Fairfax’s polling bias

May 10, 2021

Progressives should not necessarily be concerned about the Crosby Textor connection of the new SMH and Age pollster. The key question is how the polls are reported.

Should we be worried that the newspaper group chaired by Peter Costello has chosen an ex-Crosby Textor pollster to feature in their newspapers, most noticeably the Age and the SMH?

Not necessarily.

As reported on the Poll Bludger website, the new firm, Resolve Strategic, is headed by Jim Reed who is ex- Crosby Textor.

The Poll Bludger also reports relatively positively on the transparency of the initial presentation of the results.

It is wrong to assume that such a pollster will necessarily deliver more pro-Conservative results. In my experience pollsters value their reputations. In most cases they make their money on commercial activities for which their reputation for getting the right result is more valuable than their political connections.

That is not to say that some polling organisations don’t tend to deliver results more favourable to one side or the other.

It is also true that polling has frequently had a more left-leaning bias in its results in recent times and any redress of that would be welcome both from a transparency point of view and as a basis for sound planning.

It is highly likely that both Bill Shorten and Hillary Clinton would have finished their campaigns differently with a more accurate reading of the likely results.

For example, I am quite certain Hillary Clinton would have campaigned in Wisconsin if the polling had shown the extent to which it was at risk.

And in Australia many will remember the several recommendations from the excellent Wetherill/Emerson review of the 2019 ALP election failure which referred to an “industry-wide failure” by the polling industry and the consequent excessive expectations which led to campaign misjudgements.

The slight but noticeable trend to results too favourable to progressive parties was counteracted in the most recent US presidential election by the emergence of a polling organisation, Trafalgar, whose claim to fame was forecasting the pro-Trump result in the mid-west in 2016.

In 2020 Trafalgar produced results which were consistently more favourable to Trump than the results produced by the majority of pollsters.

There is no evidence to suggest that Trafalgar did any better than more mainstream pollsters such as Ipsos in forecasting the 2020 result. In their last polls Trafalgar predicted four States for Trump which actually went for Biden: Georgia; Nevada; Pennsylvania and Michigan. Ipsos got only two wrong at the same stage: Florida and North Carolina.

However, the average by which all the pollsters were wrong was similar, some too low for Biden, some too high.

The debate about methodological differences between the pollsters is arcane. However, it seems all of them in the USA and Australia have difficulty getting a representative sample because of the differing response rates between various segments of the population.

Should Resolve Strategic challenge the established polling organisations that is probably a good thing. What we should be worrying about is not the pollsters but the reporting of the results.

The journalists writing in The Australian are almost universally appalling in their reporting of polling results from Newspoll. It is an offence against common sense to pretend that a 1% change in any polling result is anything other than the typical statistical variation which is inevitable in a sample-based survey. To put these apparent variations in perspective, a 1% variation means more than 150,000 Australians have changed their voting intention. Sometimes that happens, most weeks it does not!

It is also obvious to the casual observer that the Australian’s journalists relentlessly seek and choose to highlight the elements of each poll which are capable of pro-Liberal interpretation.

This is more of a problem than Newspoll itself, which has a reasonable track record over the years.

It will be interesting to see how the Age and SMH journalists interpret the results from their new pollster.

Share and Enjoy !


Receive articles straight to your Inbox

How often?

Thank you for subscribing!