Stop those loose statistics!
I get annoyed when I hear or read statistics cited that can’t possibly be verified. I recently read in Steve Chapman’s column in the Chicago Tribune that 72 % of third-generation Hispanics living in the USA speak only English at home. I generally admire Chapman but his tossing out that statistic did annoy me. He didn’t say "according to polls" or use some such qualifier; it is virtually impossible to know the actual percentage: no one is going to contact every single third-generation Hispanic and get a response.
There are many instances of such stats being thoughtlessly thrown around, usually based on someone's estimates (or better, "guesstimates"). We hear that x million children go to bed hungry every night, the drug trade in the USA puts y dollars in the pockets of dealers, people with only a high school education make z dollars less in a lifetime of work than those who finish college, etc., etc.
But the most patently foolish one to me is it being reported, as an unqualified fact, that a certain percent of blacks/ Hispanics/ married women over 25/ white men between 25 and 40/ Catholics, etc. voted for one candidate (or party) or another in a national election.
I have heard from several sources that 7% of black people voted for George Bush in 2000, but the number increased to 13% in 2004. Clearly, no one can possibly know this. One could take the votes from precincts where the population is heavily black, but how about the increasing numbers of blacks who live in mostly white areas? One could use exit poll numbers, which are such a small and unscientific sample that they would have little validity as a proxy for the nation. Or how about small southern towns of a few thousand population which are about 50/50 white/black? There are probably only one or two polling places in those towns, making it impossible to know the numbers of the black vote.
To anyone who might say I am being picky, I say: if those users of grandiose "facts" aren't called to task, they will continue to stretch in making them up.
Please, columnists, politicians, activists, and all you others who toss these stats around. Stop it!
There are many instances of such stats being thoughtlessly thrown around, usually based on someone's estimates (or better, "guesstimates"). We hear that x million children go to bed hungry every night, the drug trade in the USA puts y dollars in the pockets of dealers, people with only a high school education make z dollars less in a lifetime of work than those who finish college, etc., etc.
But the most patently foolish one to me is it being reported, as an unqualified fact, that a certain percent of blacks/ Hispanics/ married women over 25/ white men between 25 and 40/ Catholics, etc. voted for one candidate (or party) or another in a national election.
I have heard from several sources that 7% of black people voted for George Bush in 2000, but the number increased to 13% in 2004. Clearly, no one can possibly know this. One could take the votes from precincts where the population is heavily black, but how about the increasing numbers of blacks who live in mostly white areas? One could use exit poll numbers, which are such a small and unscientific sample that they would have little validity as a proxy for the nation. Or how about small southern towns of a few thousand population which are about 50/50 white/black? There are probably only one or two polling places in those towns, making it impossible to know the numbers of the black vote.
To anyone who might say I am being picky, I say: if those users of grandiose "facts" aren't called to task, they will continue to stretch in making them up.
Please, columnists, politicians, activists, and all you others who toss these stats around. Stop it!
posted by Mycroft Watson at 2:44 PM
http://rpc.technorati.com/rpc/ping
0 Comments:
Post a Comment
<< Home