Thursday, August 26, 2004

Even LA Times Can't Spin Kerry Collapse

The Los Angeles Times has another poll out, and as usual there’s a great deal of question as to the validity and objectivity of their sampling.

You’ll recall that their June poll showing Kerry up 51%-44% was based on a sample of 38% Democrats, 25% Republicans, 24% independents, and the remainder consigned to the flotsam. Correcting for that spectacular skew put Bush up 49%-45%.

The pro-Kerry tilt isn’t quite that precipitous this time – approximately 38% Dems, 32% GOPers, and 30% independents/other – but still measurable.

What’s interesting is that even with this tilt, Bush still leads in the poll 47%-44%.

According to the internals, Bush leads among Republicans 93%-3% while Kerry leads among Democrats 81%-15%, with the independents split evenly. So, if the electorate is truly “split right down the middle,” that should mean a fairly substantial lead for the President – certainly outside the margin of error.

A KerrySpot reader crunched the numbers:

I also looked at the LA Times survey and found it really badly biased. How do their numbers add up? Something must be really wrong with their population sample. Here's the breakdowns and the math behind it.

While 3% of voters who called themselves Republicans said they would vote for Kerry, Bush drew 15% of all Democrats, and 20% of Democrats who consider themselves moderate or conservative, the poll found.

Ok, so out of their 1,597 surveyed, 782 choose Bush. If 20% of those surveyed choose Bush, but called themselves moderate Democrats, how many Democrats were included in the survey? We know 20% of those calling themselves Democrats (156) chose Bush. According to the statement, Kerry received 3% of Republicans out of his 43% (687) which equals 20 votes. Therefore, 97% of Kerry's support came from Democrats (667), plus Bush's Democrat support (156) votes equals 823 Democrats. That means that 52% of the sample's base was Democrats (832/1597). Back out the 2% bias (32 people) from the study, and the numbers are Kerry (655) compared to Bush (782). 50% Bush (782/1565) compared to (655/1565) 42% Kerry. That's an EIGHT point lead and way outside the margin of error.

FWIW, I took the 38-32-30 template, equalized the Dem and GOP proportions at 35 each, and came up with a similar result (Bush 49.4%, Kerry 41.6%).

Whatever the vagaries of one generally unreliable poll, the larger point is the direction of the trend. And if one of the Kerry campaign’s house organs shows even the five point move in Bush’s direction over the past month it’s acknowledging (IOW, since the Swiftboat Vet storm erupted)…well, it’s no wonder that Lurch is developing a Samson complex.