Bad Science Theatre 2005
Ben Goldacre's latest Bad Science column deals with a recent pseudoscientific study of homeopathy. The study simply surveyed a large group of patients who received homeopathic treatments, asking them to evaluate their treatment on a seven point scale. The majority of patients claimed to be better after the treatment, but the study lacked a control group for comparison. Goldacre explains:
Is it bad not to have a control group? Yes. …[T]hey were looking at a lot of chronic cyclical conditions, or time-limited ones, like the menopause, where people get better with time. If 70% get better that's meaningless. 99.99% of people who get a graze to their knee will get better and 99% of people who get a cold will get better. It's not enough to know that they got better. We need to know if they got more better, or better faster, than people who weren't having homeopathy.
The study is an example of the regression fallacy at work, since many ill people will seek treatment only when they are very sick, and they will likely get better no matter what treatment they receive. This is why scientific medical studies need to have a control group: to compare its outcome to that of the experimental group. If the outcome of an experimental treatment is no better than that of a placebo, then there is no reason to think that it is effective. Other, better studies have shown that homeopathic remedies are no better than placebos. In fact, most homeopathic remedies are nothing but placebos, that is, plain water or sugar pills.
Source: Ben Goldacre, "Feeling better? Or is it all in the mind?", The Guardian, 11/26/2005
- "Homeopathic Remedies for Chronic Disease", National Library for Health, 11/24/2005
- Robert L. Park, "Alternative Medicine and the Laws of Physics", Skeptical Inquirer, September/October, 1997
- Spence, et al., "Homeopathic Treatment for Chronic Disease", The Journal of Alternative and Complementary Medicine, Volume 11, Number 5 (2005), pp. 793-798 (PDF)
A recent Italian documentary has ignited a controversy over the use of white phosphorus weapons by the U.S. military in the siege of Fallujah a year ago. The documentary claims that phosphorus was dropped indiscriminately on the city. The U.S. military isn't denying that phosphorus was used in Fallujah, only that it was used intentionally against innocent civilians. However, Jeff Englehart, a former U.S. soldier interviewed in the documentary as a witness to its use against civilians, claims that he was quoted out of context:
Englehart said Thursday that some of his statements were taken out of context. He maintained that he believes white phosphorus killed civilians, though he never saw anyone burned by it while in Fallujah. "I never personally did," he said. "That's where the…documentary misquoted me. They took that out of context." … Englehart was not directly involved in the Fallujah fight. He escorted a high-ranking officer to the front lines and watched the firestorm from atop a Humvee parked on the outskirts of the city. He said he heard U.S. forces call over the radio for "Willie Pete," the military vernacular for white phosphorus. "I know I heard it being called for on the radio. That's the only proof that I have, and I talked to a reconnaissance scout after the siege while we were still in Fallujah. He said they called in for white phosphorus on human targets," Englehart said. Englehart said an Italian reporter asked him during a five-hour interview in August whether he had seen innocent civilians killed in Iraq. Englehart said he had. Englehart said the producers of the Italian documentary took his answer to that question and edited it in after a question from a reporter about whether he had seen women and children killed by white phosphorus.
If what Englehart says is true, it's hard to see how this contextomy could have been an innocent error, which casts doubt on the trustworthiness of the reporters. On the other hand, if what he now says is not true, then that casts doubt on Englehart's trustworthiness. In either case, the documentary should be treated with some skepticism.
Source: Erin Emery, "Coloradan: Incendiary Killed Civilians", Denver Post, 11/18/2005
Via: Joshua Sharf, "Blowing Smoke?, Oh, That Liberal Media!, 11/18/2005
Poll Watch, Part 2
Here's an exact replay of the Associated Press poll story of ten days ago from USA Today:
WASHINGTON―Americans' views of President Bush and his trustworthiness have hit new lows, a downturn that could make it more difficult for him to push his legislative agenda and to boost Republican candidates in next year's congressional elections. … Bush's job-approval rating sank to a record 37%, down from a previous low of 39% a month ago.
