How a minority can turn into a majority

The recent Arab Youth Survey – widely reported by the world's media and personally endorsed by the ruler of Dubai – made a number of important claims.

One was that a majority of young Arabs (52%) agree that "religion plays too big of a role in the Middle East". This was highlighted in the survey as one of its "top 10 findings".

In a blog post yesterday I argued that the 52% figure – along with numerous other figures in the survey – is probably wrong because of flaws in methodology. This is because the survey appears to have aggregated results from a series of polls in individual Arab countries without accounting for differences in population size.

Further scrutiny of the survey report – published by ASDA’A Burson-Marsteller, a Dubai-based PR firm – suggests that not only is the 52% figure wrong but young Arabs who think religion plays too big a role in the Middle East may be a minority rather than a majority.

The problem with not allowing for population differences is that it over-represents opinions from the smaller countries and under-represents those from the larger countries. In some cases this can result in a minority being wrongly reported as a majority.

A quote highlighted in the Arab Youth Survey

On the question about the role of religion, the survey report provides some additional information besides the 52% figure for Arab youth across the Middle East. It also says that in the GCC countries 61% agreed that religion plays too big a role, in the Levant and Yemen 44% agreed, and in North Africa 47% agreed:

As I explained in yesterday's blog post, the 52% figure appears to have been arrived at by adding the results of 1,400 interviews from GCC countries, 1,000 interviews from the Levant plus Yemen, and 1,100 interviews from North African countries. If so, this would not give an accurate overall picture of young Arabs' opinions.

But if the 52% figure is wrong, what would a more accurate figure be? Without access to the survey's raw data it's impossible to be sure but by using the sub-regional figures for the GCC, etc, it should be possible to get a bit closer to reality.

As a cautionary note, these sub-regional figures are probably inaccurate too if they have not been adjusted for population differences, though the distorting effect may be less than it is with the overall regional figures.

To re-evaluate the survey's question on religion, let's start by looking at population. Helpfully, the survey itself gives some population figures for the relevant countries on page 45. Added up, these are as follows:

GCC countries: 52.65 million
Levant/Yemen: 80.56 million
North Africa: 186.25 million

However, the GCC figure has to be adjusted because it includes large numbers of non-nationals (who were excluded from the survey). According to the Gulf Labour Markets and Migration website, nationals account for 51.9% of the overall GCC population which brings the figure down to 27.33 million.

That gives a total population of 294.14 million, of which the GCC accounts for 9.29%, the Levant/Yemen 27.39%, and North Africa 63.32%.

Thus, in a geographically balanced group of, say, 10,000 people (the actual number doesn't matter but 10,000 is convenient to work with) there should be 929 people from the GCC countries, 2,739 from the Levant/Yemen, and 6,332 from North Africa. 

Based on these figures, we can now adjust the Arab Youth Survey's figures for those agreeing with the statement about religion:

GCC countries: 61% of 929 = 567 who agreed 
Levant/Yemen: 44% of 2,739 = 1,205 who agreed 
North Africa: 47% of 6,332 = 2,967 who agreed 

That would give a total of 4,748 (out of 10,000) who agreed, or 47.48% – somewhat short of the majority reported in the Arab Youth Survey.