In world space everything is measured in default meter unit. Terrain, gravity, sound etc depend on it.
In device display those world space coordinates are converted to screen space coordinates.
So you can't place things in pixel. It's a variable depends on device. And world space unit is meter, you have to place things in world space. It's proven best that way. You don't want to develop a game just for one device display.
tansform.position.x is floating point number represents one portion of a 2D/3D vector which is a position vector of an object. These are calculated in world space in meter and projected in display screen in pixels. (If your camera is positioned in the origin in world space it should look like the image).