Facebook Google Menu Linkedin lock Pinterest Search Twitter

Advertisement

Gerard Henderson

Sep 25, 2009

Henderson on Gans & Leigh

In Gerard Henderson's Media Watch Dog post on September 4th he took aim at a recently released study conducted by Joshua Gans and Andrew Leigh examining political slant in the Australia

Share

In Gerard Henderson’s Media Watch Dog post on September 4th he took aim at a recently released study conducted by Joshua Gans and Andrew Leigh examining political slant in the Australian media. Toby briefly mentioned Henderson’s criticism at the time

  • The Gans and Leigh “Media slant” study is useless because it has mathematical equations in it; no research that relies on formulas can be trusted.
  • Even though that quantitative and equation-filled study should be disregarded, and even though it found that The Age leans to the left, The Age leans to the left.You don’t need statistics to prove it, though – you just need Gerard’s judgment.

  • Henderson then took another swipe at the study in his September 18th post labelling it as the quite wacky report by academics Joshua S. Gans and Andrew Leigh and going on to make ad hominem attacks on Professor Gans, based on the mistaken assumption that a link to the XKCD online comic on Professor Gans site was a representation of his research.

    Andrew Leigh took it upon himself to write to Gerard and point out this error, as well as engage him in honest debate regarding the study he and Gans co-authored.

    The source of your disagreement with Professor Gans seems to stem from the paper on media slant that he and I have coauthored. In particular, you seem concerned by our use of mathematics to express our methodology.

    Today Leigh has updated the post with a response from Henderson which included this refutation,

    I criticised the paper you co-authored with Professor Gans titled How Partisan is the Press? Multiple Measures of Media Slant. I did so because I believe that it is seriously flawed – not merely because of the use of mathematics to express your methodology.

    which makes me wonder why Gerard inserted the following two pieces of snark in his September 4th post

    Now for a full disclosure. MWD read the Gans/Leigh paper when there was a full moon – or close to a full moon. This may explain MWD’s inability to follow some of the insights of those two insightful academics. Take this insight, at Page 9, for example:

    To estimate the political position of each media outlet, we simply estimate a weighted OLS regression, in which the dependent variable P is the share of Coalition mentions by a given public intellectual i in media outlet j in time period t, and the independent variable is a vector of indicator variables for each media outlet:

    So that’s pretty clear, then.

    [Not really. Are taxpayers paying good money for this sludge or is this all a fake? Ed]

    which was followed by

    Indeed, at times, the analysis is so vague as to be incomprehensible. Take this you-beaut analysis on Page 10, for example:

    We opt to use a weighting scheme that is the product of the square root of the number of parliamentary mentions and the number of media mentions. Where p is the number of parliamentary mentions received by public intellectual i in period t and m is the number of media mentions given to public intellectual i in media outlet j in period t, the weight w given to a particular observation is:

    Using square root weights has the advantage that (unlike log weights), the weights are still defined for observations with zeros. It also captures the intuition that the standard error of the mean of a binomial variable is equal to the square root of the sample size, multiplied by the mean, multiplied by one minus the mean, ie.,

    Go on. Unfortunately Messers Gans and Leigh did just this. And, unfortunately, The Australian thought that the results of this study – which commenced as an ANU seminar – were worth reporting. This, apparently, is what passes for scholarship in the groves of the ANU.

    Henderson’s response to Leigh’s letter, particularly his unwillingness to admit he was been mistaken about the source of the urinal comic that he held up as evidence of Professor Gans’ work, is very disappointing. Henderson chose to play the man rather than the ball, and refused to admit to doing so when called on it. By choosing to demean people he disagrees with rather than debating their arguments, Gerard Henderson risks tarnishing his own reputation as a serious commentator, as opposed to some of the more partisan polemicists who usually engage in such behaviour.

    Advertisement

    Advertisement

    We recommend

    From around the web

    Powered by Taboola

    32 comments

    32 thoughts on “Henderson on Gans & Leigh

    1. Eponymous

      This is old, but I’m glad I read it. In moments of weakness, I often think GH might have some points some times. But here, he really is just saying ‘i don’t understand this, therefore it is not true’. What an incredibly arrogant position to take.

    2. Tobias Ziegler

      Henderson’s regular opinion-writing gig is with the SMH, Lev – although I don’t think that necessarily invalidates your point.

    3. Lev Lafayette

      Henderson is only the voice-box of the foolishness. The brain is the editors (and, by extension, the majority shareholders) of ‘The Australian’ who choose to pay this member of the chattering for such opinions.

