no image

Variations in Questions and Scores

Jasim Sheikh
• Islamic Relief
• Pakistan
• 10/19/17
• 1 Comment
 

Hey All,

Has anyone tried to alter the questions and their responses/scores to try to make the tool more meaningful?

I'm from the Islamic Microfinance Unit of UK-based Islamic Relief Worldwide. We have developed an Android device form that takes the responses to our central MIS where all the scores are generated. We've piloted in Pakistan and we're in an early stage of the usage but we're mainly assessing how deep our outreach is. 

However, we have noticed that some responses and their scores are a bit broad. For ex.:

10.   Does the household own a motorcycle, scooter, car, or other vehicle?
A.  No    0
B.  Yes    12

Now having a bicycle would give the person 12. Having an old worn-out motorcycle will also give them a 12. And having a car will also give them a 12. So it's clubbing individuals from highly varying income backgrounds into one group. Same is the case with some other questions.

Has anyone figured out how this broad generalization can be addressed?

Regards,

Jasim

1 Comment
Hello Jasim, Thanks for your question. The PPI scorecard is comprised of variables which are statistically found to be most strongly correlated with poverty. In the case of Pakistan, based on the 2005/06 Pakistan Social and Living Standards Measurement (PSLM) Survey data using which this scorecard has been created, the lack of ownership of a vehicle (any kind of vehicle, which is why the question is framed the way it is) is strongly correlated with poverty. FYI - we are in the process of making a change to the PPI's methodology which we believe will further enhance its ability to predict poverty. Response options for the same indicators will be allotted different points depending upon the poverty line that you plan to use. This is because the level of association of an indicator to poverty may not be the same across all poverty lines. What our change will mean in practice is that the "weight" of an indicator to the final PPI score will vary depending on the poverty line used. Our paper on the PPI's new methodology will be released later this year. We will also provide users with detailed advice on what this means in terms of changes to the way they are currently administering the PPI. Please stay tuned. One additional point related to the current question. Since PPI indicators are drawn from the National Household Income and expenditure surveys, we also need to ensure that questions are asked and answers are interpreted in the same way as they were during that survey. In the case of Pakistan, there is no guidance in the official Manual of Instructions of the 2005/6 PSLM Survey on what can be included under the broad term "vehicles" and how to treat broken/run-down vehicles. Please see appendix A in this document for more on this - http://www.simplepovertyscorecard.com/PAK_2005_ENG.pdf And finally, each country's PPI, the questions, their responses, allocated points and poverty likelihoods are derived using the country's National Income or Expenditure Survey. The PPI is calibrated using indicators that are found through statistical means to be the most powerful indicators of poverty within a country. Changing any component of the PPI will undermine its predictive power and result in poor data, and is therefore not advised. I hope this is helpful. Best regards, Sharada