Our gerontocracy


Frank Graves considers the ramifications of declining voter turnout.

Have we passed the brink from democracy to oligarchy? While on pattern with a disturbing downward trajectory in voter participation, this movement into the realm where the majority of citizens aren’t voting may be a wakeup call for those who think that elections shouldn’t be on track to a fringe activity. What may make matters worse is that this democratic decline is particularly pronounced in two sectors of our society — the young and the economically vulnerable. Although these groups have always had lower participation rates, their current anaemic levels of voting make it difficult to claim that governments legitimately speak for these large and growing portions of society.


Our gerontocracy

  1. Dot considers the ramifications of declining voter turnout.

    More opportunity for talking heads like Frank Graves to fudge/adjust his “expert” polling data to “correct” for sampling error/bias, and tell us all what it means…


    • Care to tell us just how Graves fudges his data and corrects for bias/error in order to reach what you seem to regard as biased conclusions on declining voter turnout?

      • Yeah, he wrote about it some time ago – think it was in the G&M – how he takes data from his polls, and then massages them for factors such as low voter turnout for youth, conservative bias and high turnout for seniors, use of cell phones etc. It was in his piece when he explained why he f**ked up in the last election.

        Here’s the long version of his mea culpa:

        • so it is fudging to weight to the characteristics of the actual voters in order make a better prediction of electoral outcomes? The whole point is that the gaps in May federal polling and the final outcome were mostly gaps in growing differences between those who turnout and all eligible voters. There was little  evidence of bias which is why this time around  the predicted models,  once weighted to our best estimates of the actual voter population, got the OLP and PC results within about one point of the actual outcome (in this most recent Ontario election)
           . Are you suggesting that there are not systematic differences between the population of all eligible voters and the subpopulation of those who vote? That is simply wrong. Not only are there widening gaps between the characteristics of eligible and actual voters, these are increasingly correlated with party preferences . Any casual review of the literature will confirm this as did our recent experience. We also provide the results for the entire population of all eligible voters but these data are increasingly not reliable for predicting vote outcomes. If you more carefully had  read the article you cite you would  note that the prediction error in the last election was based on things like the inclusion of cell only respondents who were both less likely to vote and less likely to support the CPC. In the recent Ontario election we continued to randomly sample all eligible voters (cell only , non internet , et, al,) What changed was the final forecast took into account the likely differences between actual and eligible voters and provided an accurate forecast . We also provided the data for all voters which is arguably a more important task than modeling the final vote outcome . We didn’t mysteriously massage or “fudge” to correct for bias. All of our treatments were clearly explained and all of our data are reported in an  open and transparent manner.It was based on careful analysis of the forecast error on May 2nd that we avoided a further error this time around. Serendipitously, we also saw some interesting shifts in the voter landscape which you are perfectly entitled to depict as banal. Your depiction of our analysis and data collection as ” fudged” and biased is untrue and unfair.Frank Graves

          • Yeah, it is. Because the correction factors are a moving target – derived from snapshots of final polling results that are obsolete as soon as you derive them. More of an art than a science, I’d argue. Why “experts” are needed to “interpret them”.

          • well the future doesn’t always resemble the past and yes the models can vary from time to time . We dont have a unified theory of voter  turnout but that concern is a far step from we fudged our biased data. As for the need for experts your commentary seems to contradict that point . I do think I pick up some things that are worth sharing with others and which improve the debate but you are certainly entitled to disagree and raise objections .  I do want to stress that we NEVER fudge our data and while these adjustments may not work next time they did work (in advance) this time.  —

          • FGraves,
            There is more than one interpretation for “fudge” and you are obviously taking my general comment personally, and using the most pejorative interpretation.
            So, let’s examine more closely what I wrote: “More opportunity for talking heads like Frank Graves to fudge/adjust his “expert” polling data to “correct” for sampling error/bias, and tell us all what it means…”

            “More opportunity” means that – opportunity
            “talking heads like” – OK, maybe you don’t like the term “talking head”, but that’s part of the turf from appearing regularly on political shows, “interpreting” your data

            “fudge/adjust” – so here I suggest two options – one negative, and one that could possibly be justified through a process that, while transparent, could equally be biased.

            I don’t believe I ever stated that you, personally “fudge” in the most negative interpretation. So, I don’t at all accept your claim that I am being “unfair”.

            But, lets get back to the relevance of daily polling and seat predictions. I have always been of the opinion, on this board at least, that they are meaningless, and serve more of a marketing ploy for firms like yours. And certainly when KO’M was here, she dined daily on your free polls and others. Check the commentary back in 2006 – I posted then as “Dot”.

