Trends - they're a favorite talking points.  Well, I've gone through my share and attacked a few, but I was doing research for a separate diary when I found some very interesting data in the process that somehow has gone unnoticed.  

I was looking at data about party affiliation over time, when I found an interesting chart from Wikipedia derived from registered voters in 2004.  That is, among the 169 million registered voters at the time, 72 million (44%) were Democrats, 55 million were Republicans and lastly, 42 were independents (decline a party or third party).  That seems right, but I wanted more and found a nice seven point scale that the Center for National Election Studies has used since 1952 which can be found here: http://www.electionstudies.org/nesguide/ text/tab2a_1.txt  Self-identification might be a better means of analyzing how people vote than their party affiliation.

I took the liberty of running some analysis on the data from 1980 on and here's some statistics jargon if you would like it (numbers given are percentages of the total population surveyed; apolitical responses are excluded due to their small weight and lack of likelihood to vote):

Strong Democrat Analysis:
Mean: 17.92
SD: 1.382

Weak Democrats:
Mean: 19.077
SD: 2.62

Independent Democrats:
Mean: 13
SD: 2.04

Independents:
Mean: 10.85
SD: 1.344

Independent Republicans:
Mean: 11.615
SD: 1.386

Weak Republicans:
Mean: 14.385
SD: 1.2609

Strong Republicans:
Mean: 11.923
SD: 2.2159

I lump the groups into three separate categories: Democrats (Strong and Weak), Independents and Republicans (Strong and Weak).  Ultimately, these are our relative averages of the general electorate.

Democrats:
Mean: 37
SD: 3.2404

Independents:
Mean: 35.462
SD: 2.6336

Republicans:
Mean: 26.308
SD: 2.2504

But do they turn out in these numbers?  Long story short, no.  Here's the turnout based on exit poll data from elections 1996, 2000 and 2004 complied from LA Times and CNN Data (numbers given are percentages; older data is difficult to find).

Democrats:
Range: 6
Mean: 40
SD: 2.1909

Independents:
Range: 8
Mean: 23.167
SD: 3.4559

Republicans:
Range: 4
Mean: 36.333
SD: 1.5055

From this data, you can see a disparity between the general electorate and the actual voting electorate - Democratic voters exceed their general electorate proportions but only slightly.  Republicans, despite being the smallest population, constitute the second largest percentage of voters in the election and the most reliable in terms of turnout size.  Independents, despite being the second largest group in the general electorate, tend to trail both sets of party-affiliated voters and are most unreliable in terms of turnout size.

By all means, if this is the case, Democrats would only need to meet the mean turnout rate and split the independent vote to keep the presidency in all three elections.  What happened in 2000 and 2004 that hurt that so significantly?  Let's examine Party Loyalty and which way Independents tend to lean. (NOTE: LA Times has a weird outlier from 1996...I've excluded it so I can keep something resembling normal distribution and frankly, it's weird when it comes to the Independent Numbers)

Democratic Party Loyalty (percentage of Democrats that voted for the Democratic nominee):
Minimum: 84 (CNN - Clinton '96)
Maximum: 89 (CNN - Kerry '04)
Range: 5
Mean: 87
SD: 2

Republican Party Loyalty:
Minimum: 80 (CNN - Dole '96)
Maximum: 94 (LA Times - Bush '04)
Range: 14
Mean: 90
SD: 5.7

Thus, Republican Party Loyalty appears to be higher than Democratic but slightly more erratic (and skewed toward Dole's awful showing in 1996).  What about independents?

Independent Votes for Democrats:
Minimum: 43 (2 instances; CNN - Clinton, LA Times - Gore)
Maximum: 49 (2 instances; Both for Kerry)
Range: 6
Mean: 45.8
SD: 3.0332 (too skewed to mean much)

Independent Votes for Republicans:
Minimum: 23 (CNN - Dole)
Maximum: 49 (LA Times - Bush '00)
Range: 26
Mean: 43
SD: 11.203 (skewed toward awful Dole showing)

Independent Votes for Third Party Candidates (Specifically Perot):
Minimum: 3 (2 instances - both 2004)
Maximum: 17 (CNN - Perot '96 (LA Times had him at 50)
Range: 14
Mean: 7.6
SD: 5.7271 (Skewed toward excellent Perot showing in '96)

Moral of the story: Without a third party candidate, independents  break in relatively even patterns between the two major parties.  However, they land on both sides of the divide - Going to Bush in 2000 and Kerry in 2004.

What does Obama or Clinton need to win come November?

- Democrats at or above their average mean (a strong base).

• Republicans below their average mean (a depressed or disenfranchised Republican base).
  
• Independents breaking for the Democrats, no matter what their turnout.
  
• A strong rightist, third-party candidate (which tends to steal independent votes from the Republicans).

Implications:

- So long as the Republicans maintain stronger Party unity than the Democrats, Democratic numbers do not necessarily give us an edge. - Independent votes can easily sway an election; it's a matter of motivating them to show up and vote for you. However, that is difficult to do and predict.

- The eventual nominee will need to unite the party immediately at or right after the convention and then work on independents.

• McCain (yes, I'll treat him like the nominee) cannot gain traction in the Republican and independent camps and cause higher than average turnout in the Republicans or cause the independents to break evenly).
  
• Without a strong third party candidate, the Democrats can't treat this like a slam dunk.  It's going to be a hard campaign based on past data.

Thoughts?