The analysis I've seen looks at divisive primaries in relation to the party with the White House. The conclusion is that divisive primaries have minimal effect on the challenging party. They do have an effect of the party defending the Presidency. In the period from 1864 to the present a presence or lack of a contested nomination in the incumbent party has forecast the outcome over 90% of the time.
Lichtman's Keys, huh? Let's look at the times the Democrats have the lost White House to the Republicans...
2000*
1980
1968
1952
1920
1896
1884*
1860
*-The Democrat won the popular vote. I don't think Cleveland had a serious challenger. Gore did have Bill Bradley, but that didn't last very long. The Democrats had divisive primaries in the remaining races.
It doesn't seem to be as good a predictor for the Republicans, but as they're not the incumbent party, it's not really relevant.