You've done a ton of work on this, and I think without a doubt you've shown that teams that win the first half are more likely to win the game. Getting off to a strong start in a game offensively helps you win more games, because it means you have more points. While this gives the defense the advantage of a cushion, the data sure seems to show that playing with a lead didn't improve their performance on any given drive last year. By scoring more points, you've improved your team's chances of winning, but improving your team's chances of winning isn't the same as improving the performance of your defense. Maybe with a greater sample size, we'd see some defensive improvement. With a greater sample size of Dallas playing with leads, we'll see more wins either way.