I have been reading some fanciful tales about presidential approval ratings lately. On the right end of the political spectrum, their collective memories are betraying their prejudices. The claim, to paraphrase, is that Bush’s poll numbers aren’t any worse than Clinton’s were. Ergo: Bush is not truly sliding into the cellar. He’s merely in a little slump.

Facts are such stubborn things. They have this inconvenient way of interfering with what we choose to believe. First of all, you will need to understand something about polls and polling. The most reliable polls are the big ones. Why? Because they have the resources to do have a large sample and to do it properly. They also have trained people working for them who don’t try to skew either the questions or the results. How can I say this? If you’re a pollster and you skew your results, you will be discarded like yesterday’s newspaper: It’s bottom of the birdcage time.

The larger the sample, the more reliable the poll results. If you were to sample the entire population (as in the census); you would be assured of 100% reliability. That is something that only the federal government can afford to do. If you have a sample of ≈1,500, you can be assured of a margin of error (MoE) of no more than 2.5 percentage points for the entire population. The smaller the sample, the larger the margin of error will be. If you are dealing with overwhelming results (e.g. 60-40%), this isn’t a huge problem. But, these if you’re faced with a close situation, as we so often are in public opinion these days, the size of the sample is very important to the reliability of your results.

Also, remember that the closer the numbers, the more divided the opinion is on a subject. If the numbers are, say 52 to 48, the electorate is very divided. If it’s 70-30, you have broad agreement on the subject.

Finally, the opinion of the moment isn’t nearly as important as trend lines. Any single poll is a snapshot in time. There are all sorts of things that will affect it before the next time the question might be asked. History effects (e.g. disasters, attacks) can change opinion on a dime. So, looking at the trends over time will tell you more about a situation than any individual poll.

Ooh, ooh, one last point: if the American public loses faith in the president’s abilities and his approval ratings go below a threshold of ≈40%, the approval cannot be regained. When Bill Clinton was president, his ratings did not fall to that level, as a matter of fact they didn’t get much lower than 50%. Even in the midst of the Lewinsky scandal his numbers were in the 60% range. The only other president who had as good ratings in his second term was Ronald Reagan. As a matter of fact, according to the AP, Bubba was always a popular guy on the playground, warts and all.

Here’s what the trend lines for George W. Bush show us: his numbers are getting as low as Richard Nixon’s did when the Watergate scandal finally broke open in ‘73. No matter what anyone on the right says, Bush has gone below the 40% threshold. The graph above doesn’t show the latest numbers; which are even lower for our Georgie. There is no way back. He is the lamest of lame ducks. Any ‘capital’ he claimed for himself a year ago has been spent.

Say good night Georgie.