Daily Archives: January 23, 2016

Confidence limits: polling organisations’ failures undermine market research credibility more widely


Things are hotting up in the opinion research environment. The climate of opinion about those who measure our opinions is undergoing significant change.

It’s an inconvenient truth, but the debacle of the UK’s May 2015 General Election casts a long shadow over research in related sectors, like marketing.

How could so many eminent polling organisations fail so spectacularly in their fundamental task?

Nobody likes to dampen the enthusiasm of the researcher, of course, and I may be a sample of one, but nowadays I can’t help wondering whether some or all of the forecasts she presents about brand awareness are worth the electronic paper they’re written on. And have attitudes to the client’s brand really changed that much, after that fairly low weight TV campaign?

Accusations of ‘herding’ have been roundly dismissed. 93.5% of polling organisations, when questioned, spontaneously replied that herding had not influenced the results of their polls. Which is quite comforting.

And it’s good to know that they’ll be changing their methods. Far too technical a topic for my little brain to understand, but it’s something to do with unrepresentative samples. So at least they’ve managed to identify the source of the problem. Good that the experts are now on the case.

Forecasting has never been an exact science, of course. Weatherman Michael Fish‘s “Don’t worry there isn’t …” response to predictions of the devastating 1987 hurricane still echoes in many memories.

But with the accumulated statistical knowledge and sophisticated computer systems that are now available, we can surely be forgiven for expecting a degree or two more accuracy in such an important sector?




Leave a comment

Filed under Media research