It would be logical to assume that machines with lower jackpots would have a lower variance - just as video poker machines with the fewest (or smallest) jackpots have lower variance.
Generally yes, playing reel slots with the lowest max payouts will often lead to the lowest variance, but I'm sure there are some exceptions out there.
Five-reel slots (penny/nickel slots) bet all lines will likely give you the lowest variance. Barring that, you probably want to look for a machine that pays out on three blanks.
Definitely true, but it's also possible the single line quarter machine would be set a 1-2% higher for return, so OP may still rather play those. And yes, paying on three blanks should help in theory. Also avoiding machines with 2X, 3X symbols, etc may help. Finding machines with nudging reels may also lower variance. But these bits of "advice" may not always be true. According to Harrigan/Dixon, Double Diamond Deluxe (which has a "positive" nudging feature, but a "negative" double payout feature) has a SD of only 6.35 bets/spin. This is on par with 9/6 Double Double Bonus video poker (6.48 bets/hand), and one of the lowest SDs I have ever heard of for a slot machine.http://jgi.camh.net/doi/pdf/10.4309/jgi.2009.23.5
One example of a machine that pays on 3 blanks though has an SD over 9 bets/spin for short coin and 10.8 bets for max coin. :( So playing one of these games may not guarantee a low-variance machine.http://wizardofodds.com/games/slots/appendix/6/
One ideally wants to see more lower paying combinations on the paytable compared to higher paying combinations as well as lower max award values. This will help lead you to lower variance slots just like it does in video poker. But unfortunately, but without exact data like that shown above, its sorta hard to tell.