      One could be forgiven for thinking that his opinions are in their interests.

    4. John Greenfield

      Phillip

      I think we would get on just fine. I am alarmed at the completely unacceptable mathematical vegetables who work as journalists. This is largely the reason for journalists thinking that summarising puff-piece advocacy press releases is a major part of their job description. No media organisation should ever hire a journo who does have a semester or two of uni maths/stats.

      On the topic of GH, I would rather stick hot needles in my eyes, than direct more than 6 or 7 nanoseconds of neuronal activity to his porridge prose. Like that entire class of baby-boomer bourgeois Australians who stayed in Australia, GH is dumb, dumb, dumb. Dull as dish-water; appears to have a barely above average IQ; never displays systemic or networked thinking; and like Phillip Adams, abuses the shit out of his privileged monopolies on some of Australia’s grandest broadcasting platforms.

      He is unequivocally and incontrovertably a pox on our polity. As are his dunderhead peers from Ackerman to Adams to Cox, and many more. I would just love to let rip on that whole class of bozos, and how they pollute our polity, but blog owners have become extremely skittish lately over legal issues.

      Personally, I would love to be put in the dock and explain methodically and without fear of contradiction that these men – and the sheilah above – should be under house arrest, banned from dribbling their intellectually low-wattage or bulb-blew-decades-ago bullshit. AND SO SHOULD YOU!

      Tobias and Jeremy have set up shop here for the specific purpose of exposing these people. It’s more miss than hit in my book, but hey it’s a completely private space, on which they can engage in public debate however they wish. Hell, I join in, so they must be on to something.

      It is because GH presents as the type who sweated like a pig to get Credits at university that my focus is much more on Gans/Leigh. While both these guys are clearly in the stratosphere, compared to my mere pogying around on terra firma, this particular research buys into the same low-rent conceptual debates and values as the Philip Adams/Ackerman/Cox crew. Surely justice requires most attention be paid to this nonsensical intellectually immature, and total Freakonomics-gone-WRONG “research”.

      I only got involved to press the point that the issues Krugman has been arguing – dismally, I might add – are quite comfortable in the same common rooms as GH’s critique. Tjhe urinals misattribution is just a clumsy red herring. The fact that AL chose to focus on it stinks of bait and switch more than genuine intellectual intercourse.

      Problem here is that most contributors are on the Luvvie side of the Culture War, and so like drones swarm to protect their queen bee du jour without much/any independent assessment of the worthiness of their buzzing time.

    5. Philip

      John Greenfield said “…I, personally, find the mathematics they use quite accessible, not much more than high school, if even that, but absolutely ridiculous given the purposes of their research.”

      I’m with you on most of this John, but that’s not the point of the post. Like you I think the maths is easy enough to follow. I found myself nodding familiarly with the use of square-roots rather than logs and the application of the Binomial distribution. And I also have doubts about the methodology – particularly (as I read it) that any mention is sign of preference. Obviously a few searching questions need to be asked of Gans and Leigh. But Henderson didn’t ask them. He wasn’t attacking the details of the methodology, he was having a go at the very idea, the sheer impertinence of using statistics to quantify bias.

      I’m a little shaken by this. Didn’t he go to Xavier College? I’m sure the Jesuits wouldn’t have let him out without the maths sufficient to understand the Gans/Leigh paper. He expects his readers to wade through interminable details of the Labor split of the ’50s, the madness of Doc Evatt and the like, yet he turns up his nose at some fairly simple maths.

      It’s quite disturbing that someone who regards himself as a leading intellectual believes that the use of statistics to try and answer difficult questions is to be sneered at.

    6. DeanL

      I think there’s a much bigger issue in all this that is being overlooked. The conservatives claim that the media is biased against them. For something to be “biased” you have to show that the something has been diverted from its “natural state” by an “outside influence”. Can conservatives tell us what this outside influence is?

      The irony in all this is that, it’s the conservatives that are the ones who are actively *trying to biase the media*. By complaining that they don’t have the representation in the media; by conducting campaings against the ABC, etc, etc, they’re actively trying to change the media and divert it from its “natural state”. That is, they’re the ones trying to bias the media.

      And how can Henderson or Bolt or Albrechtsen or any other dyed-in-the-wool Coalition supporter possibly be considered objective enough to determine media bias? Bias relative to their perspective, certainly!

      Confessions is spot on at 25.

    7. Michael

      I thought it was Media Watch-Dog.

