Voting intention opinion polls: time for a change?

The result of the Brazilian elections' first round last weekend presented great disparities when compared with the polls that preceded it. Although analyses are pointing out to "last-minute changes" or "embarrassed voting" to justify such discrepancy, not to mention some granular quantitative evaluation (in Portuguese), the context tends to reinforce the need for new approaches in this #research space dimension.

Such change is something that I had the opportunity to defend and work on, over the past years, in the field of #market and behavioral research for #innovation strategies effectively oriented to #people. Call it applied ethnography research, contextual research, or if we stretch it a bit, public participation in scientific research (citizen science). Goes without saying this is not the same as voting intention opinion polls - but that's precisely the point. Considering the times we are living, shouldn't #quantitative and #qualitative get way closer?

In the past, one of the key "divergence generators" in electoral quantitative polls consisted of sample projection. That is: if a certain #population represented X% of the national #demography, and then represented Y% within the poll sample, then (in a sort of rule of three) the total result of this demography was Z%.

What's the problem? The use of traditional #socioeconomic #demographics (#wage, #age, #gender, #location etc) as a research starting point no longer serves as a #parameter for #sample construction, findings and #analysis. The grey areas - i.e., the qualitative context - need to be taken more into account before the quantitative setup.

For instance, #identity and #behavioral clustering - where people's #worldview, their #personal values, and their #habits come to be considered for a quantitative screening - should receive an additional #methodological stimulus so that we can evaluate if it is possible to find final results less divergent with #reality.

The such approach implies even harder #work upstream, which must include controlled #experimentation. Something to be better explored and deepened (not least because there is no rocket science or major #disruptions involved in a such proposal) so that we can avoid demonizing research #institutes doing serious quant work.

And this is a very tangible way of tackling opportunistic initiatives trying to take advantage of our complex crossroads between #democracy, #populism, and fear of #change that we are experiencing more and more, everywhere.

This is breaking news already when it comes to Brazil's run-off: just one day after the first round, president Jair Bolsonaro's House leader has announced he will present a bill to criminalize survey errors (in Portuguese). Bolsonaro is seeking re-election and his final voting results were particularly dwarfed when compared with previous polls.

Still this week, the politician has accused research institutes of "meddling in democracy" (again in Portuguese). And his Justice Minister, Anderson Torres, has announced on a social network that he had determined the Federal Police open an inquiry to ascertain whether there was a crime in the institute's actions.

Would that be enough for us to explore new possibilities?