The celebrated radio quiz show star says:

There’s this study done by the Pew Research Center and Smithsonian Magazine . . . they called up one thousand and one Americans. I do not understand why it is a thousand and one rather than just a thousand. Maybe a thousand and one just seemed sexier or something. . . .

I think I know the answer to this one! The survey may well have aimed for 1000 people, but you can’t know ahead of time exactly how many people will respond. They call people, leave messages, call back, call back again, etc. The exact number of people who end up in the survey is a random variable.

1. Dan Riley says:

They called up a lot more than 1001 people to get those 1001 responses. If Ira had been more precise in his statement, the answer would have been obvious.

2. MikeM says:

Wuffo a picture of J. D. Salinger?

• Roy says:

Many of J.D. Salinger’s stories centered on the Glass family, especially Seymour Glass, and one of the facts about them in the stories is they participated in a radio quiz show for kids. If you Google you will find out that there really was such a show, though not with the Glass family. So this reference has two levels of indirection. Also if you have followed this blog for awhile, Andrew competed in math competitions in his youth.

-Roy

3. Ethan Bolker says:

1001 = 7 * 11 * 13 is a very sexy number. If you have a big number you can alternately add and subtract three digit blocks and quickly determine the remainder when divided by each of 7, 11 and 13.

• zbicyclist says:

Raising interesting questions about “1001 Arabian Nights”. I’ve always thought the “1001” in that title was very cool.

But now Andrew is telling me Scheherazade might have just had a target to aim at. Sometimes

4. Gould says:

yup, thousand & one sounds really oddball — how hard is it to randomly telephone people evenly.

but the OCD Pew pollsters get hung up on all that academic sampling trivia… and dial thousands of times to achieve their typical 5-15% response rate on their bona fide random convenience sample (BFRCS)

• Andrew says:

Gould:

No, you don’t understand. They don’t survey people sequentially. They call a lot of people at once, with a rough target of 1000 respondents. They don’t know ahead of time how many responses they’ll get. Hence, the sample size is a random variable.

And, regarding your OCD comment: Damn straight I want pollsters to care about nonresponse. That’s their job. It’s not “OCD” to want to do a good job. It’s called professionalism.

• Todd Hartman says:

Pollsters set targets for completed interviews based upon the size of the population, subsampling considerations, and so on. Often that number is 1,000 because it’s a good tradeoff between an acceptable margin of sampling error and cost. To get that 1,000 in a reasonable timeframe, the survey organization will have dozens if not hundreds of interviewers working simultaneously. As they get closer to their targeted number of completed surveys, they will slowly pull interviewers off the project and reallocate their time to other things. However, there will still be multiple interviews going on at the same time, and most systems won’t prevent new calls until the target is reached. For example, if 5 interviews are being conducted with 999 completes in the system, then the final total will be 1,004. No reason to throw out good data just to have an even number.

5. 1001 is a palindrome, and it converts from binary to the highest single decimal digit. Besides, there’s just something captivating about that 1. “One Thousand Nights” would’ve been remaindered a couple of weeks after publication. And would anyone today still be talking about “2000” (full title “2000 a Space Odysssey”)?