Thursday 5 July 2012

Understanding the PEW Survey Ratings

survey was recently conducted by the Pew Research Center which conducts public opinion polls around the world on a broad array of subjects. A range of issues was covered in the survey including approval ratings for key politicians and institutional heads in Pakistan. The approval ratings of these personalities are reproduced below:




The survey has sparked a debate on what these results mean for the upcoming elections in Pakistan. Politicians with improved numbers are quoting the results as conclusive proof that 'change' is here and that they are set to "sweep" the elections by polling 70% of the popular vote.  To understand what, if anything, can be concluded from these results we must first understand what the numbers really mean.
First and foremost we must reject the myth that the survey results show the percentage of votes that each personality will poll in an election. To confirm this, simply note that the numbers in each column add up to much more than 100% So what are these numbers then? To understand this, it is important to reconstruct the question that was asked of the survey respondents
Q) Which of the following personalities do you "view favorably" or have a positive image of? You may choose as many as you like.

__ Ashfaq Pervez Kiani
__ Asif Zardari
__ Iftikhar Muhammad Chaudhry
__ Imran Khan
__ Nawaz Sharif
__ Pervez Musharraf
__ Yousuf Raza Gillani
The people who participated in the survey were allowed to choose more than one person in the answer to the above question. So if 100 people were surveyed, the numbers in the table above show how many positive ratings each of the personalities received from those 100 respondents. Now that we understand the meaning of these numbers we can proceed to analyse their significance. 

70% of the respondents view Mr. Imran Khan, leader of Pakistan Tehreek-e-Insaaf (PTI), favorably where as 62% hold a favorable view of Mr. Nawaz Sharif, leader of Pakistan Muslim League (PMLN). Does this suggest that PTI will surely poll more votes than PMLN? To anwer this, we need to review the politics of the two parties during the last twelve months. PTI has built its campaign on the notion that all politician's in Pakistan are corrupt and incompetent (except those who chose to join it) and that PMLN and the ruling Pakistan Peoples Party (PPP) are "equally" responsible for the current state of the country. PTI leaders make anti-PMLN comments in evey public rally and in every TV appearance . It can safely be said that PTI has directed more criticism at PMLN than it has done towards the ruling PPP. PTI has also ruled out any alliance with PMLN before or after the elections. As a result of its fierce assault on PMLN, the people who may vote for PTI in the upcoming elections most likely selected only Imran Khan and not Nawaz Sharif in the answer to the question presented above. Looking at the PMLN campaign, we see that their leaders have mostly ignored PTI attacks and restricted themselves to answering allegations only when it became necessary. The center of PMLN's campaign is the corruption of the ruling PPP and not PTI. They have not explicitly ruled out a pre or post election alliance with PTI. As a result, the PMLN voter still carries a favorable image of Imran Khan originating from the 1992 cricket world cup win and the establishment of a cancer hospital in Lahore. It is therefore logical to assume that a number of respondents who are prospective PMLN voters, may have selected both Imran Khan and Nawaz Sharif in the answer to the survey question presented above. 
Simple mathematics* with the numbers in the table tells us that the number of people who selected both Imran Khan and Nawaz Sharif on this survey ranges between 32% and  62%. For the reasons presented above, it is much more likely that people who selected both Imran Khan and Nawaz Sharif on the survey are prospective PMLN voters. It can also be safely concluded that at least half of those who see Imran Khan favorably do not believe his claims about Nawaz Sharif. 
To summarize, this survey was not designed to be a predictor of election results. It was meant to gauge the approval rating of individuals. Other non-political personalities such as social worker Abdus Sattar Edhi could have been listed under the same question and plausibly secured ratings exceeding 70% 

* 62 (NS rating) + 72 (IK rating) - 100 = 32

No comments:

Post a Comment