Well, the short answer is we don't know.

Most often the only part of the animal that is preserved is the bones and teeth. With so little information being fossilized it is very difficult to answer some questions about their lifestyles or their color.

What we can do is a little bit of educated speculation, though this will not tell us what the REAL color of a dinosaur is, it can at least give some probably colors so that artists can reconstruct them.

I think that most depictions of dinosaurs are influenced greatly by what we know of modern reptiles. The comparison in many ways is a natural, unavoidable bias of sorts, but one that seems reasonable, given that one could expect that dinosaurs would in fact be similar to modern reptiles in many ways.

Analysis of bone structure offers some indication of what a dinosaur looked like in terms of musculature and shape (similar to how forensic scientists today may reconstruct what a person probably looked like based on skull characteristics).

Certainly our concept of what dinosaurs looked like in terms of color is almost certainly a matter of speculation based upon the colors of modern reptiles. The colors found in drawings of dinosaurs are probably in large part the result of artistic impression, to add a touch of realism in order to better visualize and appreciate what they might have looked like. In some ways, this is similar to how we use color to illustrate proteins, cells, organs inside the body, etc. In order to provide more comprehensible and interesting visualization models.

However, just as no biology textbook would actually describe the nucleus of a cell as being orange despite having drawn it in that color, I would be surprised if specific color characteristics were included in descriptions of a dinosaur. Certainly to draw such conclusions about color would seem somewhat irresponsible, unless very clear evidence pertaining to color was cited. My understanding is that in some rare cases some of the skin pigments are preserved with fossils, in which case such a conclusion is actually possible.