

Some people may have problems remembering something exactly, for example, the number of sexual partners they have had. This is called ‘recall bias’.

Instead, they might provide their best guess. This leads to ‘digit preference’ as they are more likely to report a number ending in 0 or 5. Some people may deliberately choose to ‘revise’ their response to make it sound better, or rather more in line with their perception of cultural norms. This is called ‘social desirability bias’.

For example, historically, men were more likely to ‘round up’ their number of sexual partners while women were more likely to ‘round down’ this number. For these reasons, it’s important to describe data from surveys like Natsal as ‘reported data’.

Couple kissing

Representativeness and reliability 

It’s also important to remember that for logistical reasons all surveys can only capture a sample (or proportion) of the whole population. The larger the sample the more accurate it is at painting a picture of the population it seeks to represent.

While Natsal is designed to get accurate estimates for the general population, and key groups like all men and all women, this isn’t always the case for minority groups such as some ethnic, gender and sexual minority groups. It’s important therefore to take account of the number of people upon which the statistics are based (this is called ‘the denominator’).

When the denominator is less than 30 people then the statistic is unlikely to be very reliable, and caution is still needed when interpreting the statistic when the denominator is between 30 and 50. For these reasons, you may find that when using the filters in the Comparing Natsal statistics tool it does not provide results for some combinations of characteristics, in which case you may want to consider broadening the age range of interest, for example.

Man on laptop

Oversampling, why and how this is done

Sometimes surveys like Natsal will ‘oversample’, that is, select more people in particular groups to take part as a way of getting more robust estimates for certain groups.

This needs to be factored in when analysing the data so that the results can still be considered as broadly representative of the population as a whole. This is achieved through statistical weighting.

Statistical weights also serve to take account of differences in how certain groups responded to participating in the survey. For example, men under 30 were most likely to refuse to take part in Natsal-3 so there aren’t as many young men in the Natsal-3 sample as we would have liked.

Couple in bed

What about the people who didn’t take part in Natsal?

While we can weight the data from people who choose to participate to try and account for those who do not, there is little that can be done to compensate for the fact that those who don’t may be different from those who do (‘participation bias’).

This is why Natsal interviewers work so hard to emphasise with would-be participants the importance of taking part regardless of their sexual experience, so that Natsal represents people from all walks of life.

It’s also important to bear in mind that estimates for much smaller groups in the overall population, such as lesbian women, are based on people who identified as lesbian and who chose to participate in Natsal. Reported Natsal statistics for lesbians therefore do not necessarily reflect all lesbian women living in Britian. 

Pride flag in bedroom


Some important things to know about statistics

Frequently asked questions (FAQs)


Rate this page

Back Next