In India, we do get elections almost every 6 months, in some State or other or elections for the entire Nation. Mostly, the elections are held at multiple phases, giving enough fodder for the 24x7 media who are literally looking for stuff. Whenever there are elections, media will be having their nice days as they have to come up with Opinion Polls results, Exit polls results and of course, the actual poll results.
How precise the opinion & exit poll results would be? What goes in these polls? Are they really scientifically designed or just a means to force the voters who are yet to decide and create a swing factor? Psephology is not a new thing to India but it became popular in the post 90s.
The term psephology is derived from Greek. Psephos means pebbles, thus the idea of counting. One of the finest political philosophers and the author of Leviathan, Thomas Hobbes, had enunciated: "All politics is geometry." With increasing focus on democracies and elections the subject gained considerable respectability and became a new research field all over the world. The 1967 and '71 polls in India were studied by the University of Michigan in collaboration with the Centre for the Study of Developing Societies in Delhi with large sample sizes and exhaustive questions. In the media, (whether you like it or not) it was INDIA TODAY (not the IT of Rajdeep Sardesai but of Nalini Singh), which pioneered the concept of opinion polls as early as 1978.
Today, we have multiple agencies conducting such surveys and polls and make good money. Each agency who conducts opinion polls or exit polls has their own method, and they are unlikely to reveal their secret recipe. Those who are aware of Statistics might know about these factors - Population, Sampling size, Sampling methods, Demographic distribution, Collation etc.,
Given the broader demography, it is quite clear that unless one prepares a systematic survey questionnaire, conducting and publishing a reliable Opinion/ Exit poll is not so easy. One cannot simply do this in a room with PCs, phones and online surveys. The survey questionnaire should have both Closed Questions and Open-ended questions.
In India, the demographic classification drills down as under:
- Male-Female
- Age factor
- New voters - Old timers
- Literates - Illiterates
- BPL - Tax payers
- Rural - Urban
- Minorities - Hindus
- Castes - OBCs - Dalits
- Party members - non-members
- those who follow current affairs - do not follow
Conducting Opinion or Exit poll on Indian elections in a country like India is complicated and difficult due to a populace comprising of myriad caste community groupings combined with multiple political parties across the political spectrum. An election survey can estimate the vote shares correctly for the political parties, but predictions can still go wrong as due to intrinsic flaws in forecasting models or due to pollsters tweaking projections based on statistical wisdom or rebalancing by media to suit their political preferences.
Sampling is another factor in this. When we conduct research about a group of people, it’s rarely possible to collect data from every person in that group. Instead, we select samples. The sample is the group of individuals who will actually participate in the research. The size of the sample is pivotal in all these researches/ studies. To draw valid conclusions from our results, we have to carefully decide how we select a sample that represents the entire group. There are two types of sampling methods:
- Probability Sampling involves random selection, allowing you to make statistical inferences about the whole group.
- Non-Probability Sampling involves non-random selection based on convenience or other criteria, allowing you to easily collect initial data.
At the outset, we need to understand the difference between a population and sample and identify the target population of our research.
- The population is the entire group that we want to draw conclusions about
- The sample is the specific group of individuals that you will collect data from
The population can be defined in terms of geographical location, age, income, and many other characteristics. It can be very broad or quite narrow: maybe we want to make inferences about the whole adult population of our country; maybe our research focuses on customers of a certain company, patients with a specific health condition, or students in a single school.
It is important to carefully define our target population according to the purpose and practicalities of our project. If the population is very large, demographically mixed, and geographically dispersed, it might be difficult to gain access to a representative sample.
The sampling frame is the actual list of individuals that the sample will be drawn from. Ideally, it should include the entire target population (and nobody who is not part of that population).
1. Simple random sampling: In a simple random sampling, every member of the population has an equal chance of being selected. Your sampling frame should include the whole population. To conduct this type of sampling, we can use tools like random number generators or other techniques that are based entirely on chance.
2. Systematic sampling is similar to simple random sampling, but it is usually slightly easier to conduct. Every member of the population is listed with a number, but instead of randomly generating numbers, individuals are chosen at regular intervals.
If we use this technique, it is important to make sure that there is no hidden pattern in the list as this might skew the sample. For example, if the HR database groups employees by team, and team members are listed in order of seniority, there is a risk that our interval might skip over people in junior roles, resulting in a sample that is skewed towards senior employees.
3. Stratified sampling involves dividing the population into sub-populations that may differ in important ways. It allows we draw more precise conclusions by ensuring that every subgroup is properly represented in the sample. To use this sampling method, we divide the population into subgroups (called strata) based on the relevant characteristic (e.g. gender, age range, income bracket, job role).
Based on the overall proportions of the population, we calculate how many people should be sampled from each subgroup. Then we use random or systematic sampling to select a sample from each subgroup.
4. Cluster sampling also involves dividing the population into subgroups, but each subgroup should have similar characteristics to the whole sample. Instead of sampling individuals from each subgroup, we randomly select entire subgroups. If it is practically possible, we might include every individual from each sampled cluster. If the clusters themselves are large, we can also sample individuals from within each cluster using one of the techniques above. This method is good for dealing with large and dispersed populations, but there is more risk of error in the sample, as there could be substantial differences between clusters. It’s difficult to guarantee that the sampled clusters are really representative of the whole population.
To conclude, a stocktaking of opinion polling in the last forty years reveals that 75% of the 833 (386 pre poll and 447 exit) election surveys correctly predicted the winning political party (ies) in India. The accuracy rates of exit polls (84%) was 13 points higher than opinion polls (71%) conducted during the elections. The success rates (aggregate of both exit and opinion polls) of polls differ quite significantly for the national and state elections. The correct prediction for Lok Sabha elections is 97% (2004 Lok Sabha polls was an outlier) while the success rate is 75% for state Assembly polls. The strike rate of such polls may not match the global standards of the polling industry, but they are not as off the mark as public perception imagine, hence it is perilous to dismiss opinion polls. The election forecast record of Indian polling firms may not match the world benchmark, but a post mortem of election polling reveals that precision in terms of vote share accuracy is at par with its US and UK contemporaries. The mathematical predictions models based on opinion poll vote share is fallible as well as fragile, but polling agencies in India guided by blue-sky thinking are trying their best to improve the craft of political forecasting and seat predictions.
