CRITIQUE #4: At least use up-to-date data!!

"We are made wise not by the recollection of our past, but by the responsibility of our future." - George Bernard Shaw

My attention moved to the IMD data, since something struck me about the ranking MCC used in appendix 7 of their report. Northenden in particular stood out as sitting pretty in the poverty stakes, sandwiched nicely between the affluent and desirable areas of Chorlton and Didsbury and ranking lower in deprivation than the city centre.

When checking the details of the library catchment areas on pages 132 for Chorlton, 138 for Didsbury and 162 for Northenden in appendix 12 of the library consultation report. I noted that in Didsbury, 1 out of the 22 Lower Super Output Areas (LSOA) were in the top 20% most deprived in the country. By comparison, Chorlton had 9 out of 26 in the top 20% most deprived and Northenden had 1 out of 4. For a second, given the IMD rankings MCC used, I even pondered whether I should buy property in Northenden if it is made up of Didsbury like affluence :-)

I just wanted to check if the details were accurate before I put down a deposit. So I got hold of the ONS statistics for IMD for 2010.

Technical Note
The Index of Multiple Deprivation is a single number which rolls up two main economic variables. Rate of income benefits and the rate of employment Benefits. It is a linear regression model and as such, includes a series of beta coefficients and a residual value which are all looked up from a table. For those who are interested, the ONS method can be found here.

I picked off the IMD values for all areas from the 2010 data and to my surprise they didn't match! I know, I know, I should know better by now given that it is MCC.

fig 1 - MCC to ONS Average IMD, together with the ranking difference (click to enlarge)

The table above shows the ranking using the IMD values from the 2010 data on the left, with the MCC ranking in the middle and the difference in ranking on the rightmost column. The highlighted libraries show the ones that have moved. You will note that the MCC correlation with this data is good, but not perfect. given this is a standard dataset used everywhere, I did wonder how this could be. More on that later.

In the real IMD figures from 2010 (reported in 2011 and updated 2013 - There hasn't been any more recent datasets published) Northenden is the 10th most deprived area of Manchester, just above Burnage.

Negative numbers mean that the library catchment areas has 'moved' (or been moved) down the rankings to allow it to be considered a more affluent area and thus have lower need. Look at Northenden, it has been moved 11 placed further down the rankings than the ONS statistics show (data source: Department for Communities and Local Government)! Burnage is 3 places lower than it should be, Fallowfield is 7 places higher than it should be. Moss Side is 4 places higher than it should be and City Library is 3 places higher than it should be.

OK, so remembering the correlation matrix from my first critique of this report, I was looking to see what this would do to the correlations around the average IMD scores. There is a small clump which strongly relates IMD to the children and young people statistic. Sure enough, when I built the correlation matrix for this newly ranked data, I got the following. I've put the original correlation matrix underneath for comparison:

fig 2 - 2010 IMD correlation matrix (click to enlarge)

fig 3 - Original correlation matrix (click to enlarge)

You can see from this that the new matrix at the top shows that IMD now doesn't have any moderate or strong correlation with the percentage of children and young people in the catchment area at all.

So what does this tell us?

I don't want to speculate too much on what has happened here, since in either case, the decision did not consider IMD at all. However, a number of things could have caused this:

• To appeal to the cynical amongst us, I could say that MCC could have attempted to 'fudge' the figures presented in Appendix 7 for areas like Moss Side, Fallowfield, Burnage and Northenden to make them seem more or less affluent than they otherwise are, possibly to avoid difficult questions around deprivation and equality impact. Giving Moss Side a Higher IMD rating could have been to attempt to decrease the overall sum of the rankings in the combined score of appendix 7, effectively increasing the chances of that library remaining open. 
• Similarly, Burnage and Northenden being regarded as more affluent than they are would mean they have less of a need for a library in that ranking.
• Alternatively, MCC has used very old data or data not otherwise available. Again, not sure why the latter would be the case, since this is a standard dataset used throughout the UK. I may dig out the 2008 published data (which will be for 2005) and check them against the MCC report.
• Poor data quality/analysis. 

Note, using the correct IMDs, the correlation matrix is now completely based on library catchment area. No other variable was really considered (remembering the total usage stats are numeric and so are dependent on the size of the catchment area). It could have been that these figures were modified to justify decisions already made or Northenden has indeed won the proverbial lottery. I am sure the place is lovely, but given economic benefit, the data in the consultation report and where it sits relative to the city centre, I suspect I might save my deposit for now.

EA