Some like to say that "statistics is the best way to lie". I believe that, while not telling the whole story, statistics can be a useful "tool", if one knows how to interpret them objectively. There are also some things in football which can't be calculated by numbers, such as off the ball movement, good positioning and the passing/ball carrying ability to connect the lines.
If you take a closer look to those stats, you'll see that key passes+assists=chances created. Which probably means that, according to squawka, a forward pass from a deeper position that gets the ball out of the defense and initiates an attacking play isn't considered a key pass statistically. key pass is a pass that creates a clear cut chance for a teammate. The ones that lead to goals are counted as assists. They both have one thing in common, they usually "happen" in the final third.
It's only logical that the holding midfielder won't get as many chances as the other midfielders to get in more advanced positions. His main job is to provide cover for his teammates. Carrick's vision and excellent passing range helped us control the tempo and improve our build up play but that can only be "proved" statistically only when you take a look at our win ratio last season with and without him on the pitch. It can be seen as a paradox but i believe it's no coincidence that all these sites mention key passes as one of Carrick's best attributes, despite their numbers telling a different story.
On the other hand, the stats show that Fellaini, although not the best player on the ball, did play his part in the final third. Not by providing assists (he has zero) but by being the only player who could hold onto the ball in the final third and generate movement from his teammates in the attacking half.