      Clearly I’m mistaken. It’s Media-Watch: Dog.

    8. confessions

      baldrick: IMO this is why any measure of media bias is completely useless unless you recruit a large representative sample of the population, show them several productions and ask them about any preceived bias. And just on the eg of emotion, there are a couple of reputable and widely used scales that measure wellbeing, whether in an individual diagnostic setting, or at a population level, which many experts dismiss as being subjective and therefore unreliable because they are based on the person’s own experiences without input from trained professionals. Damned if you do and damned if you don’t it seems.

      While any statistical analyses are not without their problems, what Henderson is holding up as a measure of bias (ie anything opposite to the howard govt) is quite simply rubbish and IMO far less rigorous than what Leigh/Gans have produced.

    9. confessions

      I look forward to Henderson’s next edition which I’m sure will include another attack on the Woman’s Weekly for including a *stand by your man* tell-all by Belinda Neal. Naturally Henderson is bound to conclude this is further proof that the magazine is biased towards the labor party. I mean, in the interest of *balance* surely the magazine should offer the same opportunity to a Liberal wife who also got cheated on by her husband, right?

    10. Alister

      John,

      You have no idea what you’re talking about. Possum is correct, and you, therefore, are not. Krugman is complaining about people who use elegant mathematical models in place of reality. Gans and Leigh used statistical analysis to describe reality. You may disagree with their basis of assessment; to line them up on the side that Krugman is disagreeing with is simply disingenuous.

    11. Tobias Ziegler

      A couple of posts back on this topic I noted my concerns about the methodology – Badrick’s first paragraph is in line with my problem. There was a US study using a similar methodology that I posted about previously because Andrew Bolt had used it as evidence of liberal bias in the American media. And I’m pretty much in line with Baldrick’s second paragraph as well – any quantitative analysis can’t be valid if the research design and measurement don’t give meaningful numbers to analyse.

      The problem is, Gerard Henderson didn’t confine his commentary to addressing issues with the “veracity of their findings”. After Gans & Leigh produced research findings he didn’t like, he went searching for anything he could use to tarnish their reputations – eventually arriving at the “urinal research”. He managed to cock up his attempt at mockery of Gans, which was hilarious, but the key issue is that he chose to attack the person rather than the research. And his correspondence with Andrew Leigh hasn’t been much better – completely failing to recognise the substantive errors he made, and inserting more snide insinuations about those who receive government funding without making a clear argument on the issues.

      I think that’s what galls me most of all about this – I think Henderson could actually advance a reasonable argument against his “opponents”, but he hasn’t managed to do it. I don’t know whether it’s because he doesn’t have the intellect to do so or because his bitter resentment of anyone who states a position against his worldview clouds his reasoning. Either way, it’s sad to watch.

    12. zombie mao for Alannah

      So,

      [no research that relies on formulas can be trusted]

      Well that all of Science, and , well, the entirety of Human Progress buggered then.

    13. baldrick

      Just a quick point from those formulae above – they use the number of mentions that said presenter gives regarding the coallition or otherwise. That is a totally useless means of evaluating so called “bias” as the number of mentions has nothing to do with whether the mentions were positive or negative. An assessment of whether they are positive or negative is in the eye of the beholder – ie. totally subjective. Therefore the calculations are rather meaningless IMO. Those “mentions” could be totally in favour or derogetory towards the subject discussed.

      Using maths to evaluate bias is like using maths to evaluate emotions without any indication of a persons subjective view of whether the emotion was positive or negative. It’s like asking “how many times have you been emotional” and then using the data to say that someone is depressed. What if you had been happy the entire time but had not been asked if the emotion was positive or negative – just whether you had been emotional.

      My commentry is based on your post material only – I haven’t read the report in it’s entirety. But if this is what they based their calculations on – I’m not putting much weight on the veracity of their findings.

    14. AR

      Gerontius “risks tarnishing his own reputation as a serious commentator,” with whom? Himself, alone, methinks.
      The position that, coz one is too thick to understand something, it’s ipso fact unclear is as old as the hills. The antiquity of an abuse/error is no justification for its continuation.
      I wasn’t here during the shameful meeja attacks on Barry Jone’s, soi disant, ‘noodle nation’ which seemed, even o/s, to have been an admirably clear (& simplified, presumably for the aforementioned jackals) example of an ideal, educated and integrated society.

    15. John Greenfield

      Dave G

      That is not correct. Krugman’s beef is explicitly with mathematics, not statistics. And even it were, it is precisely ‘well established methods’ that Krugman decries.

