Saturday, December 16, 2017

Thank you, Nigel

 Nigel Henry, lead analyst and CEO of the public opinion research firm, Solution by Simulation, wrote an open letter to me the other day in response to a comment I made on I95.5 FM radio that I could not believe a reputable pollster like him would predict the outcome of an election 16 months in advance. He used the device of the open letter because he believes “the public will benefit from understanding why such opinion research is appropriate, and how it should be interpreted”.

So his letter to me was for public consumption because of its benefits to the public in terms of better understanding of opinion research on elections. As he would explain subsequently to me in another letter, this time private, he made the first letter public because, while I might not have been responding to the polls directly, it was reported as such, and he felt it necessary to respond to “what was reported” that I had said.

Being first and foremost an educator, I thank Mr Henry for his open letter because he took the opportunity to attempt to increase public understanding of polls on elections, especially in this country. Any increase in public understanding on any matter is good since it has the potential to improve our construction and negotiation of experience, and any increase in public understanding on election polls is good as well since it has the potential to make us better interpret those polls and the political behaviour of people and parties. I teach to increase the understanding of my students on a wide variety of matters, and to enhance their cognitive and methodological skills in the acquisition and construction of that understanding. In his open letter, Mr Henry joins me in that enterprise, I think.

In my first response to his letter, I clarified my reported comment as follows:

“I didn’t hear the news report on my interview, but you need to know the context in which I made the statement. The interviewer was providing me with her sense of the results of a poll, which, it turned out, she had not carefully read, and asking me for my views on it. I told her I hadn’t seen reports about the poll so could not properly assess it. She (mis)informed me that the poll was in respect of the 2015 general elections. Speaking hypothetically, I observed what you reported in your opening statement.

It was a hypothetical comment done without any intention to ridicule. I genuinely did not believe you could make the prediction you were being reported to have made. It turns out I was right.’’

He has asked me to join him “to raise the acceptance of research, and raise the quality of analysis in the political landscape of Trinidad and Tobago”. That’s what I have been doing in my commentary on elections in the country, and have continued to do since he came on the scene. Indeed, I relied on his work last year, predicting victory for the PNM (three times) and Jack Warner in elections. 

In response to his attempt at public education via the open letter, I will now tell you what I have learnt.

First, I should not comment on secondhand information; it is better to read the original reports and then comment. Hypothetical comments from political commentators may be good for certain contexts, notably academic ones, but apparently not for the context of commentary on secondhand reports. The hypothetical nature of the comments may not be available to the consumer.

Secondly, I should observe the difference between what a poll captures at a particular point in time and what outcomes it can be used to predict. A “point in time” is the period during, and in respect of, which the poll was conducted. The pollster reports on the opinions collected during that period and analysed in respect of it, but does not make predictions on his analysis. It is people like political commentators, reporters, and the average citizen who predict—that is, make judgments about the future on the time-bound snapshot of the pollster. Mr Henry, therefore, did not predict the outcome of the PNM elections; he asked his respondents what that outcome might be “if the elections were held now’’, now being the period (more or less) in which the opinions were polled. 

Thirdly, even though pollsters do not make predictions based on their snapshot, time-bound findings, the latter, if properly contextualised, can be used (by others only, Mr Henry? Why not also the pollsters themselves?) to make predictions years in advance. For example, according to Mr Henry, “we ‘know’ (through public education) that Republican Gov Chris Christie is a better bet than Sen Ted Cruz or Sen Rand Paul to beat Democrat Secretary Hillary Clinton, and those Republicans would have a better chance against Democrat VP Joe Biden.’’ And some of us, I imagine, ‘know’, based on the results of the poll on the PNM elections, who will emerge as political leader. 

But I must observe that, in respect of some temporal distances, I would be hard-pressed to use all the particularities of context to show how a prediction will come true. Sixteen months before the next general election, I couldn’t convincingly show, though I could provide some pretty good arguments, how a prediction that either UNC or PNM or People’s Partnership will win would come true. In politics, certain events can cause opinions to change dramatically.

Mr Henry makes a very insightful statement, which bears repeating in full:

“Up to this point, we have largely relied on a tallying where each voter places a single ‘X’, as the sole measure embodying our preferences on the myriad of complex, constantly-changing issues that face us as a nation in any given five-year period.  This is a strange and outdated way of doing Democracy.  A more data-driven and evidence-based approach is now possible and must become our reality.’’

I couldn’t agree more. This is a matter that I have treated time and time again, and developed in ways that Mr Henry has not begun to broach in any public statements I have heard or read from him.  I will come back to it again, of course.

• Winford James is a UWI lecturer and political analyst