Some examples of pharm influence are outrageous, while most are comparatively benign. And while there's evidence that free lunches and other perks that drug companies provide influence prescription-writing, I've argued in the past that no studies show these perks lead to actual harm to patients (yet).
I've wondered what studies have been done on bias in other fields -- say, the media. Because there's no shortage of opinions on liberal bias, or Fox News bashing, but precious little data. More underpowered contentions were made this weekend. The Ombudsman for the New York Times, Daniel Okrent, concluded after months of analyzing the Old Gray Lady's presidential coverage there is no systematic bias toward either candidate:
If there's a commissariat at The Times ordering up coverage to help or hurt a specific candidate, it's doing a lousy job; close reading shows bruises administered to each (and free passes handed out) in a pattern adapted from Jackson Pollock. Many people want to know why the other guy's position is in the first paragraph of a story, and their side doesn't weigh in until the sixth; they don't notice when it's the other way around. Sherrie Sutton of Manhattan, who describes herself as "the only possible Bush vote on the Upper West Side," asked why Times headlines consistently use "attack" when Republicans criticize Democrats, but not when Democrats criticize Republicans. Intrigued, my associate, Arthur Bovino, determined that in the past year, headlined Republicans attacked Democrats 12 times and Democrats attacked Republicans 22 times. Ms. Sutton replied: "Statistics don't lie, and you've got 'em. Interesting, that in the face of facts, I could still feel unsatisfied that campaign coverage by the NYTimes is balanced."
This is something, but hardly the in-depth analysis many bloggers were calling for. Okrent goes on to make a broadside against fact-checking blogs that urge readers to complain to the Times (the examples he picks of eggregious complaints, however, are effectively breathtaking).
Blogger Mickey Kaus counters:
Okrent denies that "because charges of bias come from both liberals and conservatives, the paper must therefore be doing things right"--but that doesn't stop him from using complaints from left and right to balance each other out and conclude the Times isn't "systematically biased toward either candidate." Might there be other large systemic biases, or biases within various departments?
There are some broader, systematic studies out there. A Yale group report on media bias was made available earlier this year. It's summarized nicely in the Rocky Mountain News:
Two researchers have ... come up with a measure of media bias that doesn't depend on journalists' own perceptions of where they fit on the political spectrum, or on subjective judgments about the philosophical orientation of think tanks. Tim Groseclose, of UCLA and Stanford, and Jeff Milyo of the University of Chicago used data comparing which think tanks various politicians liked to quote and which think tanks various media outlets liked to quote in their news stories to estimate two ADA scores for each media outlet in the study, one based on the number of times a think tank was cited, and the other on the length of the citation...
...The predominance of liberals (however identified) in major media is well-documented, but there remains a great deal of controversy over how much that fact influences news reporting (this analysis looks only at news reports, not editorials, reviews or letters to the editor). Most journalists I know say they work hard to keep their personal views out of their news reporting (again, excepting people like me who are supposed to be expressing opinions). And most of them, I'm sure, sincerely believe they succeed. This is evidence that what they succeed best at is sounding like Democrats.
Obviously there are problems with this method, and the authors address some of them. Other criticisms are discussed here. But the point was to develop an objective measure of ideological bias, which is not easy. Most people knows bias when they see it (triggering an involuntary roll of the eyes after a particular phrase or quote from Fox News or the NYTimes). But how to quantify it, and compare to some standard (in this case, Congress)? It's a start.
But comparing journalists' attitude toward bias to that of physicians is instructive The bias of reporters, left or right, isn't really financially motivated. It's about perception from colleagues, maintaining access to sources, and promoting a specific worldview. It's also partly about reader expectations, though polls suggest readers expect less and less.
Doctors, on the other hand, have a lot to lose by giving up pharmaceutical perks. You can count the value of free meals, conference fees, office supplies, etc. Plus, the doctor loses out on new drug developments (slanted as that might be, docs are usually good enough to read between the lines, and the reps can't outright lie). Finally, the doctor's patient pool stands to lose, as many clinics have an in-house pharmacy made up of free drug company samples.
So that's what each professional group -- doctors and reporters -- stands to gain or lose from their conflicts of interest. From that, you'd think it'd be straightforward to guess which group is acting responsibly, reflecting on what goes on, asking the tough questions, and proposing action on behalf of their consituents.
But the truth is surprising. Journalists, with less to lose, tend to deny bias. Medical professionals, with so much more at stake, are leading the charge against the influence of pharmaceuticals.
Sure, the problem of media bias is less likely to kill people (at least, not directly), and solutions aren't as easy to imagine (a "fair and balanced" review board?). But biased reporters are as repugnant to the code of journalism as corrupt doctors are to the oath of Hippocrates.
I just find it gratifying that doctors are trying to protect their patients from the possibility of bias, at their own measurable expense, while many journalists continue to insist they're not biased at all.