The 2015 General Election – The End Of Market Research As We Know It?

I am going to try to avoid politics altogether, both the result and the electoral system which delivered it, since this has no place in our politically neutral Marketing Society. At least I’ll try my best, because what I really want to focus on is why the opinion polls got it wrong. This is very important because the election polls are not only the most visible and engaging form of market research for the nation as a whole, they are also highly influential on how most senior business people, those who are not career marketers, view market research. Putting it bluntly, if opinion polls can get it so wrong how can you trust any market research?

This happened in 1992. The opinion polls were predicting a clear win for Labour led by Neil Kinnock and as we know the result was a resounding triumph for John Major and the Tories. There was indeed a strong move to Labour who picked up 42 seats from a 3.6% swing in votes. The Conservatives lost 40 seats (compared to Thatcher’s landslide win in 1987) but they secured 14.1 million votes, 41.9% of a high turnout of voters, enough to give them 336 seats, a big majority. The real losers were the LibDems under Paddy Ashdown, who lost 20 seats and retained only 22. Sound familiar? So will this. The two issues of the election were the economy, we were in a recession, and immigration. The Tories campaigned on both issues with scare stories about high taxation, high inflation and “opening the floodgates to immigrants”, under Kinnock’s Labour party. It worked. There was the infamous gaff of Kinnock arriving in Presidential style to a pre-election Labour rally, then tripping over and falling into the sea the following day. John Major went round the country with a little wooden soap-box, and the Sun newspaper apparently won it with their high profile support for the Conservatives. So was it the policies, the fear factor, Kinnock looking a bit pompous and accident prone, Major looking like a decent guy, the support of Murdoch’s press? Who knows – all of the above – but the fact is that opinion polls spectacularly got it wrong, and did so right up to the moment real votes were cast.

I can remember clearly in the months that followed the 1992 election being in meetings and presenting market research data to support some recommendation or other and seeing the cynical look on the faces of the non-marketers in the room. Some of them came right out and said it – “Yes, but market research said Labour would win”.

I predict this will happen again. Research clearly tells us that the green pack is preferred. Yes and research told us emphatically that it would be a hung parliament. So, my fellow marketers, all of whom will want to continue using market research to inform decision-making, you better get your arguments together. And I want to help you, even though part of me is celebrating the end of market research as we know it. Even though I do regard 99.9% of market research as another “higher form of ignorance” (see my recent post).

Back in 1992 we used the following arguments. The polls did not ask the right question. They asked which party you would vote for, not which leader you prefer and which policies are most important to you. Overall the polls showed that the majority of people thought they would vote Labour, but they also thought Kinnock was a bit of a pillock and they did not trust Labour with the economy or immigration. That is what actually drove where they cast their vote in the secrecy of the polling booth. Opinion polls in the USA are more accurate because their electoral system separates party preference from Presidential choice. Any anyway, you should never use quant research on its own, focus groups would have told you that claimed voting intentions masked an underlying lack of confidence in Labour and their leader among the key socio-demographic groups Labour needed to win over to secure a victory.

The same arguments can be applied to this last election – even the same issues and the same groups. Let me illustrate this with the purest form of qualitative research. I met up with an old colleague, a middle class Northerner and lifelong Labour supporter, just a week before the election. Let’s call him Tony. For Tony, support for Labour is visceral, he admitted that. He could never imagine not voting Labour. I did not try to dissuade him, I simply said, “Miliband, Balls, SNP, fragile economic recovery, really, I mean really?” I promise you, he gave out a long sigh. Yes, he said, I guess you’re right. No idea how he actually voted but you get my point. Party allegiance is not the same as your actual vote when you consider the real issues and the real options, and as you wrestle with the reality of our electoral system. You are not just casting your vote for your local MP, you are also casting your vote for who governs the country in a first-past-the-post system. That alone makes a fool of Opinion Polls – and yes, I am aware that Lord Ashcroft and his team developed a way of asking the questions that separated the local MP choice from the national choice but it didn’t work did it? His final poll put Labour and Conservatives both on 33% and he was publicly saying a hung parliament was by far the most likely outcome.

So let’s summarise – market research is fine but a) you’ve got to ask the right questions b) you’ve got to segment the respondents and c) you have to use quant and qual to get the full picture. Right? Wrong. You have to accept that 99.9% (that is statistically significant by the way) of all market research never gives you the full picture and will never reliably predict outcomes for one very important reason. Humans. It fails to take account of the human condition, our genetic programming and how the brain actually works.

Not only do we not know what we think, we do not know how we think. So how can any question posed in a market research context ever predict or even fully explain anything?

I could elaborate but time’s up (I have a word count limit I have to stick to). So let me just close by saying the 2015 Election should be the end of market research as we know it, and good riddance. It should now be the age of market intelligence.

Leave a Reply