Again, the margin of error is 3%, so the drop is within the margin, though the margin is not reported within the story itself, nor at the bottom in fine print; instead, you have to click on a link which takes you to a separate page containing such details. The details page is very informative, but most people probably won't find it, bother to read it, or understand it if they do. So, it's up to the reporter to accurately interpret the poll findings for the majority of readers.
These two stories taken together illustrate another problem with poll reporting: Different newspapers and news channels sponsor their own polls, and don't report the results of competing polls. However, while one poll may not show a statistically significant change, if most other polls taken about the same time show a similar change then that may be good evidence of a real shift in public opinion. The major media sources may not do such comparisons, but the Mystery Pollster does:
And then there were nine. With the latest survey out last night from Gallup/CNN/USAToday, we now have nine national surveys released this month and all nine show lowest ever job ratings for President Bush. … Compared to other surveys conducted just a few weeks ago, the most recent round represents another significant drop.
- Susan Page, "President's Ratings Hit New Low in Poll", USA Today, 11/14/2005
- "USA Today/CNN Gallup Poll", USA Today, 11/15/2005
Acknowledgment: Thanks to fallacy hunter Michael Koplow for submitting this example.
One of the oldest movie blurbs from my files is from a 1988 movie called "Cop", and the blurb was: "Cop: A Thriller!" The blurb was simply the title of the review, with an unjustified exclamation point added. This is an example of a seldom used trick of the blurb writer: not unjustified exclamation points, which are used all the time, but quoting the review title. The trouble with using the title of a review in a blurb is that the titles of newspaper stories are usually added by an editor, rather than the reviewer. Moreover, in the case of "Cop", the term "thriller" was only used to describe the type of movie reviewed; adding an exclamation point makes it sound as if the reviewer were praising the movie for being thrilling.
I've noted a few other instances of this sort of contextomy recently, but here's one from an ad for the current movie "Chicken Little":
"YOU'LL FALL FOR 'CHICKEN LITTLE'"
THE WASHINGTON POST
Interestingly, this was only one of five blurbs in the ad, and each of the other blurbs was attributed to both the publication and the reviewer. I suspect that this was the only blurb attributed solely to the publication because the ad writer knew that the title probably wasn't written by Stephen Hunter, the reviewer. If so, why quote an unknown editor who may not have even see the movie, but is just trying to come up with a catchy title for the review?
- Ad for "Chicken Little", Go!, 11/11/2005, p. 17
- Stephen Hunter, "You'll Fall for 'Chicken Little'", The Washington Post, 11/4/2005
The Thing Without Feathers
"One man illustrated proper logic with this syllogism:
All men are mortal.
Socrates is mortal.
Therefore, Socrates is a man.
I raised my hand. 'Birds are mortal too, aren't they?' I asked, hoping he would correct his error.
'Yes,' our teacher agreed.
'So Socrates could be a bird?'
He smiled benignly. 'No. Socrates doesn't have feathers.'"
Source: M. Garrett Bauman, "But Can You Teach?", Chronicle of Higher Education, 10/28/2005
Pall Mall cigarettes were once advertised as "A longer, finer and milder smoke." The question is, longer, finer and milder than what? … "Magnavox gives you more" doesn't tell you what you get more of. More value? More television? More than they gave you before? It sounds nice, but it means nothing, until you fill in the claim with your own words, the words the advertiser didn't use. Since each of us fills in the claim differently, the ad and product can become all things to all people, and not promise a single thing. …
Some years ago, Ford's advertisements proclaimed "Ford LTD―700% quieter." Now, what do you think Ford was claiming with these unfinished words? What was the Ford LTD quieter than? A Cadillac? A Mercedes Benz? A BMW? Well, when the FTC asked Ford to substantiate this unfinished claim, Ford replied that it meant that the inside of the LTD was 700% quieter than the outside. How did you finish those unfinished words when you first read them? Did you even come close to Ford's meaning?
Source: William Lutz, Doublespeak: From "Revenue Enhancement" to "Terminal Living", How Government, Business, Advertisers, and Others Use Language to Deceive You (Harper & Row, 1989), pp. 95-96.
