One of the gimmicks he used to market his company was that he developed his sample by pairing all the 3-digit phone prefixes in the market with final four digits drawn from a random number table. Compared to other survey companies, who pulled their sample numbers straight out of the phone directory, our phone operators blew through more not-in-service numbers, but the client wouldn't care about that; they just cared that we reached more new numbers and unlisted numbers than our competitors. New numbers and unlisted numbers, so the argument went, skewed toward the holy grail: young, mobile, professional listeners.
This morning, DailyKOS regular Meteor Blades jumps into an interesting discussion that's been brewing in the background for a few years now:
whether pollsters are undercounting Barack Obama's support by millions of voters because they are failing to survey cellphone-only users, a growing portion of the population, especially the population of young adults most likely to have only cellphones and which showed a strong preference in the primaries and caucuses for the Illinois Senator.
Reviewing a Salon.com article on the topic this morning, he concludes:
If Maslin and Brown are right, pollsters who continue to take the easy path this election year could wake up red-faced on the morning of November 5.
Some day, if I live long enough, I'll probably read an article speculating that polling results may be undersampling one candidate's supporters (Meghan McCain's?) because they aren't reaching the younger voters who get their messages via subcutaneous transponders instead of the old-style 3-D holovids everyone else has.