I was over at Donkey Rising today, reading through Ruy Teixeira's latest interesting post. I made it about halfway through the post until I found the sort of link that is a real gold mine for a numbers junkie such as myself: the archive of long-term data from the National Election Survey. I'll have to finish Ruy's post later, because I am so entranced by the numbers I found in that link.

One of the most intriguing long-term graphs I saw had to do with voter knowledge concerning which party controlled the House of Representatives. With the exception of 1974, every two years since 1970 the National Election Survey has asked the populace the following question:

The results were as follows:

Year	     '70   '72	 '76   '78   '80   '82
Incorrect     50    36	  39	41    29    68
Correct       50    64	  61	59    71    32

Year	     '84   '86	 '88   '90   '92   '94 
Incorrect     45    67	  41	51    41    30
Correct       55    33	  59	49    59    70

Year	     '96   '98	 '00   '02   '04 
Incorrect     27    34	  49	72    46
Correct       73    66	  51	28    54

Looking into the internals of these numbers, "independents" and "moderates" have almost always scored noticeably lower on the knowledge scale than liberals, conservatives, Democrats and Republicans. I don't find this surprising, as it backs up my general thesis that many of the people who consider themselves "independents" or "moderates" do so primarily because they are less politically engaged than the rest of the population and are not even aware of their strong similarities to one coalition or the other. It also isn't surprisingly that as someone's income and education level rise, the more likely they are to know the correct answer, and that this has an impact on other internal cross-tabs in the survey.

Leaving aside what isn't surprising, what really blows my mind is just how much these results vary from one election cycle to the next, even though control of the House has only changed once during the entire survey period. As you can see, voter knowledge on this topic was particularly low in 1982, 1986, and 2002. By contrast, it was particularly high in 1980, 1994, 1996, and 1998. For some reason, knowledge of who controlled the House dropped from 73% in 1996 to just 28% in 2002, even though there was no significant change in the composition of Congress during those six years.

Why was voter knowledge so unbelievably low in 2002? Were people confusing control of the Senate with control of both Houses of congress? That seems to be a possibility, as the lowest years of voter knowledge (1982, 1986, and 2002) all occurred during times when control of the two branches of congress were split between the two parties. Incorrect answers also seem to take a large jump after one party wins the national popular vote in the Presidential election but does not end up controlling the House of Representatives (see 1970, 1982, 1986, 1990 and 2002). The three lowest years of voter knowledge occurred when these two factors combined with one another. When the party that controls the House of Representatives loses the popular vote in the Presidential election and does not control the Senate, apparently very few people are aware that that party actually controls the House of Representatives.

I think this means two things. First, people pay significantly more attention to the Senate and to the Presidency than to the House of Representatives. I guess that isn't surprising. More interestingly (and optimistically), it should also mean that many more voters will believe that Republicans control the House in 2006 than believed it in any election cycle since 1998. After all, people know that Bush won the popular vote in 2004, and that Republicans control the Senate. Thus, they will probably at least assume that Republicans also control the House. Considering low congressional approval ratings, this should result in an anti-Republican sentiment that would be reflected in generic congressional ballot polls. With Republicans consistently behind in such generic ballots by sizable margins, and seemingly unable to improve upon a support level in the high 30's or low 40's, there appears to be evidence that this is indeed happening.

Voters tend to never like the job performance of Congress, and I suppose it thus isn't surprising that whichever party they think is in power tends to do poorly in Congressional elections. In 1980, 1994, and 2002, most voters (over 65%) thought Democrats were in charge of the House, and Democrats suffered real losses as a result. In 1982, 1986, 1996, and 1998, most voters (over 65%) thought Republicans were in charge of the House, and Republicans suffered real losses as a result. In all the years in between, voters didn't really know who was in charge of the House (no consenus of 60% or higher), and there was no significant change in seats. If I am right about the knowledge patterns I listed above, most voters should know that Republicans are in charge in 2006, which should result in Republicans losing seats in 2006. Will it be enough seats for Democrats to take over? I don't know, but it should be an encouraging sign nonetheless.