Reading post-poll surveys

Praveen S
Friday, June 25, 2021

The Left Democratic Front (LDF) won a decisive mandate, clinching 99 seats out of 140, in the recently concluded Kerala elections. The election result was interesting as it ended the four-decade-old trend of alternating governments between LDF and the United Democratic Front (UDF). There are many theories on the reason behind LDF's success. There are some interesting empirical shreds of evidence too examining the significant points regarding voter perception on which LDF scored above UDF. For example, the Lokniti-CSDS post-poll survey.
In the Lokniti-CSDS post-poll survey, when asked whether the LDF should get another term, 51% responded that it should, 27% answered that it should not, and 22% did not respond. Many analysts take the post-poll data as clearly indicating that the second term for LDF was a byproduct of public perception that the government had done a reasonably good job, supported by a higher assessment of the work done by the LDF government on different sectors. Respondents rated that almost all sectors of governance as 'improved' compared to 'deteriorated'; for example, medical facilities at government hospitals, condition of government schools, condition of roads, supply of electricity, supply of drinking water, public transportation, and law and order.
A natural question that comes to our minds when reading such surveys is should we believe them? Yes, you should consider them if they are coming from reputed organizations; No, you should not consider them since they are all biased in a way. Since reputed organizations truly care to protect their reputations, the process involved in collecting such surveys will be transparent and free from technical errors. They would have procedures for randomly contacting samples from different populations and standard ways of contacting and recording data. Since Lokniti-CSDS is reputed in this field, we can believe in the data collected and the results we obtain by analyzing the data. However, we added a ‘no’ because we should also be able to understand the human biases that may prejudice all these responses. In the rest of the article, we explain how the bias might be acting in this case.
An objective measure of performance on sectors on which perceptions have been assessed in the survey may showcase a different picture. Why do I think the objective measure would be different? Because all organizations/governments are limited by their resources in investing in these welfare activities. They are all bound by a resource allocation framework through which they judiciously allocate their limited resources on several accounts such as health, roads, food security, IT, communications, etc. When the investments in one of the sectors go up, one has to limit their expenditure in other sectors. That means because of the resource limitations, it is impossible for anyone to overinvest in all the sectors. Therefore, if you are comparing different organizations, you should see that some governments would have overinvested in some of the sectors than others. This means that unless LDF had a tremendous resource advantage over UDF, it is impossible for them to overinvest in all sectors. And most likely, the developments in each sector will be proportional to the resource investments in that sector unless the government will have the efficiency to collect a higher return on such investments. Since the state government staff who implement projects using these resources are the same, it is unlikely to be the case.
Then, how should I read the survey results? Why does the survey show all positive for LDF? This is where human biases come into the picture.  As per Nobel laureate Kahneman's works on human biases, when we are posed with a difficult question as to evaluate a government's work on health, the question is so complex to a respondent, and they need so much information to arrive at a valid answer to the question. To avoid the complexity involved in answering such a complex question, our minds replace this difficult question with an easy one. The most likely question that would replace this question would be how much do we like the government. If we like the government more, we would respond favorably to the question.
I hope I made my perspective clear on how we should interpret such survey results. Yes, most people like the LDF government now over the UDF government, not necessarily that the LDF government has overperformed in their work on all these sectors.