Punjab Forecast: Prannoy Roy’s Nate Silver-Donald Trump Moment???

Tuesday, February 07, 2017

When Dr Prannoy Roy switched from his usual votes, seats, shares and swing projections to dishing out probabilities, he was actually playing chicken à la Nate Silver of the website www.fivethirtyeight.com, an American statistician and writer who analyzes baseball and elections. He is the editor-in-chief of ESPN's FiveThirtyEight and a Special Correspondent for ABC. Silver is not merely a living icon in the US but the world over. But Silver still finally suffered his Waterloo moment with Trump’s victory.

Vote shares, seat shares, swings are measurable, reflecting a higher degree of certainty and thus lend themselves to public accountability. Probabilities are vastly different. Like Silver, Dr Roy too now can take the route to argue that his probability of AAP winning 57.5% (read here) theoretically is merely a forecast and not a prediction. The difference between the two terms is that essentially a forecast is not a certainty but just “an estimate of risk”. What is overlooked is that Dr Roy’s probability implicitly also gave AAP 43.5% chances of losing!! In short, whichever way the result went, a pollster who dishes out probabilities still can claim his forecast was spot on just as Silver continues to do so with his probabilities of the US Presidential elections last year. Never mind that millions of Nate followers took his numbers literally ended up feeling betrayed and deeply let down by their prophet! Trump’s victory also heralds the death of their prophet!

Then there is Michael Moore, internationally acclaimed documentary producer. Being a diehard Left liberal with a deep seated hatred for Donald Trump is a characteristic trait he shared with Nate Silver. But there the comparison ended. Right from the primaries when Nate dismissed Trump as a non viable candidate, Moore in contrast, even in the early days of the US primaries, warned the liberal movement of the dangers of the rise of Trump and campaigning for Left Liberal Bernie Sanders who he felt had the best prospects of trouncing Trump. But unlike Nate who used numbers, Moore relied on more qualitative interpretative arguments. It is easy to get carried away with numbers and as a result, the liberal movement preferred to embrace Silver’s forecasts with both hands and chose to dismissively ignore Moore who in hindsight is now widely accepted had been perceptibly right on target.

Today in India we have a choice. Either go for dazzling numbers dished out by the likes of Prannoy Roy, Yashwant Desmukh etc or go by the field reports of old time style journalism from the likes of Shivam Vij, Prashant Jha, Rohini Singh and so many like them who provide us interpretative answers and not so much solid but meaningless numbers. I have many times jokingly tweeted to Yashwant Deshmukh that I have not much faith in the numbers that his CVoter dishes out but have complete admiration for his more substantive analytical election related articles. This I suppose holds equally true for all truly psephology freaks, rather than numbers, go by the analytical reports of the likes of Yogendra Yadav, Sanjay Kumar and CSDS rather than seductive topline numbers of Today’s Chankaya, AC Nielson and their likes.

Provided below are extracts from an excellent article by Marlena Trafas in the Weekender on lessons learnt from the last US Presidential Elections 

While there are many explanations for how Trump got elected, two stands out in particular for how well they exemplify the failure of numbers to fully explain the complex world we live in. One explanation comes from resident Numbers Guy, Nate Silver, the founder and editor-in-chief of FiveThirtyEight, a site that “uses statistical analysis — hard numbers — to tell compelling stories about politics, sports, science, economics and culture.” The other explanation comes from liberal filmmaker and author, Michael Moore.

A week after the election, Silver appeared on The Daily Show and was interviewed by the host, Trevor Noah. The interview began a bit awkwardly as Noah says to Silver, “Let’s talk about polling. You’re not a pollster, you aggregate the polls. This entire race you were wrong.” To that Silver responds, “No,” and the entire audience erupts with laughter.

This audience response makes perfect sense, as it is now clear that all the predictions and forecasts and probabilities that expressed very little chance of Trump winning the presidency were wrong. Now, Silver wouldn’t say they were wrong. In fact, his entire interview with Noah is a defense of the polls he used, saying there’s “not an easy television explanation,” saying these were “some of the best polls,” and that his “forecasts” were actually not “predictions,” but “an estimate of risk.” He describes percentages and “if this, then this” explanations based on calculations, maintaining that if there were a 30 percent chance of rain, people may bring out their umbrellas, thereby suggesting that people should’ve been worried about the 30 percent chance of Trump being elected.

Despite Silver’s reasoning, many everyday people took his numbers to mean that Trump in fact had no chance of winning. Silver argues that people shouldn’t have thought that based on the numbers he provided. It’s this frustrating, circular, contradiction that leads to Noah’s exasperated ending joke, “I’m gonna choke you right now, Nate Silver.” If one watches the interview, it’s clear that all the smart talk Silver delivers never really feels like a true answer or explanation.

When compared with Moore’s “5 Reasons Trump Will Win,” Silver’s numbers become abstract and meaningless as they translate why Trump will win into zeroes and ones. Moore, unlike Silver, uses words — qualitative evidence — history and sociology to explain the human behavior and feelings that would lead to a Trump presidency.

Moore’s reasons range from “The Last Stand of The Angry White Man” to “The Depressed Sanders Vote,” and reading through them on his website makes it it hard to discount the fact that they make sense — definitely more sense than Silver’s mathematical explanation. Moore’s explanations are ones people seemingly only came to fully realize after the election. But Moore made his prediction in the summer of 2016. And very few people seemed to take him seriously. The days before the election I heard hopeful, confident talk of a Madam President, not wariness about the “Angry White Men” who would vote for Trump because they felt left behind by the current cultural-political moment.

Put simply, the guy who used numbers and data and stats incorrectly predicted the election and the guy who used words, history, and qualitative facts correctly predicted the election., Leading up to Nov. 9, many hopeful liberals put their faith in Silver’s numbers rather than Moore’s words to explain how humans would act that Tuesday. The numerical data was trusted and the numerical data let a lot of people down.

While there are many lessons to take away from the election of Donald Trump, one is that numbers can lie. Math and stats and data points don’t always tell it like it is. Sometimes you need interpretative answers rather than the “correct” answer. Sometimes the guy spouting words and using simple addition has a keener sense of the forces that govern our society than the one using statistical analysis.

Read the full article: Here

You Might Also Like


Like us on Facebook

Follow us on Twitter