      Possum

      Krugman’s beef is most certainly not confined to financial pricing models – an area, which by the way, his comments (on for example, CAPM) way depart from the quality of his highly mathematical work in trade theory – but are primarily concerned with macro.

      Krugman is explicitly contemptuous of the mathematical Real Business Cycle Models, Inflation Models, New Keynesianism, and so on. He also takes stabs at micro inflation models that are linked to changes in the money supply. We could list countless Krugman rejections of mathematical economics, but that is surely a tired parlour game by now.

      Krugman’s essential point is that too many have equated impressive mathematics with proximity to economic “reality”. Most of Krugman’s criticisms in this area are more op-ed bloviating than the quality – highly mathematical – scholarship of his earlier days, for which he won the Nobel.

      Gerard Henderson has made some quite reasonable sneeers on the level of mathematics used to represent parliamentarians mentioning of Fairfax-Luvvie votes on “public intellectuals” as a variable determining the Left/Right sway in the media.

      The consistency of Henderson’s sneering with Krugman’s is obvious. The cogency of their respective takes on mathematics and its ab/use in Leigh/Gans-type Freakonomics flights of fancy are parallel.

      You will find it very difficult to support one, while vilifying the other. I, personally, find the mathematics they use quite accessible, not much more than high school, if even that, but absolutely ridiculous given the purposes of their research.

      It is precisely this type of economic modelling that Krugman is hoping will vanish that the entire economics profession of the last 30 years has subjected its graduate students, producing the catastrophic economic crisis he so melodramatically insists they have reaped.

      Personally, I think his ravings are those of a narcissistic Culture Warrior Luvvie, who would do well to check out some of his own course outlines before he starts throwing topological stones. 😉

    16. surlysimon

      And John, Gans and Leigh’s paper has bugger all to do with economics, unless I missed something and the ABC is actually a bank?

    17. Sisyphis

      “Gerard Henderson risks tarnishing his own reputation as a serious commentator.”

      That would indeed be “his own reputation”.

      A reputation shared by … ?

      (Disclosure. I went to lectures that Gerry also attended at Melb Uni in the mid 60’s. He looked like everybody’s somnambulant great uncle even then.)

    18. Matthew of Canberra

      I don’t see how anyone can determine “bias” objectively, unless the object openly admits it.

      I can see how one might plot a source’s position on certain issues, I can see how one might plot positive/negative stories about particular parties, and I can even see how one might correlate a particular source’s favor towards one side of politics or the other, as it’s manifested at any given time.

      The trouble is – none of that actually indicates _bias_. A slant to one side or the other might just indicate honest reporting.

      If the ABC were plonked down in, say, NAZI germany or fascist italy, and if it continued reporting according to its brief, I would imaging (hope?) they’d be pretty consistently at odds with the dominant political outlook in those countries at the time. To any mainstream italian or german observer they’d appear biased. According to any statistical analysis of their reporting they’d surely appear biased. In actual fact, they’d just be right. And there’s the rub.

      Any measurement of bias has to take into account a hypothetical set of fixed point norms, be they issues or political parties, and they’ll vary by time and place. Any USEFUL measurement of bias would have to take into account accuracy and honesty – and I’m not sure I’ve heard of a study yet that’s done that.

      Give it up, I reckon. Fox can produce opinion numbers to back up its claim of being fair and balanced. Who am I to argue with millions of americans?

      Lets just go back to arguing about which end of a boiled egg to crack.

    19. Possum Comitatus

      John,

      What Leigh and Gans have done here isn’t actually the type of econometric modeling that Krugman is in any way, shape or form, actually talking about.

      Krugman’s beef is with financial pricing models – which are effectively an exercise in describing variance and volatility.

      What Andrew and Josh have done is basic social science research – ask a question, design a methodology, gather the results as data, then throw them through the process of basic statistical analysis.

      This sort of Ordinary Least Squares regression used to analyse observable reality isnt remotely in the ballpark of using the type of threshold-GARCH/TARCH models on financial asset prices and running an auto-buy/sell regime on the basis of the forecasts – that’s the sort of thing Krugman is talking about.

      I despair sometimes when people take this weird view that maths is evil – you do realise that the alternative to empirical analysis is “making shit up”?

    20. Dave Gaukroger

      Professor Krugman’s argument about economists’ infatuation with statistics doesn’t really apply here though. Gans and Leigh aren’t trying to build a model out of theoretical data points, they are analysing observed data using well established statistical analysis.

