Numbers, primarily. If you'd like to read some actual analysis, this is an excellent page: Leinsdorf.com
Three samples from that page:
In the Governor's races, Perot's voters cast 18% of their ballots for the Republican candidates; 56% of their ballots for Democratic candidates, 17% for independent candidates, and 8% did not bother to vote for Governor. If Perot's voters had voted for Bush and Clinton in the same proportion that the voted for the Republican and Democratic candidates for Governor, Clinton's lead would have increased by 7.5 million votes.
Perot's voters voted overwhelmingly for Democratic Governor candidates, and only marginally in favor of the Republican candidates for the House and Senate. Perot's voters favored Republican Senate candidates by 2.28%, and Republican House candidates by 2.69%. Because Perot's voters were only 1/5th of the total, that translates into about another 500,000 votes or 0.5% for bush if they had voted in a two way presidential race the same way they voted for the Senate and House. That is about 1/7th of the margin by which Bush lost.
This analysis can be further confirmed by comparing the 1992 and 1996 results where Perot's vote dropped by 10 million compared to 1992. By comparing the vote totals for Clinton in both years with Bush's and Dole's (assuming Dole voters and Bush voters were the same voters) it is possible to conclude that in 1992 Perot's presence on the ballot cost Bush: Montana, North Carolina, Colorado and Georgia. However, Perot cost Clinton: Florida and Arizona in 1992. So, in 1992, Perot cost Clinton 32 electoral votes while costing Bush 37 electoral votes. Bush lost by 100 electoral votes, so 5 more would not have given him victory.
“Coat tails” do matter, and you can not know how those voters would have voted, if Perot was NOT on the ballot -—
People absolutely do NOT vote straight ticket, much of the time. There are always more people voting at the TOP of the ballot than vote for the bottom of the ballot/state legislative candidates. However, it is a political fact of life that, for an incumbent, there is a “pro incumbent” vote and an “anti incumbent” vote, so that Perot split the “anti incumbent” vote, considerably, to the benefit of CLINTON! Beyond the obvious Perot mess, there have been countless times that Conservatives split the conservative vote, handing an election to the liberals.