            The “correction” factors you select are based upon criteria that you have determined are valid – be it age, sex, ethnic background, geographic location, whatever. But, there could be any number of other factors, and fleeting as well.

            Just because industry standards or rule of thumb or convention suggests the factors developed give accurate predictions of past votes, it does not necessarily follow that the same holds for the future, no matter how you adjust them.

            You obtain raw data. Then you attempt to convert it into useful info. Call it sausage making, call it cookie baking, call it making fudge, it’s all basically the same process.

            And what different pollsters are attempting to do is claim they are the best bakers – whereas it is debatable whether the whole exercise is anything more than free advertising for the bakery.

            I’ll wait for the cake or whatever to come out of the oven on election day. And I’d suggest others do as well. But they won’t.

          • if you read our last poll of the Ontario election you will note that we make many of the same points. I detest the increasingly flawed yardstick of how close the last poll came to the election result that everyone will know anyway in less than 24 hours. ( I subtitled the last poll our least important). I also agreed that the  future doesn’t need resemble the past but there are no agreed upon cookbooks for creating accurate predictions (which we both agree are kind of pointless). I have argued that good polling does far more important things than predict the horse race outcome but you would be hard pressed to find much evidence of that in media coverage
            I am not that convinced that your explanation  on alternate meanings of fudge is  that convincing but let’s just leave it at that and note that it appears that we agree on many points . thanks for the clarification .

        • I see Graves got here first.[ how do they do that?] Nevertheless that’s not a mea culpa, not in my estimation; it is more in the way of a thoughtful post mortem.
          That said my answer from a purely layman’s perspective of one of his questions – does polling effect the outcome and is that simply providing useful information to the voter is yes and yes, but…I’d prefer no polling afer a certain point and this kind of post mortem afterwards. Sorry if that cuts into your business Mr Graves. I’m concerned the over professionalization of politics in general[ too much spin. possibly deliberate voter supression tactics, etc.,] is part of what is turning people off politics.

          • In our post mortem report we formally tested the hypothesis of whether the polls influence the outcome of that election and the answer was that in this case the net impacts were immeasurable . Some were affected but it was a small number and there was no clear pattern favouring any party . This conclusion that polls are reflective not causal is consistent with the academic literature.
            By way of background we do no paid political work for any party at EKOS and our business is not based on elections or political polling . There may or may not  be over professionalization of politics but we are not involved. Interestingly, if media polling were banned , the parties would be the only ones with access to polling (to use and spin as they saw fit with no accountability to public polling). 
            The professionalization of politics is certainly a broader issue . I think that one of the key problems we now encounter is a growing disjuncture between the interests of parties (to get elected) and the overall public interest. It’s hard to fault parties for pursuing success in the electoral marketplace. It’s equally hard to fault the public for being mistrustful when party interests eclipse the overall public interest. We may need to be considering institutional reforms to democratic practices to narrow the gap between party and public interest.(e.g. mandatory voting would render vote supression  strategies useless). It is notable that the steep decline in trust in government over the past 40 years has been led by declining trust in political parties (and politicians).  The key driver of this is precisely the conviction that public interests have been compromised.

          • Thx for that. I wonder to what degree this public conviction that the parties are more concerned with their interests foremost is a kind of closed loop – is it getting worse as voter turnout drops? Other pundits[ Hebert] seem to feel those numbers are deceiving – people will come out when they feel compelled to – as in the prorogation saga. I don’t tend to feel so sanguine myself. 
            Manditory voting might do it, certainly Australians seem to have adapted and now see it as being not a big deal. However i heard tales of people voting strategically or basicallly throwing their second or third[?] picks away on say the communist party, which laughably distorted their share of the vote. This was back in the eighties so i’m a little dated and not at all sure if this was simply a quirk of their voting system or the manditory vote.[ not sure if it was manditory back then ,either so i may be off base here]  i don’t suppose manditory voting in our system would run into similar problems.

  2. These people will start to vote when parties start addressing their concerns.  When you’ve got a McJob and are living hand to mouth, you don’t care about macro-economics. You don’t care about generalities around increasing employment opportunities, investment into Canadian companies, longer prison sentences, new ways to avoid paying taxes you don’t pay already by saving money you don’t have anyway, or any of that crap.

    You care about how you’re going to balance your rent, electric bill, and student-loan payment. You care about why the government is taking so much money from your paycheque now and keep it until May or June when they give some of it back. You worry about whether your GST rebate will arrive before your mom’s birthday so that you can pay your car insurance.  Non-refundable tax-rebate for a bus pass?  How about a tax rebate for what they charge at the local Esso?

    Until the parties start dealing with the things these groups of people care about, there’s no reason for these people to bother voting.

Sign in to comment.