The ticking data bomb

In a recent op-ed in another newspaper, the writer highlighted the importance of solving our country’s ballooning population growth, since it is pretty much uncontrolled and likely to cause (is causing?) severe issues if it continues to go unaddressed. The writer used various data points to reinforce his argument. I agree with his core thesis that the population problem needs to be dealt with very seriously and immediately, before we hit an iceberg.

So how do we go about fixing important issues? By analysing them. To analyse, we need to have access to data. Is our data correct and reliable? The answer, sadly, is no. Can our analysis be well-formed and likely to solve our issues in the most prudent way possible when the underlying data is unreliable? The answer, again, is no. What to do then? I believe the answer lies in strengthening our capacity to collect accurate data. The writer of the said opinion piece used relevant health-care indicators to reinforce his argument, but the quality of data is questionable.

There are serious discrepancies in our healthcare data. Ironically, I found that one government publication reported different numbers when compared to another publication for the same year and the same health indicator. Let me point out just a few of these inconsistencies

According to the Pakistan Demographic and Health Survey (PDHS) 2006–2007, the number of ‘Delivered by skilled birth attendants’ cases in Punjab is 38 percent, whereas according to the Pakistan Social and Living Standard Measurement (PSLM) it is 32 percent for the same year and the same province. Which one do I believe? How can you spend millions to fix issues when you’re confused about which data points to consider?

Another one: according to the Multiple Indicator Cluster Survey (MICS) 2008, Khyber Pakhtunkhwa had 38.5 percent of deliveries at health facilities, whereas according to the PSLM it comes in at about 27 percent for the same year. Again, the delta between the two indicators is huge and leaves one confused.

There is also a discrepancy for an indicator called ‘postnatal consultation’ between the PSLM and MICS for 2008-2009 for KP. According to the MICS, the percentage of births followed by postnatal consultation in KP is 13 percent, whereas the PSLM reports that the figure is 23 percent.

Given the discrepancies in the above indicators one can infer that indicators like ‘fertility rate’ and ‘contraceptive prevalence rate’ could also be possibly prone to errors.

The unreliable data was checked just for provinces; it can be possible that district-wide data is as incorrect as provincial data. If that’s the case, it will be practically impossible to make any timely progress on any important issue. A deserving district may get an unfair share due to the inaccuracy of available data. Likewise, an undeserving district may get unfair attention. Although, here I’ve taken the liberty of assuming that our policy formation is based on evidence and data. But even if, by good happenstance, our leaders rely on data, this reliance will unfortunately result in unfair distribution of resources.

Given these inconsistencies, which publication do we trust? Can we trust such institutions to carry on other data collection exercises? The PSLM is produced by the same institution that carried out our census. Given such discrepancies, can we believe the census numbers?

Can we conduct any meaningful research given these discrepancies? I don’t think any meaningful analysis can be conducted without having access to correct data. Without meaningful analysis, no insights can be gained, and without insights I’m not certain how impactful policies can be formed. And without impactful policies, we cannot help our people out of misery.

The government needs to really take data collection seriously. For one, it will be hard to believe the tall claims they make without backing them up reliable data. A simple glance at data made available by a bureau of statistics of a Western country shows that they are light years ahead of us in both the accuracy and volume of available data.

To make any real progress, we need to strengthen our data collection mechanisms. No substantial analysis can be formed on the basis of unreliable data.

Email: wyounas@lumsalumni.pk

Twitter: @wyounas