Popular vote since 1992

[Nichlemn]

Yawn, cherrypick starting dates and you can conclude anything. Start from 1968 and it's Republicans who have the upper hand... or go back to 1960 to make it Democrats again... or 1952 for Republicans again... or 1932 for Democrats again... and so on.

[Nichlemn]

Yawn, cherrypick starting dates and you can conclude anything. Start from 1968 and it's Republicans who have the upper hand... or go back to 1960 to make it Democrats again... or 1952 for Republicans again... or 1932 for Democrats again... and so on.

I am aware that dates can be used to draw one's own conclusion. If you looked at the three presidential elections in the 1980's you'd think that the Democratic party didn't exist.

Still, political eras are a real thing, and most political scientists can look back and determine periods of time when one party had the upper hand.
 

1) fifth party system: FDR's election created the new deal coalition. From 1932 to 1967 the Republicans only won 2 presidential elections.

2) sixth party system: There is debate as to when it ended, or if it has ended at all, but most agree it started in 1968 with the election of Richard Nixon. I personally believe that this era ended in 1992. Between 1969 and 1991, the Republicans won five presidential elections out of six. That's definitely a trend.

3) I don't know if you can call this a new era or a transition period, but since 1992, it's become increasingly more difficult for Republicans to win presidential elections, while it's become easier to win control of congress.

-since 1992, the democratic candidate's electoral vote total has never fallen below 251, while the republicans electoral total has never risen above 300. The GOP actually received less than 200 EV's in three out of the last 5 elections. That is most certainly a trend.

As Sean Trende points out, if elections were decided by coin flip, we should expect runs of consecutive wins in pretty much exactly the same way as they have turned out.

If these relationships mean anything, they should have some predictive power. I haven't done anything fancy, but just a quick correlation exercise in Excel shows that past election results had no correlation with the average results of the past five elections (or any other number). Just think about it intuitively by going back through the elections and comparing it with the preceding five elections.

[Nichlemn]

Here are the results (represented by 0s and 1s for the winners, broken up into political eras) from a country's elections, can you tell which one?

0   0   0   0    
1   1   0   1   1   
0   0   0   0   0   0   
1   1   1   0   1   1   1   1   1   0   1   1   0   0   1   1   1   
0   1   0   1   0   1   0   1   0   0   1   0   1   
0   0   0   1   0   0   0   1   0   1   0

The first political era lasted until election 4, where Party 0 won every election. Then, from election 5-9, Party 1 emerged on the scene, winning 4/5 elections, only to collapse in election 10 and usher in an era of uninterrupted dominance by Party 0. (Some historians consider these first three periods as one era of Party 0 dominance interrupted by a brief Party 1 period, as Party 0 won 11/15 of these elections). In the next era, lasting from election 17 to 32, Party 1 was the dominant party, winning 13/17 elections. The next era (starting at election 33 and continuing until election 46) was unique as the "anti-incumbent era": the incumbent party was defeated in all but one election. The 46th election marked the beginning of a Party 0 reign that continues to this day, winning 8/11 elections and never losing office for more than one term.

Source: 56 coinflips, random.org

[Nichlemn]

Except American elections aren't decided by coin flips and are determined by national factors.

Obviously they're not, my point is that if the results of a series of coinflips can produce a number of "eras" of "one party dominance", it calls into question the theory that similar periods of actual one party dominance indicated that one party was actually stronger rather mostly benefiting from good luck. Furthermore, previous elections may have little or no predictive power.  

[Nichlemn]

The problem with your simulation, Nichelm, is that the "eras" you isolate are extremely different in their duration. Of course if you make the right cuts it might look like a pattern, but what's also interesting with American political eras (at least as I understand them) is that they last for roughly the same amount of time, that is, between 35 and 50 years.

Turn the first three eras into one and then they're pretty much the same.