Discount stores often introduce new merchandise, like a new kind of cookie, at a special low price. A psychologist predicted that this practice would actually lower sales. To test the idea, twenty-five pairs of stores were selected, matched according to location, sales volume, advertisements, and merchandise display.

All 50 stores introduced a new kind of cookie. For each match pair, one was chosen at random to do so at a special low price for two weeks before raising the price to the regular level; the other store in the pair introduced the cookie at the regular price. Total sales were computed...
In 18 of 25 pairs, the store which introduced the cookies at the regular price turned out to have sold more of them than the other store. Can this result be explained as chance variation? Or, does it support the psychologist's prediction?

I'm just wondering, why is the Standard error 2.5?