Because pixels are not square. Most of them at least.
Depending on a complex combination of source of the image, components of the projection system, dimension of the screen, standard adopted on that device and kind of image that you are watching the pixels may be square or not, more over square pixels may be converted to non-square and viceversa.
The explanation is a bit complex and I think that the best so far is to read Programmer's Guide to Video Systems and Square and Non-Square Pixels, written by Chris Pirazzi.
Some relevant information from those deep, technical documents are:
In particular, the geniuses who designed the standard definition (480i, 576i) digital electrical standards decided to make every line have 720 non-square pixels, which we'll talk about next, and these are the exact same 720 non-square pixels that you are often forced to deal with by various video devices on PCs.
Fortunately, sanity prevailed for HD (1080i and 720p), whose digital electrical signals all have square pixels.
To the above I'll add, unless you have a WUXGA monitor where the pixels get a bit stretched, getting again non-square pixels.
So, this gives us an incontrovertible way to answer the question of "how non-square are non-square pixels?" If 480i square pixels are sampled at 12 3/11 MHz, and 480i non-square pixels are sampled at 13.5 MHz, that must mean that each non-square pixel has aspect ratio:
12 3/11 MHz / 13.5 MHz = 10 / 11
Another element on the complex equation mentioned, the pixel aspect ratio can be seen in the tutorial Understanding the use of square vs non-square pixels in AE (Adobe After Effects). This article focuses on the pixels itself and not than much on the electrical aspect. In the next quote we can see how some people and programs know about the differences and how to work with it.
The drop dead simple way to work in a square pixel comp is to use 648 X 486 for your SQUARE PIXEL project and then just re-size to 720 X 486 NTSC/DV aspect ratio comp for rendering.
Another reference is found on the [Apple - Pixel Aspect Ratio] part of the manual of Final Cut Pro, where we can read something like:
The ITU-R BT.601 specification makes it possible to transmit either NTSC or PAL information in a single signal. To achieve this goal, both NTSC and PAL video lines are sampled 720 times. In both NTSC and PAL video, the frame displayed has an aspect ratio of 4:3, yet neither 720 x 486 nor 720 x 576 constitutes a 4:3 ratio. The solution to this problem is to display the pixels (the samples of light intensity) taller-than-wide, or wider-than-tall, so that they fit into a 4:3 frame. This results in the concept of “rectangular pixels”—pixels that must be stretched or squeezed to fit in the 4:3 frame. Most SD video devices actually use 704 or 708 pixels for picture information but stretch these pixels to 720 when recording to tape.
Finally, we can't avoid at least one mention of Wikipedia, so here we have an article about Pixel aspect ratio which mentions both aspects, electrical signal and device resolution. The talk section of that page is long, but has some interesting comments.
You may have square pixels on a square capable device, those will look square and everything will be OK, but any other combination, will make things non-square, and considering all the involved elements, it's not surprising to have non-square pixels most of the time.