Previous Month | RSS/XML | Current Weblog | Next Month

April 21st, 2004 (Permalink)

Poll Watch

Here's how CNN reports their latest poll results:

WASHINGTON (CNN) -- President Bush increased his lead over Sen. John Kerry in a CNN/USA Today/Gallup Poll released Monday…

Bush led Kerry, the presumptive Democratic nominee, 51 percent to 46 percent in the survey of likely voters, which was conducted Friday through Sunday. The survey interviewed 1,003 adults, including a subsample of 767 respondents deemed most likely to vote in November…

The previous CNN/USA Today/Gallup Poll, conducted April 5-8, showed Bush leading Kerry 48 percent to 45 percent among likely voters…

With the current survey's margin of error of plus or minus 4 percentage points among likely voters, Bush and Kerry remain locked in a dead heat more than six months before the November election.

So which is it: Is Bush leading, as the headline and first paragraph say, or are Bush and Kerry "locked in a dead heat"? Given that the margin of error for likely voters is + or - 4%, Bush's "lead" in both the current and previous poll is within that margin. Thus, the only "lead" that Bush has is a statistically insignificant one. The two presidential candidates have been tied in the polls for at least three months.

I often wonder whether reporters don't understand the margin of error, or add the phony "horse race" details in order to make a dull story more interesting. The contradiction in this story suggests the latter.


Resource: Ana Marie Cox, "Poll Analysis Roundup", STATS, 4/21/2004

Update (5/12/2004): Eugene Volokh has pointed out that CNN's subsequent reporting on polls continues to ignore the margin of error, and he has a follow-up from a reader who gives further evidence that this is done knowingly in the interest of more exciting stories.


April 19th, 2004 (Permalink)

What's New?

I've added a "wild" new example to the examples file, as well as adding an explanation of "wildness" to the introduction.

Source: Stalking the Wild Fallacy

April 18th, 2004 (Permalink)

Blurb Watch

The movie ad writers are up to their old trick of quoting a part of a review that refers to an actor as if it referred to the movie as a whole. The following blurb is from the ad for the new movie The Girl Next Door:

"Sexy and Engaging."
A. O. Scott, The New York Times

Reading the review, however, reveals that these words refer to the actress Elisha Cuthbert, who plays the title character, rather than the movie itself, which is panned.

Source: A. O. Scott, "The Perfect Girl, Except for the Résumé", New York Times, 4/9/2004

Fallacy: Quoting Out of Context

April 16th, 2004 (Permalink)

Not Ready for "Primetime"

ABC News' "Primetime" program Thursday night had two reports on supposedly supernatural topics, one on reincarnation and the other on a "psychic" detective named Carla Baron, who claims to be able to help the police find missing persons. In the report, Baron leads a police officer and the television crew on a snipe hunt to find a missing college student.

The way that "psychic" detectives typically work is by making many vague and general statements, thus maximizing the chance of some random match between what they say and what is discovered. The "Primetime" program illustrated this technique, though inadvertently, I suspect:

"In the Song case, Baron accompanied Quiñones and Sprinkle on a drive around eastern Pennsylvania. She said she was looking for some railroad tracks, water and 'some sort of generator.'

"At one point, she said she got a painful 'vision' with some vital clues. 'I saw rocks on the bank,' she said. 'There is going to be a part of the bank, that has rocks…bigger stones…It's not just water, right at the bank, right at trees.'"

How likely is it that the student will be found somewhere near railroad tracks, water, trees, and rocks? How far away would they have to be not to count as "near"? The most specific thing in this description is the generator, but the "some sort of" qualifier makes it less so. Would the generator in a car engine count? Wherever the student eventually turns up, you can bet that Baron will claim success.

On the program, despite Baron's "help", no sign was found of the missing student.

Source: "Help From the Other Side: To Find Missing College Student, ‘Psychic Detective’ Enlisted", ABC News, 4/15/2004

Fallacy: Vagueness

April 9th, 2004 (Permalink)

Head Line

Jane Fonda to teens: Use head to avoid pregnancy

Source: "Headlines", The Tonight Show

Fallacy: Equivocation

April 8th, 2004 (Permalink)


Q: "I don't see any reference in your website to the phenomenon of the Circular Reference, which is related to a Circular Argument. Yet your glossary includes an excellent example, as I found when I tried to find out what the term 'distributed' meant. Perhaps you could elucidate: what is a 'distributed' term? It doesn't really help to give examples wherein one or both, or neither term is 'distributed' if you don't make clear what it is about the term that makes it so, or not so."—Paul & Sherry Stone

A: There is no actual circularity in the definition of "distributed" given in the Glossary, rather it is defined by the positions of terms in categorical propositions. This may seem like an unsatisfactory definition since it doesn't explain why these particular positions in these specific propositions are considered "distributed", nor why the English word "distributed" was chosen for this property.

I intentionally didn't offer a more informative definition for two reasons:

  1. It isn't needed in order to use the concept to detect fallacies in syllogisms: all you need to know is which terms are distributed and which are not, and you can tell this from their positions.
  2. The usual textbook definition of "distribution" is vague and difficult to understand, so that it's not worth giving.

