Not enough black friends to go round …

After reading yet another story about the state of community / religious (for which read ethnic or race) relations, I thought I’d do my first post about race. This should have been questioning the logic of categorising people in fixed, homogenous and overlapping groups (sometimes described as billiard ball cultures). This would be the natural place to start. However, it’s late at night and deconstructing the concept of race will need a bit more effort, so instead I’ll comment on the state of interracial friendship in the UK (despite the problematic nature of this idea) as an indicator, however flawed, of integration.

Every now and again, we read stories from pessimists showing how divided we are on grounds of race and from optimists that show how we are all one happy family. The evidence comes from knowledge of people’s friends, relationships and family formation.

This story starts with a survey. It was conducted by the internet pollsters YouGov (done some good work, some bad, I was on their panel at the very start but quickly got bored of filling in the forms) on behalf of the CRE (again, some good work, some bad, but their heart is in the right place). The headline finding was:

CRE survey shows little integration among UK’s white majority community with ethnic minorities

(http://www.cre.gov.uk/Default.aspx.locID-0hgnew01y.htm)

The 2004 press release begins: ‘Most white people (94%) have few or no ethnic minority friends, while nearly half (47%) of non-white people, say most or all of their close friends are white.’ This was found by asking people how many of their 10 to 20 close friends were white or from ethnic minorities.

Unfortunately, what this analysis fails to take account of is the extremely important fact that most people in Britain are white. Precisely because white people make up the vast majority of those living here (about 9 in 10) the average person should have about 10% of their friends being BME.

Indeed, we can imagine a world, and calculate what would happen, where there is perfect integration. If we take the 2001 census figure of  7.9%, and people choose their friends randomly from the population (so ignoring barriers of class, geography, age as well as ethnicity), we find that for the question asked everybody should have a majority of friends who are white (whether white or BME).

 

BME friends

from 20 friends

from 10 friends

0

19.28%

43.91%

1

33.08%

37.67%

2

26.96%

14.54%

3

13.87%

3.33%

4

5.06%

0.50%

5

1.39%

0.05%

6

0.30%

0.00%

7

0.05%

0.00%

8

0.01%

0.00%

9

0.00%

0.00%

10

0.00%

0.00%

11-20%

0.00%

N/A

 

If everyone picked 20 people, 19% would have all white friends, 33% would have one BME friend, 27% would have 2 and 14% would have 3. The number having 8 or more would be 0.008%.

If we compare the YouGov survey (using the 2005 results) to this we find that white people have more BME friends than would be expected and BME people have fewer white friends than would be expected, a different spin on these results.

Yes, more white people have no BME friends than this model predicts, but the number saying their friends are all or mainly white is 95% when it should really be 100% to the nearest whole percentage.

Amazingly, the number surveyed saying they have half or more friends who are BME is 5%, 600 times more than expected!

And, of course, if race is no barrier then this should be true for the BME population too. Almost every one should have a majority of white friends. By the style of analysis of the CRE report, BME people are less integrated with the white population than the white population is with the BME population, adjusting for population size. This last sentence, however, is nonsense. The problem with the CRE’s analysis is that it treats integration as something only to be done with BME people, i.e. having BME friends is integration. But surely if you are say, of Bangladeshi origin, integration means having friends who are white, black African and so on. The survey doesn’t even ask about what kind of BME people these friends are – they could all be Bangladeshi and how integrated is that? (I won’t go into the badly worded question that doesn’t allow us to know exactly how many friends.)

Not only is this new analysis conceptually nonsense, it also doesn’t quite fit. Surely if we take two populations and partially mix them up, how can one be less integrated with the other than vice versa? That is, the black people have white friends but the white people don’t have black friends. Surely someone is lying, whether to big themselves up or to hide something. A little bit of maths shows this up nicely, in the same way as the sex partner conundrum (http://exploringdata.cqu.edu.au/sex_prtr.htm) in which the men say they have had more partners, on average, than the females.

Using the YouGov survey, and assuming BME and white respondents picked the same number of people on average (let’s use ten), it’s easy to work out the number of interracial friendships (or more accurately white-BME friendships). I think it’s fair to say ‘almost all’ can mean eight or nine, ‘most’, six or seven. In this case the number of white-BME friendships, according to the white people, is from 44.4 million to 66.6 million. According to the BME people the figure is somewhere between 21.1 million and 24.8 million. Either the respondents aren’t representative, or they are lying, or they have different meanings of close friendship. Whichever, the results of the YouGov survey aren’t valid, and the analysis just makes things worse.

Advertisements

Leave a comment

Filed under Statistics and simplicity

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s