You have seen this many times before: a baseball team, leading in the late innings of a close game, will take out one of their players who has defensive deficiencies and replace him with a player who is better at defense but not as good of a hitter. The idea is, now that we have a lead, we will place a priority on preventing the other team from scoring rather than on scoring more (unnecessary) runs ourselves.
I would put it this way: we have a line-up on the field that has demonstrated it’s superiority to the other team by scoring more runs than they do, so let’s alter that line-up. Let’s remove a player that helped give us that lead and replace him with a player who is less effective.
We know he is less effective because otherwise he would have started the game.
Does this make sense? It makes as much sense as the guard against doubles defense late in the game and the sacrifice bunt. In other words, it doesn’t make sense.
You have an offensive player who, say, bats .280 and walks a little, but costs you some defense. He’s not going to miss every defensive play. Nor is the defensive player going to hit .100. Just how often will the difference in defensive capability matter? At third base, it will affect a small number of hits that are slightly out of the reach of an average or below average defender. But then, the third baseman is likely to bat once in the last 3 innings– perhaps twice. How much difference will an extra .100 make on the offense?