It's pretty much compulsory knowledge for baseball fans that relievers' ERAs aren't on the same scale as starters' ERAs. A good reliever, while usually a worse pitcher than a good starter, will generally have a better ERA. There are a number of reasons for this, but the primary ones are that hitters fare substantially better each subsequent time they face a pitcher in a game, so facing a hitter only once a game suppresses overall production off a pitcher, and that relievers can concentrate all their strength and focus into a single inning in most cases without having to worry about pacing or setting hitters up for later at bats. It is also true, however, that not all relievers' ERAs are on the same scale either. There are some cases when a pitcher has additional factors depressing his ERA, the most consistent and extreme such example being the situational lefty.
One advantage in the situational lefty's favor is the platoon advantage. He faces a disproportionate number of favourable match-ups. In 2008, left-handed hitters OPSed .077 points lower off of left-handed pitchers than right-handed pitchers, so that's a pretty significant advantage to have. This advantage is pretty well-known, as it's basically the entire reason for having situational lefties in the first place.
A less obvious but potentially just as important advantage, however, is that one/two-out pitchers often find themselves coming into games more often with one or two outs and less often with no one out. Every hit or walk allowed is less harmful the more outs there are at the time: for example, in 2007, a runner was likely to score from 1st base 40% of the time with no outs, but only 13% of the time with 2 outs. Since a pitcher's ERA only covers what happens to the baserunners he allows, coming into the game with 2 outs is like pitching an inning in which the opposition only gets 1 out as far as his ERA is concerned: the batters he faces only get 1 out to cross the plate before his slate is wiped clean instead of 3.
To see just how big a difference this makes, let's look at two case studies: one of a good left-handed specialist, and one of a bad one, and see how their ERAs would look adjusted for these factors. To do this, I looked at the baserunners given up by each pitcher distributed over the number of BF with each number of outs and with each platoon situation and then looked at how much the expected runs allowed would change with those baserunners redistributed over an average number of BF with each number of outs and an average platoon split.
214 BF, 50 H, 8 2B, 1 3B, 0 HR, 21 BB, 3 HBP
0 out: 63 BF
1 out: 80 BF
2 out: 71 BF
vLHB: 100 BF
vRHB: 114 BF
As you can see, Beimel actually faced more righties than lefties, though still a higher percentage of lefties than average, and had a moderate advantage in pitching late in innings. Based on these numbers, Beimel gets a modest drop in ERA. His ERA, assuming he allowed hitters the same rates of production, would have been about 5.0% higher had he been a standard 1-inning reliever, with about a 60% of that difference due to pitching mostly later in innings and 40% due to the platoon advantage.
131 BF, 34 H, 8 2B, 1 3B, 2 HR, 20 BB, 1 HBP
0 out: 36 BF
1 out: 44 BF
2 out: 51 BF
vLHB: 64 BF
vRHB: 67 BF
Like Beimel, Flores faced more righties than lefties, but still more lefties than average. He had a slightly better platoon advantage than Beimel as well as a bigger advantage in appearing later in innings. Converting him to a 1-inning pitcher bumps his ERA 9.4%, with about two-thirds of the difference coming from the number of outs when he pitched and one-third from the platoon advantage.
As one would expect, the bad reliever was used in more situations that reduced the effects of his poor pitching, namely later in innings where his baserunners had a smaller chance of scoring. This is one measurable area of the elusive managerial value-managing the bullpen to reduce the negative effects of your worst pitchers. In Flores' case, Tony LaRussa saved his team 1.5-2 runs by how he distributed Flores' innings compared to an average distribution, which sounds like nothing, but that's just from redistributing 25.2 innings. Without context, it's not clear how substantial this type of bullpen management is, but with enough legwork, it would be possible to compute these kind of numbers for entire staffs and start to get an idea of which managers are best at it. For now, we can use these numbers instead to better consider how relievers compare to each other when we look at their ERAs.