Pandas married with city-wide ceremony in Thailand
It's the first step onto the slippery slope to gay panda marriage!
Here's how an Associated Press report on a recent poll begins:
WASHINGTON―President Bush's public support has eroded to its lowest level yet, with the Iraq war dragging on, a top White House aide facing felony charges and the White House rushing to replace a failed Supreme Court nominee.
That certainly sounds bad, but the third paragraph adds:
A new AP-Ipsos poll found the president's approval rating was at 37 percent, compared with 39 percent a month ago.
If you're a savvy poll watcher, your alarm bell should be ringing now. According to the poll, the approval rating has dropped two points in a month; but almost no polls have margins of error (MoEs) as low as two points, and most national polls have MoEs of three points or higher. Of course, reporters usually bury the numbers near the end of the article; sure enough, the sixteenth paragraph reads:
The poll was conducted by telephone Oct. 31-Nov. 2 among 1,006 adults nationwide. The margin on [sic] sampling error is plus or minus 3 percentage points.
So, more accurately, the change in approval rating in the last month is within the margin of sampling error, despite "the Iraq war dragging on, a top White House aide facing felony charges and the White House rushing to replace a failed Supreme Court nominee." Notice, also, the use of juxtaposition and the word "with" in that first paragraph to suggest that the insignificant decline in the approval rating was caused by these three unfortunate events. So, there are two fallacies in this report:
- A statistically insignificant change is treated as an event: fake precision;
- The nonevent is explained as due to other events which occurred at the same time, despite the fact that there is no other evidence given that they are responsible: cum hoc.
I would like to be able to end this entry by saying that my approval rating of the AP has dropped to a new low, but that's not true because I don't expect any better.
Source: Will Lester, "Public Support for Bush Slips to New Low", Associated Press, 11/4/2005
Resource: Fallacy Watch: How to Read a Poll
Q&A: Contradictory Premisses
Melanie Lounds sent an email recently about "the fallacy of contradictory premisses", noting that she could not find it listed in the Fallacy Files, and asking whether it was listed under another name. Though some other works on fallacies―such as Madsen Pirie's Book of the Fallacy―include a fallacy by that name, it isn't to be found in these files under that or any other name.
What are contradictory premisses? Two propositions are contradictory when they must have opposite truth-values, that is, one must be true and the other false; for instance, "Pat is pregnant" and "Pat is not pregnant" are contradictory. So, an argument with contradictory premisses would, strictly speaking, have to be one with only two premisses. However, I suspect that the fallacy would be said to be committed by an argument with a larger number of premisses when some pair of them is contradictory.
What's wrong with contradictory premisses? Any set of propositions which contains a pair of contradictories―including the pair itself―will be a set of contrary propositions, that is, at least one member of the set must be false. For instance, "Pat is pregnant" and "Pat is male" are contrary propositions, but not contradictory since they might both be false. So, any argument with contrary premisses will be unsound. For this reason, the supposed fallacy of "contradictory" premisses could be a subfallacy of "the fallacy of contrary premisses".
Why isn't either of these fallacies listed? Arguing from contradictory premisses is certainly a logically bad thing to do, but I doubt that it's a very common occurrence. People are likely to notice that propositions contradict one another, so it is unlikely that they will mistakenly reason from them, or think that an audience would accept such an argument. This is certainly true when one of the contradictory propositions is the negation of the other, though contradictoriness is harder to spot when the propositions are more logically complex. While it's possible that contradictoriness among a large set of premisses might go unnoticed, I can point to no real-life examples of arguments which commit this mistake. So, the supposed fallacy would fail to meet one criterion of a logical fallacy, namely, that the error be a common one.
However, it is more likely that contrary, but not contradictory, premisses would go unnoticed, so there is a possibility that reasoning from contrary premisses may be a more common mistake. However, I still have seen no uncontrived examples. So, I will end this answer with my usual plea that if you know of any such example, please send it to me!
Resource: Madsen Pirie, "Contradictory Premises", The Book of the Fallacy