Research News Live

MRX and polling under scrutiny: failure vs accuracy

Polling failures are a recent discussion around the world and a big concern in many countries. Nobody questions the importance of polls for democracies and how useful they are when building public policies. Nevertheless, the lack of accuracy in some elections questions them and affects the prestige of our profession. But this is not true, and we haven´t been good enough in defending polling and our industry. Some researchers may give less importance to polling because they only represent 6% of the total spending on market research according to the ESOMAR GMR 2018, but the exposure they have in the media makes them relevant and a showcase of our industry, so they do contribute to the image of market research in general.


As Jon Puleston said in his presentation in the ESOMAR Congress in Amsterdam, polling is not getting worse, it has new challenges, but throughout the years, polls kept the same accuracy rate. Puleston analyzed polling results in 40 countries related to 430 events/elections. It´s quite interesting to mention that 85% of the polls analyzed were correct in predicting the outcome. UK election (2015), UK referendum (Brexit), US elections (2017) are considered failures but that´s not accurate, different techniques are used to measure effectiveness and the error is in the eye of the beholder. To go deeper on this, I recommend to read his paper “Are we getting Worse at Political Polling?”.

To this global context, we need to add new successes. An example could be the latest Peruvian Elections for City Mayor (October 7, 2018). The graph shows how accurate results were in comparison to official results. It´s true that the closer to the election is, the most accurate the poll will be. One week before, undecided were 46% of voters, but polls were able to predict the trend and announce the winner.


In countries like Peru, polling is a challenge. Voting is compulsory, 24% of voters live in rural areas (includes towns located in the jungle and highlands and over 12 hours travelling by road or river to reach them), so to reach them, polls are done face to face. There is a blackout period of 7 days before the election (the good news is that there is a law project to reduce it) and also publishing polls related to intention vote are regulated (database and a special report should be delivered to the authority in charge).

Despite those challenges, results were accurate, the media and the market in general recognized that. There is no mystery for this success. Like other research projects, the team and the design of this project were crucial. The team had a mix of experienced researchers and young millennials, engaged with the challenge and willing to give their best. The project design followed a couple of principles. The first one is that we needed accuracy, so the sample was very important because it should represent the country, better a smaller sample well designed than a bigger one difficult to obtain. Second, the timing, events in the campaign could change the intention vote, so technology helped as to have results in real time.

Therefore, it´s time to celebrate success, to make some noise and show how accurate polls are in countries like Peru. I believe this will contribute to trust and confidence on market research.


Urpi Torrado





The Australian Market and Social Research Society is linked globally to 45 associations through its partnership with the Global Research Business Network (GRBN) and the Asia Pacific Research Committee (APRC). Click here to read about the AMSRS global network. This article is originally sourced from GRBN website.

About The Research Society 1084 Articles
The Research Society is the peak body for research, insights and analytics professionals in Australia. It has a diverse membership of individuals at all levels of experience and seniority within agencies, consultancies, client-side organisations, the non-profit and government sectors, support services as well as institutions and the academic community. As well as over 2,000 individual members, the Research Society has 125+ company and client partners, with the number continuing to grow. The Research Society research professionals and company partners commit to and are regulated by the Research Society Code of Professional Behaviour.

Be the first to comment

Leave a Reply

Your email address will not be published.