Odds and ends and trends - National elections

A couple of different posts from different places in the last week or so have got me playing with data, this time on US national elections.


Trends in the United States House Of Representatives

A post from John Hawkins last week about the trend in Senate seats for each party made me curious about what the House of Representatives looked like. The clerk of the House has a lot of information on-line, but most of it's in PDFs, and it's difficult to access the information. The last 7 elections, however, are available in HTML, so I was able to construct a database containing that information, all of the national election data, since 1992. (Presidential electors, US Reps, US Senators - no state offices.)

I also was able to find the seats by party, so I looked at that first. (I'm sure that the professionals have access to all of this in super-duper databases. I'm not a professional, so I don't - I had to make my own. Is there an elections database out there with free access?) Anyway, I've looked at the Republicans representation in the House for the last 60 years or so. In most cases, I'm looking at it as a percentage of the 2-party (Republican, Democrat) cohort.

There are a couple of things that show themselves. One is that the Republicans look to have hit a low point in 1958, with a gradual climb up from there, with a couple of sharp "debacle" elections. The first was 1964, with Goldwater at the top of the ticket, and coming off of the Kennedy assassination. The second is 1974-1976, elections during which the Republicans were pummeled by Watergate. I don't have vote totals for those years, and can't be sure what happened votewise. What I can be sure of is that the representation seemed to have stabilized with the Republicans holding about 40-42% of the seats. Until 1994. I remember being shocked that the Republicans took the house - looking at the trend, I'm even more shocked than I was then.



There was a huge shift in 1994. The percentage of Republicans did NOT drift up. It was stable in the low 40s until 1994, and then it jumped to the low-mid 50s. And stayed there.

And, interestingly, it doesn't appear to be the case that the Republicans vote support was better than their seat representation and re-districting re-aligned things. I don't have vote totals earlier than 1992, but on the next chart, I've got both the percentage of votes cast, and the percentage of seats held, and they track almost perfectly.



The percentage of seats that they hold is marginally smaller than the percentage of the vote that they get, and has been in each of the last 7 election cycles. But it doesn't look like there was some huge Republican congressional vote out there that wasn't represented - it looks like a lot of people shifted their votes in 1994, and then continued to vote Republican.

Until we look at the next chart. There were actually far fewer Republican voters in 1994 than there were in 1992. It's just that there were even fewer Democrats than that.




It makes sense that both parties would draw less support in the off-year election of 1994. And it becomes clearer when we sort the off-year elections together and the presidential-year elections together.



The Republicans did make a big jump from 1990 to 1994, and the Democrats dropped. Whether those groups actually contains party-switching is not clear from this data, but presumably there was exit-polling to give the parties some idea about it. I suspect that this was the election that did finally break the "Democratic habit" for many conservatives in the south and, as such, represented a re-alignment.

Because the Senate only has 1/3 of the seats up every two years, it is far less susceptible to short-term impacts than the House (as they were designed). The power of incumbency then makes stability in both houses the default condition. I suspect that's why the Senate chart shows a trend over time more clearly than the House chart does - the House is affected greatly by a "short, sharp shock." There certainly was one in 1994. Hillary-care perhaps? Gays in the military? Whatever it was, it looks like a tipping-point was reached.

Which should absolutely not cause one to assume that things can't tip back...




Behold the power of Cheese! ... er ... Gerrymandering!

The other thing that started me on this was a post from Alexander McClure at PoliPundit.com, looking at The House of Representatives. His take is that

There are 218 safe Republican seats.
There are 184 safe Democratic seats.

There are 15 Republican seats that either are lean Republican or toss-ups.
There are 18 Democratic seats that either are lean Democratic or toss-ups.

In short, for Democrats to win control of Congress, they would have to hold every Democratic vulnerable seat and win every vulnerable Republican seat in addition to one Republican seat that does not seem at all vulnerable right now.

Does it sound impossible? It is.

