Discussion concerning a possible independent run for president by Donald Trump has brought back the ghost of Ross Perot in 1992 to the Republican faithful.  It seems an article of faith in the Republican Party that Perot cost President Bush the election and threw the race to Bill Clinton.

However, this is not an accurate analysis of the effect of Perot running in 1992.  To use an analogy from AT contributor Herbert Meyer, we need to adjust the prism through which we view events.

First, take a look at trends in place as we went into 1992.  Here is the trend in total voter turnout for president:

1972: 77,744,030
1976: 81,540,780
1980: 86,509,678
1984: 92,653,233
1988: 91,594,809

Here is the trend in winning candidate vote:

1972: 47,168,710
1976: 40,831,881
1980: 43,903,230
1984: 54,455,472
1988: 48,886,597

This analysis tells us that in 1988, George Bush was already a weak candidate, but as usual, the Electoral College served to amplify the vote result.  Therefore, a weak election victory became a solid win of 426 electoral votes to 111 in the Electoral College over an even weaker candidate.  Far fewer people voted (perhaps as many as five million fewer) in 1988 than we should have expected based on the trends in place.

We can now see the results of 1992 compared to what one might expect in more typical circumstances.  In 1992, there were 104,426,659 votes cast, which put us back on trend or even above trend.  (The vote total in 1996 was 96,275,640.)  The impact of Perot running was to have some 10 million more people vote than would otherwise have been reasonably expected to vote based on the 1988 and 1996 numbers.  This accounts for over half of the Perot vote.

The other conclusion to make is that the remaining 9 million Perot votes were therefore "pulled" from another candidate.  Which one?

Bill Clinton got 44,909,806 votes in 1992, compared to 41,809,476 for Dukakis in 1988, so that election on election, the Democrat votes increased by some 3 million, while percentage of vote total declined.  The important point here is that the number of Democratic voters did not decline, and the increase was almost half of the Bush 1988 victory margin.

The Republican total of 39,104,550 shows a hemorrhage of 9.7 million votes.  Not all of Perot’s 19.7 million votes came from the Republicans.  The other 10 million had to come from somewhere.  The author posits that those are the people who would not have voted if Perot had not been in the race.  Note that already at this point, we are accepting the assumption that no Perot voter would have voted for Clinton if Perot were not in the race.  Such an assumption discounts the Reagan Democrats, who could have seen Clinton as a moderate worthy of their vote.

However, it is the Electoral College that matters, and Clinton got 370 electoral votes, so Bush, with Perot not running, would have had to move 102 of them away from Clinton for a victory.  Would all 9.7 million votes have gone back to the Republican total if Perot had not been there?  Is that a reasonable assumption to make, and what would be the impact on the Electoral College?

Let us look at Ohio, with 21 electoral votes.

1988 election total was 2,416,549 (R) and 1,939,629 (D)
1992 election total was 1,894,310 (R) and 1,984,982 (D) and 1,036,426 (Perot)
1996 election total was 1,859,883 (R) and 2,148,222 (D) and 483,207 (Perot)
2000 election total was 2,350,363 (R) and 2,183,628 (D) and 111,799 (Nader)

The reasonable conclusion is that Perot did cost Bush Ohio in 1992, but Clinton was also a weak candidate there, losing some votes to third-party candidates, with the Republican vote in Ohio trending down anyway.  It should also be noted that Dole did not get any of that lost vote back for Republicans.  Thus, we change the Electoral College vote to 349 for Clinton and 189 for Bush.

Georgia was won by Clinton by almost 14,000 votes, where Perot got over 300,000.  We shift those 13 Electoral College votes to make it 336 for Clinton and 202 for Bush.  In Louisiana, Clinton wins by some 83,000 votes, where Perot had just over 200,000.  Let’s say half the Perot vote stays home and the other half goes to Bush.  Thus, Clinton 325 and Bush 211.  Tennessee is won by Clinton by just over 93,000 votes, and Perot got just less than 200,000.  With the same assumption, Bush barely wins there, and the College changes to Clinton 314 and Bush 222.  Kentucky would move 8 votes to make it Clinton 306 and Bush 230.

Now we go out west, where Perot frequently got over 20% of the vote.  Colorado would shift, and now Clinton 306 and Bush 238.  Likewise for Montana, and the Electoral College stills stays with Clinton 303 to 241.  The same methodology would move Nevada to a Bush win, and the margin shrinks to 299 to 245.

Now we hit the wall.  In remaining states, the Perot vote would have to go substantially to Bush.  A number of Perot voters not participating of staying home in excess of 30% or 40% does not provide enough votes to close the Clinton victory margin.  To win, Bush would have to win states worth 25 or more electoral votes.  In West Virginia, almost 90,000 of Perot’s almost 109,000 votes would have to break for Bush in order to win those five electoral votes.  New Hampshire is the only Northeast possibility, which would shift the Electoral College to 295 Clinton and 249 Bush.

In summary, Ohio is the most favorable state for Bush that went to Clinton, and one can go down the chart to Arkansas, where Clinton got over 50% of the vote anyway.  Let us remember that California was won by Bush in 1988 and lost in 1992.  I’d suggest that not even fervent Republicans will suggest that Perot cost a Republican win in California, as that loss has been permanent since 1988.  Likewise with the result in Illinois.  Clinton’s victory was the result of turning states in the South and West away from Republicans.

A reasonable statistical analysis is that Perot took votes from Clinton in the Northeast and other liberal strongholds and took votes from Bush in the South and middle of the country.  The method used here of 50% of the Perot votes stays home, and 50% goes Republican is the most favorable assumption for Republicans that even approaches reasonable.  But those states were not enough to change the Electoral College result.  The loss of the Midwest put Bill Clinton in the White House.  Clinton still would have won as he did in 1996, and the Democrats did in 2008 and 2012.

Full disclosure: I was part of Perot's 1992 campaign staff, and we took a close look at  the results to assess how much "skew" occurred as a result of Perot's candidacy.

I would suggest to my Republican friends that the path to victory in the presidential race is to field a candidate who will appeal to voters in swing states and will cause those who would otherwise stay home to be motivated to vote.  Given public anger at the political class, this candidate with appeal is likely to be someone without a long history in the political arena.  While it is much more difficult to mount an independent run for president today than in 1992, that fact should really motivate Republicans to mount a primary process that yields an attractive candidate.