    21. John Greenfield

      Dave. G.

      I am neither endorsing nor rejecting Gerard Henderson’s views, but his comments on the Leigh/Gans use of mathematics are particularly apt to Nobel Prize winning economist Professor Paul Krugman’s comments this past month. In his regular NYT column/blog, Krugman had this say about economists’ reliance on fancy mathematics:

      As I see it, the economics profession went astray because economists, as a group, mistook beauty, clad in impressive-looking mathematics, for truth..As memories of the [Great] Depression faded…the central cause of the profession’s failure was the desire for an all-encompassing, intellectually elegant approach that also gave economists a chance to show off their mathematical prowess.

      http://www.nytimes.com/2009/09/06/magazine/06Economic-t.html

      Regardless of what one otherwise thinks of Gerard’s argument, there is no doubt his broader point is at the epicentre of global debate on the role played by mathematically-focused economists in causing the GFC.

      So perhaps a dose of one’s own medicine, and a little more eye on the ball than the man might be appropriate here as well? 😉

    22. Philip

      I’m horrified by Henderson’s sneering at the use of statistics to detect media bias. So statistical analysis is something to be mocked? Gosh. How else should claims of media bias be tested if not with statistics? I’m still saucer-eyed at his comments.

    23. Dewgong

      Add comments 8 and 9 on this blog together, and I think that alone shoots Henderson’s credibility to pieces.

      {My tongue was very firmly planted in my cheek when I authored comment #9 – Dave}

    24. Dave Gaukroger

      Dear Jason Wilson,
      I refer to your comment on the weblog Pure Poison, dated September 25th, where you claim that I have invited Andrew Leigh out for lattes (plural). I would draw your attention to my electronic mail to Professor Leigh, as published in issue 29 of my Media Watch Dog, where I quite explicitly offer to, and I quote, “shout you a coffee”. Please take note that I have mentioned neither lattes, nor the purchase of more than one coffee.

      You might think about apologising to people who relied upon you for an accurate description of this situation. However, I would not request this – nor expect it to happen.

      Yours sincerely

      A media commentator in pursuit of accuracy.

    25. confessions

      From the MWD of today:

      [In Issue 26 and Issue 28 MWD discussed the paper written by Joshua Gans and Andrew Leigh which purported to demonstrate that there was no political bias in the Australian media, including the ABC.]

      Didn’t the LEigh/Gans paper show media bias at the ABC – towards the howard govt? Or am i thinking of some other paper?

    26. confessions

      Reading through Henerson’s reply there’s this:

      [Yet many journalists/commentators openly opposed the Howard Government on the invasion of Iraq, the non-ratification of the Kyoto Agreement, industrial relations (including waterfront) reform and national security (including the handling of the Haneef Case). Any paper on media bias in Australia during the Howard Government which does not cover such issues is intellectually shallow. ]

      Is he really saying that just because journos presented material contrary to the howard govt position on those issues, that constitutes bias against howard govt? And that that is a more rigorous investigation to be taking than the mathematical modelling? I think he should stick to parsing woman’s weekly.

    27. Catsidhe

      You may have noticed that there is a lot of “wasted taxpayer money in academia”, according to Gerard.

      You might think this reflects more on the limitations of Gerard’s understanding than on any actual deficit in research quality; I couldn’t possibly comment.

    28. Jason Wilson

      Did you see the latest MWD? Hendo has invited Andrew Leigh out for lattes.

    29. Dave Gaukroger

      I think that Henderson’s dismissive comments about the maths involved are rather silly, but his ad hominem attacks on Gans, and refusal to own up to them, are even more egregious failures.

    30. surlysimon

      So is Gerard say any research with mathematical equations in it is useless? Maybe we should tell the Nobel committee and John Forbes Nash.
      If Gerard is right why don’t we close the maths, physics, chemistry, engineering and computing departments of every university. Maybe Gerard wants us all to go back to doing “Classics” degrees (in Latin of course).
      The struggle to use mathematical equations to explain and understand is as old as learning.

    31. Tibor

      I think the key is to realise that ‘ijt’ in each formula should be read phonetically as ‘eejit’.

    32. Greg Weeks

      So anything that Gerard doesn’t understand is unworthy of being reported, something which merely “passes for scholarship” and further proof that academics are all wasting taxpayers’ money? What a very silly attitude to take.

    Advertisement

    Telling you what the others don't. FREE for 21 days.