There will be much discussion about Prothom Alo’s recent survey—and let it be so. Analysis is always enlightening. The current piece, however, is not an analysis of this particular survey. It is election season, and the word 'survey' is something we will hear most frequently. This article attempts to shed light on all kinds of pre-election surveys from every quarter. Surveys that have not yet been conducted, but are in preparation, will also not be excluded from this discussion. Until now, we have never presented a truly candid conversation about surveys. Why we haven’t is a matter of sheer astonishment. The truth is, while elections are serious political events, they are equally a form of entertainment. Surveys are the glittering costumes of that entertainment.
Imagine a staged historical fantasy play called Nawab Siraj-ud-Doula. No matter how compelling the story, if Siraj-ud-Doula is dressed in a simple lungi and panjabi, even the best story falls flat. But if he is adorned in dazzling royal silks and sherwanis encrusted with diamonds, emeralds, and rubies, with a crown on his head and a sword at his waist, the story, though somewhat weakened, becomes a thoroughly enjoyable spectacle. If the election is the Siraj-ud-Doula character, the survey is its glittering costume. Surveys make elections more enjoyable.
Sociologist Erving Goffman believes that society itself is a theater. His “social dramaturgy” theory explains society in terms of backstage (behind the scenes, out of view) and frontstage (onstage, visible). Each institution, essential activity, and form of entertainment fits within this framework. Pre-election surveys are best treated as entertainment rather than guidance. The prevailing thought, “The results are almost certain; the survey shows it,” and the political discussions on television and social media should also be regarded as entertainment.
In many countries, pre-election surveys have repeatedly proven inaccurate. Yet people like surveys, just as pollsters favour counting. The 1992 UK election is still cited as a classic example of the “shy Tory" effect. Almost all surveys suggested the Conservatives were losing, yet they won by a large margin. Surveys were again proven wrong in 2015. All of them predicted a “neck-and-neck race,” but when ballots were counted, the Conservatives won comfortably.
A dramatic instance occurred during the 2016 Brexit vote. Most surveys showed support leading for “Remain,” staying in the European Union. However, the final referendum resulted in a “leave” victory. Experts later conducted meticulous analyses. They concluded that telephone surveys yielded a uniform type of response. Respondents, eager to hang up quickly, were not serious about stating their true opinion. Online surveys, on the other hand, often matched the actual outcome, as respondents took their time to answer thoughtfully. Telephone surveys are also less trusted; respondents fear that their responses may be recorded or that they may be identified.
In the United States in 2016, national surveys correctly reflected Hillary Clinton’s popularity, but key state-level surveys showed Trump far behind. Similar survey failures were observed in 2020 and 2024. Analysts on talk shows explained that many voters, particularly Republicans, participate less in surveys. Many are older, hold traditional views, and treat their voting intentions as strictly private. They even doubt the confidentiality of “anonymous” surveys, reducing their reliability.
In India’s 2004 elections, major surveys predicted that the BJP-led alliance would return to power. In the end, the Congress-led alliance won the election.
Surveyors rarely get the chance to be proven wrong. Most polling organisations conduct surveys under the leadership of statistics experts and are fully aware of the limitations and weaknesses of their methods. So, is it possible that neither the survey results nor the election outcomes are actually wrong?
It is very possible. “Voting behaviour” or the behaviour of voters is one of the most complex and diverse phenomena. In Bangladesh, more research is needed on voting behaviour rather than just opinion polls. As far as I know, there is a notable study by Professor Yahya Akhter, the current Vice-Chancellor of Chittagong University. I have heard of a few other significant studies, but I could not locate them.
In a country like Bangladesh, a major issue is the refusal to accept election results. No party leader has yet pledged that they will accept the outcome, even if the upcoming election is free, fair, and accurate. One problem with surveys is that they can create a mindset among political activists and supporters that “we are going to win.”
Many believe that such surveys can encourage voters to favour the party predicted to win. If that were true, the repeated survey failures in so many countries would not exist. Today, we hear much more about “swing voters” than before. This group is not blindly loyal to any party. Around the world, partisan blindness is declining. The number of swing voters is rapidly increasing, and their influence is virtually limitless. It is difficult to predict when, why, or how their opinions, judgments, and choices will shift. In Bangladesh, this group has also grown significantly. In the upcoming elections, it will be the swing voters, not surveys, who will determine the outcome.
For decades, sociologists have conducted comparative research to understand the diversity of voting behaviour worldwide. Political scientists Seymour Martin Lipset and Stein Rokkan, in their seminal work Party Systems and Alignments (1967), demonstrated that voter behaviour in European countries is not uniform. Historical divisions—workers vs. owners, church vs. state, centre vs. peripheral regions—create distinct “cleavages.” These social fault lines shape party-based voter identity differently in each country, resulting in complex and varied voting patterns.
This idea was further developed by Karen Brooks and two co-researchers in their study Cleavage-Based Voting Behaviour in Cross-National Perspective. They showed that voter behaviour cannot be explained by a single universal rule. Cleavages—religion, race, gender, regionality, inequality, class, social mobility, etc.—strongly influence voter behaviour.
American political scientists Russell Dalton and Christopher Anderson conducted a project called Comparative Study of Electoral Systems. Using this data, they edited the book Citizens, Context, and Choice. Being able to account for cleavages is referred to as the “context.” Political scientists Pippa Norris and Ronald Inglehart have called it the “cultural backlash.”
Cleavages are usually multiple, but the largest cleavage determines voter behaviour. In Bangladesh, the largest cleavage—the context and cultural backlash—can be summed up as “trustlessness,” meaning the lack of trust and confidence in political parties. If the 2026 election is free and fair, the winners will be those who can establish trust and credibility among Bangladesh’s swing voters.
#Helal Mohiuddin is Professor of Sociology at Mayville State University, North Dakota, USA
*The opinions expressed are the author’s own.