The RMIT-ABC Fact Check unit did a fact check on Chris Berg and my claim in our recent book Against Public Broadcasting: Why and how we should privatise the ABC and repeated in an article in The Spectator and on Sky News.
Professor Davidson’s claim is flimsy.
I will let readers decide for themselves on this point.
In making the claim, Professor Davidson referred to a study published in 2013 which surveyed 605 journalists from a variety of organisations on their voting intentions.
Fifty-nine of these journalists were from the ABC, and only 34 of them answered the question on voting intention, with 25 either undecided or electing not to answer.
Of the 34 who did answer, 41.2 per cent, or 14, said they would vote for the Greens.
But experts told Fact Check that the ABC sub-sample was too small and the rate of undecided and non-response too high to be able to draw accurate conclusions from the survey on ABC journalist voting intention, let alone voting intention of all ABC employees.
This may well be what “experts” told the RMIT-ABC Fact Check unit. But what did the author of the study say? The author of the study said: “It means that even though only a smaller number of journalists answered the voting intentions, which does increase the margin of error, it is still reasonable to conclude that there is a marked difference between the voting intentions of journalists at the three major media organisations”.
Upon releasing the findings in 2013, the author of the study himself, Folker Hanusch, inserted numerous caveats about using sub-samples of the survey, including that the margins of error would be larger than those for the total sample. Professor Davidson neglected to include any of these important caveats in making his claim.
Upon releasing the findings in 2013 at The Conversation it was reported: “Sub-samples of journalists’ responses to some questions – such as voting intentions – are likely to have a higher margin error, however, appropriate statistical methods were used in testing for differences between sub-samples to take account of the smaller sample sizes”.
Experts contacted by Fact Check, including Professor Hanusch, also took issue with comparing the results of the survey with larger, more stable studies of the voting intention of the general population, such as Newspoll.
We did not compare the results of the Hanusch study to Newspoll.
Whilst the survey found that Australian journalists in general tend to skew left, it showed no evidence that ABC journalists were five times more likely to vote for the Greens than the general public, and experts contacted by Fact Check said they did not know of any other recent studies which canvass the voting intentions of ABC journalists.
Indeed the Hanusch study did not find ABC journalists being nearly five times more likely to vote for the Greens than the general public – Chris and I calculated that statistic using the Hanuasch study and the 2013 election results. The RMIT-ABC Fact Check unit has (incorrectly) criticised us for comparing the Hanusch study to Newspoll so they know that we had augmented the data for our analysis.
University of New South Wales statistician Jake Olivier compared the survey results with those of Newspoll over the same survey period, and found ABC journalists were 2.4 times more likely to vote for the Greens.
This is the same RMIT-ABC Fact Check unit that just three paragraphs ago criticised me for not talking about margins of error and the like that has now neglected to include Jake Olivier’s full analysis. Let’s go to the tape: “Using this model, he calculated that ABC journalists have a 2.9-fold increase in the odds of preferring the Greens over Newspoll respondents, with a 95 per cent confidence interval between 1.6 and 5.2”. That range includes our estimate of nearly five time (actually 4.74 times). Note, however, we didn’t use Newspoll – we used actual election results.
But he cautioned that the sample size of ABC journalists was too small to make strong conclusions about this result.
I can’t possibly comment on whether Jake Olivier’s analysis has correctly adjusted for small sample sizes. It is not peer-reviewed research. All I can say is that Folker Hanusch’s peer reviewed research did claim to have made use of the appropriate techniques.
Fact Check could find no research on the voting intentions of all ABC employees including those outside of the news division.
False claim – Chris and I spoke specifically about ABC journalists, the RMIT-ABC Fact Check unit took it upon themselves to interpret our claim to include all staff given our proposal to privatise the ABC by allocating shares to all employees.
Now I could go on and do the whole of the RMIT-ABC Fact Check article, but I think people can get the idea.