As those of you who follow various pollsters know, Rasmussen Reports numbers can be a little funky. That said, they are strongly suggesting that Bush's poll numbers began nosediving early this week.
http://www.rasmussenreports.com/...
Before I get into the gory details, let me say that the trend is more interesting than the absolute numbers. But, ... to give you some way of judging the absolute numbers, Rasmussen's Bush approval numbers tend to run about 4% to 5% higher than the average poll. The discrepancy is largely due to the fact that RR polls for Strongly Approve/Somewhat Approve/Somewhat Disapprove/Strongly Disapprove rather than a binary approve/disapprove.
So ... Rasmussen polls nightly (500 people a night) and each day reports the three day rolling average from the previous three nights. Well, on July 25 the nubmers (reflecting polling rom 7/22--7/24) dove from 42 to 37. A pretty big drop, but ... could it have been the result of a single outlier night? Hard to say at first. But, the next two days also came up with 37. Still, because of the rolling average, a single outlier night will impact results for three days. So, I eagerly anticipated today's number ... which was 38. Now that we've had four days worth of results, we can do some real analysis.
Since today's result covered 7/25--7/27, we can average this with the 37 reported on July 25 to get 37.5 for the six days from 7/22--7/27. You can do the same with earlier RR results to get a reading of 44.0 for 7/16--7/21. That's a 6.5% drop from one six day period to the next! Since the sample size for each six day period was 3,000, this almost certainly suggests a big drop in underlying approval in the two periods. The MoE on each result is around 1.8% using a 95% confidence level. Skipping a little math, we can say with 99.9% certainty that there was at least a 2% drop in the underlying approval between the two periods. That takes into account that rounding might be contributing to the drop ... i.e., today's 38 might really be 38.49 .... I haven't taken the time to do the math, but I suspect that at a 95% confidence level there was a 3% drop ... and, again that's assuming the maximum possible contribution from rounding.
Now, 3% may not sound like a lot, but it is. Pollkatz's Bush Index does a poll of polls every two weeks. The only time since 12/31/04 with a change of at least 3% from one two week period to the next was in early November with the Rove indictment/Miers nomination. So, even a 2% drop -- if it really happened -- would be a big deal.
So far, no other polls are suggesting a drop of this magnitude. But, it looks like the big drop probably began on Monday. And, I haven't seen any other polls out there where most of the data has been on or after Monday. It will be very interesting to see what happens with polls coming out over the next few days.
If there has been a precipitous drop, what caused it? It would have to be Lebanon ... plus all the negative associations that come with it (Iraq, gas prices, etc.) Specifically, the unfortunate reports that Hezbollah is proving to be a formidable foe for the IDF and the implication that there is not going to be a quick resolution may have tiggered the approval plunge.
Let me close by reiterating that RR can be a bit squirrely. It tends to zig when others are zagging. On more than one occasion I've sworn never to look at RR again. But, this change is awfully big. We'll know soon enough if its real.