Polls, margins and majorities

This article was originally on a blog post platform and may be missing photos, graphics or links. See About archive blog posts.

Some readers were confused or accused The Times of bias after reading the May 23 poll story presenting results of Californians’ opinions about same-sex marriage.

Jim Nores of Santa Clarita referred to the headline, ‘Californians Barely Reject Gay Marriage,’ when he wrote, ‘It is not until you add the numbers yourself do you find that numbers are actually 52% to 41% with 7% don’t know! That is a spread of 11 points! Please tell me how the word ‘barely’ can be used to describe those numbers. The headlines and first paragraph are clearly intended to mislead the reader into thinking that the numbers were a lot closer than they really were.’ (In some editions, the headline read, ‘Californians Slimly Reject Gay Marriage,’ and online the word was ‘narrowly.’)

Nores was talking about results for the question, “Do you approve or disapprove of the California Supreme Court’s decision last week to allow same-sex marriage in California?” The results among voters and nonvoters combined:

Strongly approve: 29%
Somewhat approve: 12%
Somewhat disapprove: 10%
Strongly disapprove: 42%
Don’t know: 7%

In a follow-up note, Nores added, ‘I am sure if [Barack] Obama beats Hillary [Clinton] by the same margin, the descriptor would be ‘landslide,’ not ‘barely.’’

Of another poll findings, several readers thought -- as one put it -- that it was an ‘obvious example of bias’ to cite as a ‘bare majority’ what they saw as a 19-point lead among those who want to outlaw gay marriage.

Those readers were asking about a different question, asked of registered voters: “A proposed amendment to the state’s Constitution that may appear on the November ballot would reverse the court’s decision and state that marriage is only between a man and a woman. If the election were held today, would you vote for or against the amendment?”

The results: 54% for; 35% against.

Other readers, too, were stumped at why the 54% to 35% was ‘barely.’

As editors and the reporter emphasized afterward, the word ‘barely’ did not refer to the margin, it referred to the majority -- ‘a different statistical measurement,’ as state politics editor Cathleen Decker, who wrote the story, put it. And, editors point out, in a poll in which the margin of error is plus or minus 4 percentage points, 54% is a bare majority.

But as evidenced by some of the notes that came in, when it comes to polls, people usually consider the margin between the two figures to be key.

Here’s what the first line of the article said: ‘By bare majorities, Californians reject the state Supreme Court’s decision to allow same-sex marriages and back a proposed constitutional amendment aimed at the November ballot that would outlaw such unions, a Los Angeles Times/KTLA poll has found.’

But for some readers the confusion came in the second paragraph, which said, ‘The poll suggests the outcome of the proposed amendment is far from certain. Overall, it was leading 54% to 35% among registered voters. But because ballot measures on controversial topics often lose support during the course of a campaign, strategists typically want to start out well above the 50% support level.’

Why did the poll story focus on the level of support, rather than the margin?

California Editor David Lauter explains, ‘For ballot measures, the key question is: Is it over 50? The telling example would be a ballot measure that leads 40-20 with 40% undecided. We’d feel in that case that the measure is in potential trouble despite the large margin because it would be substantially under 50. Sometimes, of course, ballot measures gain support during a campaign. That hypothetical measure with 40% support certainly wouldn’t be doomed. But experience has shown that voters need to be persuaded to vote yes on things: their default position is no. Undecided voters tend, over time, to turn into ‘no’ votes, and measures often lose support as time goes on. So we generally focus on what percentage of voters are in the ‘yes’ camp.’

Times Poll Director Susan Pinkus repeated those facts in another e-mail to a reader then gave a specific example: ‘The amendment has to pass by 50% -- so 54% is not a big majority among registered voters and smaller even among all Californians -- 52%.’ Referring to the earlier, voter-approved measure that said: ‘Only marriage between a man and a woman is valid or recognized in California,’ Pinkus continued, ‘Prop. 22 in 2000 passed by 61%. So what we were trying to say is it could go down to defeat when you look at the different demographics (some would vote yes, especially the youngest voters, but it is under 50%), and the attitudes of Californians about same-sex relationships are now more tolerant. In my experience polling on initiatives or amendments to the Constitution, if it doesn’t start out with large numbers, it is difficult to get more people to vote for the measure.’

Some readers, it seems, would have benefited from an explanation of why the gap between the two figures in this case was not what mattered when it came to forecasting how the amendment might fare this fall. What might have helped, said Lauter, was a sentence that said, ‘The margin may seem impressive, but at this stage of the campaign, it doesn’t mean as much as the percentage of people who say they would vote yes.’