rewarding nakedness

From CNN.com. First, a late entry to the superfluous statistic of the year contest:

"'Paris Hilton' is the No. 1 fake name used by people calling for pizza deliveries, according to a survey of Domino's Pizza drivers in Washington, D.C., released Monday by the pizza delivery chain. And 38 percent of those using the name of the socialite model ordered pepperoni topping.

Is 38% a lot? What is the non-Paris-Hilton % of pepperoni pizza orders? Can someone call Domino's for me on this? Rob? Is there some sly innuendo here I'm missing? Am I being desperately unhip yet again?

But then, a couple of paragraphs later, a more intriguing statistic:

According to the survey of 630 drivers, nine percent of people who answer the door in the nude tip more than 20 percent, compared with 2 percent of people in pajamas."

Say this difference was indeed real some artifact of the delivery-driver-survey-method. Hypotheses? I suppose the first question is whether one thinks this is a between-person or within-person difference: that is, is it that people who are more likely to order a pizza and answer a door nude are more likely to be big tippers (more generous even during those times when they are wearing clothes), or that people are bigger tippers when in the situation where they are answering the door nude (as opposed to occasions when they are clothed). If the latter, is it because when somebody answers the door nude they are more likely to be in a good mood? Or, are they less concerned with getting change back since they don't have anywhere to put it?

Update: Rob replies:

Here are 1998 stats on Domino's pizza toppings:

The most commonly used toppings (excluding cheese) were:
(1) Pepperoni (24 million pounds)
(2) Italian Sausage (13 million pounds)
(3) Mushrooms (9.3 million pounds)
(4) Ham (7.6 million pounds)
(5) Hamburger (6.7 million pounds)
(6) Onions (4.7 million pounds)
(7) Bacon (2.3 million pounds)

These seven toppings all add up to 67.6 million pounds. If we add an
"Other" category and throw in a couple more million pounds to round up to
70 million pounds, and we assume that 1998 is like any other year, and we
make the additional assumption that one order of any given topping will
equal the same weight, and then we also assume that the likelihood of
ordering an "extra" amount of any topping is equal across toppings, and
then we make one final assumption that any other problem associated with
making an estimate based on the above data magically disappears, then I
would guess that of all Domino's pizzas ordered, 34.3% of them have
pepperoni on them.

So what does this mean? Well, if we can assume that Washington D.C. is
like any other city (the city used for the "Paris Hilton study"), it means
that people who use the name Paris Hilton when ordering Domino's Pizza are
more likely to order pepperoni on their pizza (38% of them did this) than
those who do not use the name Paris Hilton.

Yes, but is the difference significant?

To me it is.

No, is the difference statistically significant?

Oh, I don't really know about that.

One methodological note: if we don't round up to 70 million pounds, and we
just use the 67.6 million pounds as the denominator, we find that the
percentage is 35.5%, still below the Paris Hilton threshold. Conversely,
if we inflate the denominator to 75 million pounds, we get 32.0%, and then
we really do need to start getting into the "Paris Hilton/Pepperoni"
subculture theories, not to mention putting together a separate ASA
section devoted to this topic...

Thanks Rob!