My inclination is to agree with that last line. But those charts should prove a cautionary tale to Republicans who feel like getting cocky. If you'd said in June of 1993 that "in order for Republicans to win control of congress, they have to hold every Republican vulnerable seat and win every vulnerable Democratic seat...Does it sound impossible? It is," I'd have agreed with that, too. Yet we can see what happened in 1994...

I'm not certain which seats he's putting in which category, and I've not looked at all of the individual seats. Anyone reading this should assume (as I do) that he knows what he's talking about, and I'm just a guy with a spreadsheet. But I've spent a little time looking at trends, and there's no question that, as a general rule, congressional seats continue to get safer and elections less competitive. All seats, both parties. In 1992, there were 76 congressional races in which the winner took less than 55% of the two-party vote total. In 2004, there were only 21. Out of 435. 414 of the 435 members of the House of Representatives took 55+% of the two-party vote in their elections last fall.

The following were the only "closely-contested" seats in 2004:

Closely-contested seats

State	District	Winner	Percent of 2-way vote
NORTH CAROLINA	11	Republican	54.90%
INDIANA	2	Republican	54.88%
INDIANA	8	Republican	54.50%
NEW MEXICO	1	Republican	54.44%
OREGON	5	Democrat	54.39%
CONNECTICUT	2	Republican	54.22%
NEW YORK	26	Republican	54.12%
MINNESOTA	6	Republican	54.03%
MISSOURI	3	Democrat	53.94%
SOUTH DAKOTA	At Large	Democrat	53.75%
CALIFORNIA	20	Democrat	53.40%
COLORADO	4	Republican	53.27%
WASHINGTON	8	Republican	52.44%
CONNECTICUT	4	Republican	52.43%
COLORADO	3	Democrat	52.06%
TEXAS	17	Democrat	51.92%
GEORGIA	12	Democrat	51.81%
ILLINOIS	8	Democrat	51.70%
PENNSYLVANIA	6	Republican	51.01%
NEW YORK	27	Democrat	50.39%
INDIANA	9	Republican	50.25%

This table shows the trends over the past 7 elections. You can see that not only have the number of "highly competitive" races dropped, the number of "no contest" races has increased. Over 75% of the elected representatives in the last 3 elections took over 61% of the two-party vote, meaning that they won by 20+% points.

Competitive Elections? I don't think so...

Year	HighlyCompetitive	Competitive	AlmostCompetitive	NotVeryCompetitive	NotClose
1992	7.65%	8.16%	13.27%	16.58%	54.34%
1994	8.82%	9.89%	8.82%	16.04%	56.42%
1996	8.02%	8.73%	8.49%	12.97%	61.79%
1998	3.82%	4.41%	8.82%	20.29%	62.65%
2000	4.02%	4.83%	7.51%	13.40%	70.24%
2002	3.65%	4.21%	7.30%	9.55%	75.28%
2004	1.61%	1.88%	8.04%	13.14%	75.34%

You can see the competitive races, which weren't very common to being with, petering out, and the "not close" races increasing...


There have been 458 congressional seats contested over the last 15 years, 435 per year. (The extra 23 are seats that either appeared or disappeared in 2002 following re-apportionment.) Of the 458 districts, 114 of them, nearly 25%, have elected the same representative in each of the last 7 elections. 252 of them, 55%, have elected representation from the same party in each of the last 7 elections. There have been 3045 congressional district elections, won by only 866 different people.

Congressional Elections 1994-2004

	1994	1996	1998	2000	2002	2004
Changing parties	14.7%	10.6%	4.8%	3.9%	16.1%	10.1%
Changing representatives	22.8%	20.0%	14.0%	13.3%	37.5%	19.3%
Incumbent Victory	77.2%	80.0%	86.0%	86.7%	62.5%	80.7%

All told, incumbent Representatives have won 81.9% of the elections over the last 6 cycles. In short, there's very strong reason to believe that whoever holds the House will continue to hold the House (duh!), absent the out of power party being able to "nationalize" the elections (as the Republicans did in 1994) or serious extra-curricular events (like 1964 and 1974.)