If only X could vote

Based on CNN's exit polls from the 2016 election by state, in the states where these polls were conducted, here are what the results of elections would have been if only members of group X could vote. Light color (red/blue) = less than 5% win, medium = 5-10% inclusive, and dark = 11% or more (if a state margin is between 10 and 11%, it will be medium color; this only applies to Oregon in the first map).

Let's start by taking the actual 2016 EV results (recall they were 306 Trump, 232 Clinton) and grey out all the states that CNN does not have exit polls at all for. As you can see plenty of states are grayed out--mostly ones Trump won in the Great Plains/Western area, as well as some Trump-loving southern states and a few liberal ones in the Northeast. The 2016 exit poll included 28 states, excluding 22 states and DC. In total, we see Trump with 224 EVs and Clinton with 187, so no one has enough to win here:

Note that in the hypothetical scenarios below I will be giving all of Maine's 4 EVs to the same candidate (even though this might not actually happen, and didn't actually happen in 2016).

If only men could vote, Clinton would lose 62 EVs in these states, and would win only 125 EVs. Trump, meanwhile, would gain all 62 of these EVs and be able to win the Electoral College even without the excluded states (most of which he actually won anyway). These Trump gains/Clinton losses come from Minnesota, Nevada, Washington state, Colorado, New Mexico, Maine, New Hampshire, and Virginia--Clinton won all of these states but more men voted for Trump than for her in each of them:

But if only women could vote, unsurprisingly, Clinton would do extremely well, flipping many states Trump won and getting a whopping 367 EVs: 

If only white people could vote, the results would be even more favorable to Trump than they were with men. Clinton would hang on to ME and WA, but lose NY, NJ, and even her birth state of IL! The result would be Trump getting a whopping 333 EVs, way more than enough to win:

What about if only white women could vote? We've heard a lot about white women since 2016 because most of them voted for Trump that year (despite most women in general voting for Clinton, as usual for Democratic candidates). The results reveal some strange patterns: some states flip in Clinton's favor (IA and WI) compared to the actual results, while other states (e.g. NJ, MN, and IL) actually flip the other way. In all, Trump gets 275 EVs, just enough to win, and Clinton only gets 136. This represents a Republican gain of 51 EVs compared to the actual results (again only based on the states w/exit polls):

Also, it seems obvious that if only white men could vote, Trump would absolutely crush Clinton almost everywhere, no? Indeed he would have, and the result is so lopsided even I was surprised by it: 

Yes, you're reading that right: the only state (included in CNN's exit polls) where more white men voted for Clinton than Trump is Washington state! So of course Trump would win easily in this scenario, with a whopping 399 EVs compared to only 12 for Clinton.

If only people aged 18-45 could vote,  Clinton, no surprise, would do very well, winning all but 8 states included here. This would result in her getting 339 EVs, 152 more than she actually got in the included states. Oddly, one state would actually flip from  Clinton to Trump if only those under the age of 45 could vote: Maine.

But if only people 45 or older could vote, we would see Clinton lose a bit of ground relative to her real performance.

Finally, if only white people with college degrees could vote, Trump would get 212 EVs (12 less) to Clinton's 179 (8 less). Note that Pennsylvania is marked "other"; this is because the white college graduate vote was split exactly evenly (48-48) in that state. 

In conclusion, ranking demographic groups from most to least Republican gives this result:

1. White men
2. White people overall
3. Men overall
4. White women
5. People 45/older
6. The American electorate overall
7. White college graduates
8. People under the age of 45
9. Women overall