There are all kinds of problems with looking at a stat like that in a vacuum, Russ. For one, it's not pace-adjusted. For instance - points per 48 minutes isn't pace-adjusted, like points per 100 possessions would be. But, hey, apples... oranges... it's all fruit.
82 games has another pair of stats listed under "off" and "def" that shows the number of points per possession while that player is in the game. It shows that the Spurs offense scores 1.05 points per possession while Splitter is in, but the defense gives up 1.06 points per possession. (105 points per 100 possessions scored, and 106 points per 100 possessions given up) So if the Spurs give up more points than they score while Splitter is in the game, how can his overall +/- be positive? The answer is because the Spurs get more possessions than their opponents, while Splitter is in. One stat is pace adjusted, one isn't.
82 games has another stat called "Net 48". Basically, it's the net points (+/-) for a player, per 48 minutes on the floor. Everyone on the team is positive, except for Leonard, Neal, and Joseph. From that we can conclude that by getting rid of those three players, the Spurs would win every game. Hey... maybe that's true. I've just unravelled the mystery of the Spurs' losses through statistics.