And a bonus for non-official MBTI instrument,
-the MBTI instrument is meant to be administered by a trained professional
These people verify your results through interaction post-test. While these people can't catch everything (e.g. an INTP highly interested in the human condition who becomes very expressive and friendly in order to connect with people so that they may help his voracious need for understanding. Essentially a case in employing Fe in the interest of Ti, though this is highly unstable and will likely collapse within a month. But I digress...), the validation catches most things such as an uncanny indecisiveness from an "ESTJ" or a highly opinionated "ISTP."
That being said, people who are in denial will also have great difficulty getting an accurate reading anywhere, both through test taking and studying the theory. So studying the theory behind the test isn't exactly a cure-all either. Hope that helps.
-test-takers can be confused as to what the questions are asking
Such as typical behavior vs. preferences (possible, though such a person is likely miserable if they do not coincide) and perhaps ambiguity due to vagueness of the question (i.e. one likes people who are empathetic in some situations and efficient in others)
-subjects who have taken the test once can easily manipulate the results
One can look at questions and tell what they intend to measure. People who cling to a type (e.g. one thinks INFJ is cool and caring and mystical, so she stretches the truth a bit to bias the test towards INFJ, despite being ISFJ or something). Or someone is thinking about how someone is different from them and answers relative to someone else (i.e. one INTP compares himself to another INTP friend, one who is actually antisocial, and unconsciously misclassifies himself to distance himself from that)
There are a few problems with the MBTI instrument (and a few more with online imitations):
-the test attempts to deduce multiple interacting preferences at a time
For example, some introverts might be "to the point" with people because they are introverts, not because they tend to use thinking processes. Likewise, an ISFJ might not agree with either choice ("here and now" vs. "what lies ahead") because he's focused on the past (Si vs Se) and thus chooses "what lies ahead"--an 'N' trait--due to lack of a better option
-the test can be easily confounded through the subject's mood
Because it measures preferences, people who are feeling rather "philosophical" at the moment may be introspecting much more than usual and thus not accurately describe their natural preferences. Similarly, single-minded people might be focused on something that makes them behave against their natural preferences to achieve their goal.
Talking about my sig? I deliberately made it other than INTP to make a statement.
I think I'm INTP because I have difficulty with arbitrary decision-making (sometimes use algorithms). Also, I work up until the deadline on papers not because of procrastination but because of excessive analysis, never picking an idea until I absolutely must (there's always a better idea around the corner).
I see schedules more as suggestions, a cure for not knowing what to do. If anything, I'm more likely ENTP than INTJ considering I love new ideas more than pretty much anything else. I haven't found a group of N-types, so I don't know if I'd like to hang out with groups. I'm pretty expressive and am usually bored rather than drained from social interaction, but if no new ideas are to come from it (aka drinking party) I'll take books to people any day. Definitely introspective though.
Does that answer your question or is there something else about my type that seems off?