It depends on how you look at it. The difference in effect is approximately linear in terms of pops/team/season. But if you're talking in percentages, it's a curve.
You can estimate it this way: each trainer level + or - yields about 1 or 2 more/less pops/team/season (for all 18yo's and no team training). So, say you're able to get about 36 pops/team/season with a lvl5. You'll get about 34.5 pops/team/ssn with a lvl4, 31.5 with a lvl2, or 39 with a lvl7. So the difference is linear.
But if you think about it as percentages going from a lvl2 to lvl4 is an increase of 9.5% but going from a lvl4 to lvl6 is an increase of 8.7%.
The differences are small, but when you compound them over time, they can build into significant ones. When I was making the Training Simulator, I checked on this a little. And if you're building an elite player it can make a big difference. Say you want to make an elite LI Guard and start with a 50 TSP 18yo. If you use a lvl6 trainer you can get a 124 TSP player in 7 seasons. And if you use a lvl4 trainer that same player is only 118 TSP. Those last 6 skill points can make a huge difference. The first player is world class, and the second is a good D.1 player. But most of that difference (4pts) is made in the first 4 seasons.
The kicker is that the cost of a trainer goes up exponentially. So the return on your investment is a pretty steep downward curve as you increase your trainer's level.
What ends up having the largest effect on your training is your ability to get trainees the full minutes each week. And unless you're training an elite player, you don't need to blow all your cash on a high level trainer.
The old adages about training are still valid:
1) Get your player 48+ min in the right position.
2) Speeding up training is most important when you have young (21 or younger) players.
3) Did I say get you player 48+?