Exploring the secret science behind opinion polls

Express July 09, 2010

ISLAMABAD: Newspapers around the world regularly publish results of opinion polls on a wide range of matters. These polls claim to accurately represent the views of millions of people, although they are often based on interviews with no more than 2,000 respondents.

This leads some readers to wonder: how can a sample of just 2,000 people accurately reflect the views of 170 million?

It was this major question which was addressed in a five-day course on research methodology organised by Gallup Pakistan and hosted by Iqra University.

The course ended on Friday, with participants including professors of public health and business management, students of the social sciences as well as representatives of NGOs, who had gathered from all over the country.

On the first day, Dr Gilani of Gallup Pakistan startled his already curious audience
when he said: “The most counter-intuitive secret of the polling business is that if you’re surveying the
whole country, you need not interview more than two thousand people; a sample of two thousand is about as good as a sample of two hundred thousand.

On the other hand, if you’re surveying the incomparable smaller population of just one district of the country, you will still need a sample of two thousand.

In reality, the size of a survey sample has no direct numeric relationship with the size of the population being surveyed.”

The trick in scientific sampling, he explained, lies not in how many people you interview but in whom you interview, and how you interview them.

According to the speaker, law of probability infers that if applied uniformly throughout the procedure of data collection, from identifying the sample to taking the interviews, the final result will corroborate to the actual representation of the people.

Participants were enlightened on this “complex art” through the first part of this workshop.

The rest of the course was essentially about teaching the participants the scientific method of choosing a representative sample. Once such a scientific sample is chosen, the researcher can foretell the opinion of the whole nation, without having to go to everyone.

That said, the art of appropriate sampling was only one component of this course.

Other components included questionnaire design, conducting fieldwork and research model-building. In all modules, participants received not only instruction but also hands-on professional training.

Perhaps the highlight of the course was the fourth day when all participants
went to the village of Tarlai, adjacent to National Park Area, Islamabad and got an opportunity to hone their freshly-acquired research skills - an exercise which might have been exciting for the participants, but possibly rather irritating for the villagers.

On the final day, all the groups made presentations on their selected areas of interest.

The workshop was also addressed by guest speakers, Bilal Khan, a research consultant and Dr Muhammad Islam, Advisor to the Dean, Iqra University, both of
whom shared their valuable expertise with the participants.

The Dean of Iqra University, Dr Jamil Ahmed, concluded the course and awarded certificates to a rather satisfied bunch of participants.

Published in The Express Tribune, July 10th, 2010.


Replying to X

Comments are moderated and generally will be posted if they are on-topic and not abusive.

For more information, please see our Comments FAQ


Most Read