The Pixelate transition starts with the first video or image, blurs it like the pixelate effect (by dividing the image into squares, and filling each square with the average color of all of the squares), and increases the size of the squares over time. Halfway through (at maximum pixelation), it switches to an equally-pixelated version of the second video, and then decreases the size of the squares over time, until at the end what remains is a perfectly-sharp other video or image.
This value defines the maximum size of the averaging squares which the Pixelate transition uses. The larger the squares, the fewer of them in the video (and the most detail lost).
By default this transition starts from the video in Video Track 1 and ends with the video in Video Track 2. Check this option to transition from Video Track 2 to Video Track 1.