Some election year resolutions
By Doug Fisher
Happy News Year! It seems appropriate as we begin 2006 with a midterm
election, a controversial war and presidential hopefuls starting to jockey
for 2008 all guaranteed to produce a bountiful harvest of headlines.
As we head into this election year, however, let's resolve to pay closer
attention to a few things that are always sticking points.
Presumptive: Let's stop using this word when referring to those
expected to run or get their party's nomination. It's too easily confused
with the more negative "presumptuous." I regularly give a usage
test with a sentence like this: That was very presumptive/presumptuous
of you to think I'd do that. About half the class gets it wrong. Do we
want to use a word that has the potential of being so widely misunderstood?
There are plenty of good alternatives: expected, likely, etc.
Reform: We will be bombarded by tax reform, immigration reform,
education reform welfare reform and more. But "reform" is a
politician's friend. It instantly puts a positive spin on the most controversial
proposal. Is it reform if my welfare benefits will be reduced, or if I'm
a renter and my sales tax goes up to reduce someone else's property taxes? "Use
the word reform with care, the stylebook reminds us," writes
Paul R. Martin, the Wall Street Journal's language steward, "because
one person's idea of reform may be another person's idea of exploitation."
We're journalists, so let's not mindlessly adopt the politicians' terms.
Martin says "overhaul" and "revision" are more neutral.
Even simpler, try "changes."
Margin vs. ratio: Get these terms correct. Ratio refers to division
of the larger by the smaller. Margin is subtraction of the smaller from
the larger. If Filbert beats Cashew by 32,000 votes to 16,000, Filbert
won by a 2-1 ratio, not a margin. The margin is 16,000. And under AP style,
the hyphenation on the ratio numbers changes before and after the term:
It's a 2-1 ratio but a ratio of 2-to-1.
Report polls properly: First, let's stop calling them "snapshots" of
public opinion. We haven't frozen opinion in that moment; all we've done
is take one estimate of what opinion, a fluid concept, it is likely to
be. The value of polls comes from watching changes over time and trying
to discern trends.
Second, don't reach beyond what a poll actually measures. One large newspaper
reported that a poll on church attendance tells us we take religion seriously.
As one observer later wrote: "This poll doesn't measure how seriously
people take religion. It measures self-reported attendance at religious
services. They are not the same thing. Ask a preacher."
If it is a telephone poll, beware that excluding those without telephones
or with unlisted cell phone numbers can skew results. Online polls also
have serious potential problems.
Third, remember polls are affected by events around them and report that
context. Had you asked people how safe they felt on Sept. 10, 2001, and
again on Sept. 12, 2001, you would likely have found a big difference.
Fourth, know who is doing the poll and the potential bias.
Finally, let's correctly report margin of error and its meaning. Columbia
Journalism Review's "Campaign Desk" has labeled it journalists' "Margin
of Ignorance"
A poll is a pretty good estimate if put together following various rules,
but an estimate just the same. And one reported at the 95 percent confidence
level, the general standard, means that one time out of 20 any differences
could be entirely by chance. (Polls at less than 95 percent are generally
substandard for news use.)
Even those other 19 times, the numbers are only estimates and each
number can shift by the margin of error. For instance, if Filbert
has an estimated 52 percent support, with a 4 percent error margin,
it could be as high as 56 percent or as low as 48 percent. Some guidelines:
– If the numbers are within one margin of error, it's too close
to call. If Cashew's reported number was 48 percent, then Cashew actually
could be as high as 52 percent and Filbert as low as 48 percent.
– Within two margins of error, you still can't say one candidate
is winning. Let's say the poll reported Filbert at 53 percent and Cashew
at 47 percent. The uncertainty means Cashew still could have 51 percent
to 49 percent for Filbert. This is where journalists usually make their
mistake by saying one candidate is winning or that a candidate is "outside
the margin of error."
– At greater than two margins of error, one candidate "appears" to
be leading the other. Why "appears"? Remember, one time out
of 20, the numbers could result from mere chance and be totally off base.
– And margins of error are best reported as percents, not percentage
points.
If all this drives you nuts, then the link to the National
Council on Public Polls, www.ncpp.org/home.htm, should be in your browser. Its "20
Questions a Journalist Should Ask About Poll Results," will help
strengthen your reporting.
