muon, what your apologia for TRUMP fails to note is that in each case, it was the foreign-seeming delegate of his that placed worst of the three in that district. That can't be explained away by pointing out the strengths of other candidate's delegates. All that does is explain who was best positioned to take advantage of the racial bigotry of TRUMP voters unwilling to cast votes for delegates with strange names.
The drop-off from the presidential vote to the delegate vote was about 20%.
If all voters who voted for a presidential candidate, and voted for delegates, voted for three, it would be:
80% for 3; 20% for 0.
If some voted for only one delegate; then for every 3% who vote for one, the number who voted for three decreases by 1%.
79% for 3; 3% for 1; 18% for 0.
78% for 3; 6% for 1; 16% for 0.
70% for 3; 30% for 1; 0% for 0.
The last is quite unlikely since most of the delegate voting is fairly balanced and it would be quite unlikely that those who voted for one would vote selectively for different candidates. That would only happen if the presidential campaign carefully instructed voters which candidate to vote for - and they would not do that.
So we can estimate somewhere in between:
75% for 3; 15% for 1; 10% for 0.
In some cases, the vote-for-one vote might split fairly evenly: say 6%, 5%, 4%, which would give three candidates 81%, 80%, and 79% of the presidential vote. In other cases it might split 8%, 7%, 0%, or 10%, 4%, 1%.
If you are being "selective", you are not likely to vote for an odd sounding name. You are also more likely to pick the first candidate. Trump delegate candidates are less likely to be recognizable to ordinary voters (see the vote for Adam Kinzinger, Jim Oberweis, and Darrin LaHood for contrast). State senators did pretty good downstate, where (1) people are more likely to know who their legislators are; and (2) the legislators are Republican.
We can measure the consistency of voting for the three delegate candidates in each districts, using the standard deviation divided by the average. Typically, the Rubio and Kasich slates had more variance, followed by Cruz and Trump. Rubio and Kasich are more likely to have known candidates, who even less informed voters would recognized. They might also gain some cross-over votes.
An extreme example is Rep. Adam Kinzinger, who as a Bush delegate in IL-17 finished 7th, with almost 3 times the number of votes as his two slate-mates combined. This was not a vote for Bush, but a vote for Kinzinger. It also appears to have upset the voting pattern for all the presidential candidates. If you vote for a Bush delegate, you can only vote for 2 delegates of your favorite candidate.
Incidentally, in IL-13 the combined vote for the three Cruz delegates was greater than for the three Trump delegates. Voters plumping for Hartmann secured him one of the three spots.