However, in researching your question, I discovered a clear and more understandable definition:

A term in a categorical proposition is distributed if and only if the proposition implies every proposition that results from replacing the term with a more specific term.

For instance, the subject term "mammals" in "all mammals are animals" is distributed because it implies "all cats are animals", "all dogs are animals", "all humans are animals", etc. In contrast, the predicate term "animals" is not distributed because the proposition doesn't imply that all mammals are cats.

Thanks for the difficult question, Paul and Sherry.

Source: Thomas Mautner, A Dictionary of Philosophy (Blackwell, 1996), p. 112.


So far, no one has solved this month's puzzle and claimed the prize. So, I'm extending the deadline until the 14th (do the puzzle instead of your taxes!) and giving the following hint: The judge let the guilty person go because the judge did not want to punish an innocent person. Also, keep in mind that the prize will go to the best answer received by the deadline.

Source: April Fool's Puzzle

April 6th, 2004 (Permalink)

What's New?

April 3rd, 2004 (Permalink)

Orwell on Blurbmeisters

"Publishers have got to live, like anyone else, and you cannot blame them for advertising their wares, but the truly shameful feature of literary life before the war was the blurring of the distinction between advertisement and criticism. A number of the so-called reviewers, and especially the best-known ones, were simply blurb writers. The ‘screaming’ advertisement started some time in the nineteen-twenties, and as the competition to take up as much space and use as many superlatives as possible became fiercer, publishers’ advertisements grew to be an important source of revenue to a number of papers. The literary pages of several well-known papers were practically owned by a handful of publishers, who had their quislings planted in all the important jobs. These wretches churned forth their praise—‘masterpiece’, ‘brilliant’, ‘unforgettable’ and so forth—like so many mechanical pianos. A book coming from the right publishers could be absolutely certain not only of favourable reviews, but of being placed on the ‘recommended’ list which industrious book borrowers would cut out and take to the library the next day. …

"Even reputable literary papers could not afford to disregard their advertisers altogether. It was quite usual to send a book to a reviewer with some such formula as, ‘Review this book if it seems any good. If not, send it back. We don’t think it’s worthwhile to print simply damning reviews.’

"Naturally, a person to whom the guinea or so that he gets for the review means next week’s rent is not going to send the book back. He can be counted on to find something to praise, whatever his private opinion of the book may be.

"In America even the pretence that hack reviewers read the books they are paid to criticize has been partially abandoned. Publishers, or some publishers, send out with review copies a short synopsis telling the reviewer what to say. Once, in the case of a novel of my own, they mis-spelt the name of one of the characters. The same mis-spelling turned up in review after review. The so-called critics had not even glanced into the book—which, nevertheless, most of them were boosting to the skies."

Source: George Orwell, "As I Please", Tribune, 6/9/1944

April 2nd, 2004 (Permalink)

Check it Out


Update (4/4/2004): Here's another case of ignoring the margin of error in order to "manufacture" news. Ana Marie Cox of the Statistical Assessment Service debunks reporting of a recent Gallup poll.

Source: Ana Marie Cox, "Handicapped by Plus or Minus Three Points", Statistical Assessment Service, 3/30/2004

April 1st, 2004 (Permalink)

April Fool's Puzzle

The prize for the solution of this month's puzzle is a copy of Irving Copi & Carl Cohen's classic Introduction to Logic—specifically, the eighth edition of 1990. Here is the puzzle:

Two people were on trial for the same crime. The jury found one guilty but the other not guilty. The judge addressed the guilty defendant as follows: "This is the most unusual and difficult case that I have ever presided over. Despite the fact that your guilt has been established to the jury's satisfaction beyond a reasonable doubt, justice requires me to release you. You are free to go." Why did the judge let the guilty defendant go?

This month the prize will go to the best answer received by April 7th. The answers will be judged based on both the solution to the puzzle and a logical explanation of why the puzzle is tricky. Send your answer to the The Fallacy Files.

Update (4/15/2004): We have a winner! I judged the best entry to be the following one from Tim Davenport:

"The defendants are Siamese twins. The only penalty available to the judge was imprisonment, not a fine. The judge had no choice but to let the guilty twin go rather than force the twin found not guilty to serve the sentence with his or her sibling."

Runner-up is Timothy Roscoe Carter with the following clever solution:

"The crime of which the two persons were accused was conspiracy, and the state argued that the two conspired with each other. In some jurisdictions, such as California, a conspiracy only occurs if there is a mutual agreement among at least two persons. If this case was in such a jurisdiction, under the facts of the puzzle, the judge would have to let the 'guilty' person go. The jury may have even rightly decided, on the facts, that there was reasonable doubt for one defendant, but not the other. Even so, the fact that there was only one 'conspirator' means there was, legally, no conspiracy, and therefore the 'guilty' defendant was innocent because no crime had occurred."

Source: Raymond Smullyan, What is the name of this book?: The riddle of Dracula and other logical puzzles (Prentice-Hall, 1978), p. 10.

Previous Month | RSS/XML | Current Weblog | Next Month