Silver lays out the cause-and-effect question as follows:
There were two periods that account for most of the decline [in Obama's poll numbers]. One was the period immediately following his inauguration until about the first week in March; Obama's ratings fell by about 5 points over this interval. The other was a longer period from the end of May through mid-August, during which time Obama's approval declined by 9 points or so. Those two periods collectively account for about 14 points of the roughly 17-point decline that Obama has experienced.
He quickly writes off the economy as a prime driver:
Was it the economy? Undoubtedly, the economy is a significant part of the story. The employment reports that came in during this period showed the economy losing about 300,000 jobs per month, which is really, really bad. Still, it wasn't as bad as the reports that came in during April and May, which had the economy losing 500-600K jobs per month, and during which time Obama's numbers were rather steady. (A more robust indicator, perhaps, is consumer confidence, which had a bit of a reversal in June and July before picking up again, but the correlation there is still fairly weak.)Silver then tests various events, such as the summer grilling of the health care bills, the Sotomayor nomination, Franken's fight to get seated in the Senate, etc. -- and finds no strong correlations.
I'm no mathematician. But if you assume a slight perception lag and step back to look at the big picture, it seems to me that unemployment numbers match up with Obama's poll drops -- in aggregate, if not in lockstep. The Dec. '08 unemployment rate was 7.4%; by March it was 8.6%, a 1.2% increase over four months -- explaining Obama's first slide down the pedestal fairly well (though a steep drop in March, per Silver, overlaps with a stabilization in Obama's poll numbers). From January to February it increased .5% - a huge spike, matched only by the spike from April (8.9%) to May (9.4%). It's true that that second big spike came in a month when Obama's numbers were holding steady (as does the March spike), and that the unemployment rise leveled off in June, July and August, when Obama's approval ratings were falling. But again, assuming a perception lag, it's not surprising that Obama would suffer a poll drop basically a month after the horrific unemployment rise from 7.4% in Dec. '08 to 9.4% in May '09. During Obama's second poll plunge unemployment continued to rise, albeit more slowly. In that long hot summer, perhaps the depth of the trough the country had fallen into really (pardon the expression) sank in.
By way of perspective, Obama's poll numbers have held up quite well compared to those of Reagan, who also endured a steep rise in unemployment over his first two years, with a rise in '82 that roughly matches that of 2009. When Reagan came into office, unemployment already stood tall at 7.5%. It was still at 7.5% in May, when Reagan was shot by a would-be assassin -- at which point his approval rating climbed to 68%. By December '81, unemployment had spiked to 8.5% -- half the percentage rise that the U.S. endured in Obama's first 6 months -- and his approval rating was down to 49%, just about where Obama's is now. By December '82 unemployment had soared to 10.8%, and while it eased off to 10.4% in January '83, at that point Reagan hit his polling nadir of 35% approval (two months after the Republicans lost 27 House seats). By May 1984, however, unemployment was back down to approximately where it was when Reagan started his term - 7.4%. The rest, of course, is history -- and Republican myth.