According to Baseball Reference, in 2016 there were 127 plate appearances containing sacrifice Bunt Attempts at-bats with 3rd base occupied and 1 out (precisely).
Of those 127, 61 RBIs were recorded on 56 successful SHs (not all of which recorded RBIs, 1st/3rd can still be a SH when 1st advances to 2nd). 3 times a runner reached on error. 20 hits were recorded, meaning the player was attempting a sacrifice bunt but reached base without an error or a fielders' choice.
11 strikeouts were recorded on sacrifice bunt attempts, 2 double plays, and a total of 51 non-sacrifice outs (meaning they did not rule it a successful sacrifice - so it was 71 "at bats", 20 hits 51 outs, and 56 sacrifice hits which don't count against ABs.
The overall WPA (win probability added) was -1.5 for the season (meaning about -0.01 per attempt). RE24 was a -15 (meaning the overall run expectancy for the season was that they scored 15 fewer runs than they would have otherwise).
This data isn't perfect, it's not really possible to get perfect with the free BR data, but what it tells me is that squeeze bunts are slightly negative but close to neutral - and in situations where you only need one run, they may be useful.
Limiting it to situations that are clearly squeeze attempts (3rd or 2nd/3rd only) and adding a few years' data, you have 198 attempts from 2011-2016 with 2nd and 3rd or 3rd occupied (not 1st) and 1 out exactly.
122 RBI were recorded, on 99 SHs and 27 hits. 22 times a FC/SH was recorded without recording an out (meaning they attempted to throw the runner out at home and failed), and 7 times the runner reached on error, meaning 22+7+27=56 times no outs were recorded and 198-56=142 times an out was recorded (and three of those 142 times, 2 outs were recorded). Total RE24 for the period was -7 (again, meaning this was more or less even with not doing it, but very slightly negative).
122/198 is not bad, 61.6% success rate - far higher than most things in baseball really on the hitting side. However, because the run expectancy with runner on 3rd, 1 out, is 0.9223, and 1.32 with 2nd/3rd occupied, and then the run expectancy with 2 outs is very low (0.36 for 3rd, 0.57 for 2nd/3rd), you give up a lot (more than half) of run expectancy for a failure even if its just the runner is thrown out at first.
My interpretation of this: if you need one run (either to tie or win) and is late in the game (9th, or maybe 8th), and have a competent bunter, it's probably worth doing it - the reason the RE for the play in general is so high is because you can score several runs, especially 2nd/3rd.
But in the 2nd or 3rd or 4th, or even 7th? You're preventing yourself from having a big inning by giving up the out, and you're only scoring a run about 2/3 to 3/4 of the time anyway. It's less dumb than a regular sacrifice bunt with one out, which takes you from RE of 0.526 to RE of 0.317 even when it succeeds (comparing on [this page] the RE of 1-0-0 with 1 outs to 0-2-0 with 2 outs), but it still doesn't really look like a good idea.
Add to that the specifics of the players. Jayson Werth was up for two of these opportunities, and while he's not the player he once was, he is probably more likely to have a SF than to Sac Bunt successfully. Lobaton was up for the one in the 2nd inning, and while he does seem like a better choice for the squeeze bunt, he's also the 8th batter (meaning Scherzer was up after him), so the RE for the 39% of the time he fails to score an RBI is nearly 0.