Pre-poll surveys: Methodology is the culprit
By B. K. Kuthiala

IN social science research parlance opinion poll is a kind of random sample survey conducted to collect the opinion and attitude of people with regard to a given subject. The random sample surveys are based on a proven and well-established theory of probability and random distribution. The theory on the basis of which random sample surveys are conducted envisages that a small fraction of the randomly selected sample will represent the total population in terms of distribution of opinions.

The key term is the random selection which according to the probability theory puts a condition that every unit of the population under study must get an equal chance of getting selected. Any element of subjectivity or bias in the selection process causes deviation from the actuals. Even a minor element of subjectivity will lead to totally non-representative findings. Therefore, if in a random sample survey the organisers for the sake of convenience give more weightage to the respondents in the urban area or in villages near urban areas, the study does not remain representative and cannot be relied upon. There is no scope for convenience for the sake of economy or time in such studies.

If all elements of the population under study get equal chance of being selected and if the questions asked are proper then there is no doubt that the findings will represent the total population with a reasonable degree of deviance. The sample need not be very large; although larger the sample, greater would be the validity of the findings.

During the last 50 years, the use of qualitative methods of investigation borrowed from social sciences have been increasingly used in journalism also. In 1935, Fortune magazine published what is considered to be the first scientific poll conducted by a news organisation. Today most of the newspapers and magazines dealing with current affairs and almost all radio and television channels are using sample surveys in one form or another.

Media academician Meyer in 1973, for the first time, formally introduced sample surveys as a tool of journalism. He wrote a book titled Application of Social Sciences Methods to the Practice of Journalism. On the recommendation of a professor of journalism, Everet Darnis, Meyer changed his mind and called these methods "precision journalism".

Through the years the use of social science techniques for the collection of news has been given a number of names, including new journalism, computer assisted journalism, scientific journalism and qualitative journalism. But there seems to be a general agreement between media academicians and media practitioners that "precision journalism" is a term that describes more adequately the process of random sample surveys for news collection.

In India in recent years, "precision journalism" has also been called "Non-VIP Reporting". The assumption in this nomenclature is that most of the news originates from the people who have a status higher than that of a common man. Precision journalism basically deals with how masses perceive various issues and how they either plan to behave or actually behave.

With a new name and a distinct identity, precision journalism got established in 1970 rather strongly. According to a study by Rippey in the USA only one-third of the daily newspapers were not doing opinion polls. In India also daily newspapers either sponsor their own opinion polls or the opinion polls conducted by professional agencies are reported by them. Some of the newspapers and current affairs magazines regularly conduct surveys and report them prominently.

The problem with most of the pre-poll surveys conducted during previous elections had been that in order to expedite the process of data collection, the sampling process has been so compromised that there was an inbuilt bias in favour of electorate living in metropolitan cities, urban areas and rural areas near urban centres. This type of bias definitely leads to distortions and gives a bad name to the whole process of opinion polls. Similarly for exit polls if the selection of the booths from which the data is to be collected is done in such a manner that all booths get an equal chance of being selected irrespective of the convenience of the investigators the exit polls are bound to reflect the trends of actual polling.

Another important aspect of opinion polls is the kind of questions that the investigators ask. For a opinion poll at the national level, the questions asked have to be framed very carefully and tested a number of times before being actually used for the survey. In some of the surveys conducted recently, the questions asked were of the type "have you stopped beating your wife?" The possible responses provided are ‘yes’ or ‘no’. In both cases the information received is distorted. One of the opinion poll agency assumed that people will not vote for defectors and without asking whether they would vote for defectors or not the question asked was, "Why will you not vote for defectors?"

Though the methodology of conducting opinion polls is based on scientific principles, conversion of this information into the number of seats is a difficult and tedious tasks. So far there has been no authentic methodology to arrive at these magic figures. The reason is that a large number of factors are operational and no single methodology can be applied. There would be differential influence of various factors in each constituency, region and also at different points of time. The tried formula of cubes has proved to be more wrong than valid. But when the predictions about the number of seats proves wrong, the blame is put on the opinion polls. If the opinion polls confine themselves to reporting only what they find, the distribution of pattern of votes, they would perhaps retain their credibility. The psephologists have to work harder to develop a foolproof methodology.