In the physical sciences, a model is developed in order to describe a phenomenon. Music theory is a model. It is used to interpret music in order to better understand how it is perceived. In empirical science, models are developed by looking for patterns. Music theory is no different. We see patterns like major chords or the perception of keys and we gives names to concepts and investigate their appearance.
Unfortunately, music theory is rarely presented this way. Instead, it is presented as a set of guidelines to which you may or may not conform. This is not how experienced theorists look at things (sorry for the appeal to authority). The misconception starts with the most basic concept, the major scale. The major scale is used to describe a fundamental "bias" in note choice. You're more likely to see a note in the scale rather than out of it. Thus, when composing, the note you want to use is probably in the scale. If someone says that using a note out of the scale is "theoretically wrong" then they are misconceived. We can describe out of scale notes by saying where out of the scale they are, and the case is closed. This method of seeing notes based on a hierarchy centered around the major scale is called tonality. It is the basis of our model of music.
Another huge misconception comes with chord progressions. Often times the language in articles contributes to this. "These are the chords you can use." What a bunch of crap; you can use whatever chords you want. No one's going to stop you. Instead, when diatonic chord progressions are presented, it should be noted that they are the chords that belong to the major scale and consequently are the most common. Chances are, when you're looking for the right chord, it's going to be diatonic. Using a non diatonic chord isn't "wrong," it's just less common.
Once you stop looking at theory as rules or guidelines and start looking at it is a tool that shows patterns, you'll find that you can analyze music much more clearly and write music more proficiently. Also, to digress for a second, theory is not necessary. It is only a tool! It's a damn good tool, but a tool nonetheless.
