In general I'm very sympathetic to this point of view, and there are some good examples in your post.

One bad example, in my opinion, is Eliezer's recent procrastination post vs. the survey of "scientific research on procrastination." I read the chapter, and it appears to mostly cite studies that involved casual surveys, subjective description, and fuzzy labeling. Although there are many valid scientific endeavors that involve nothing but categorization (it is interesting to know how many species of tree frog there are and what they look and sound like even if we do not make any predictions beyond what is observed), categorization should at least be rigorous enough that we can specify what we expect to see with a modicum of precision.

When a biologist says that frogus neonblueicus has neon blue spots and chirps at 500 Hz, she will give you enough information that you can go to Costa Rica and check for yourself whether you have found one of the rare neonblueicus specimens. Although there will be some controversies around the edges, your identification of any particular frog will not correlate with your political biases or personal problems, and repeated observation of the same frog population by a few different researchers will tend to decrease error.

When a psychologist says that procrastinators can be divided into "relaxed" types and "tense-afraid" types, the "science" being done is not merely descriptive, but also horrifyingly vague. What does it mean for a human to be "tense-afraid" when "procrastinating"? The three paragraphs or so of context on the topic give you enough of an idea of what the researcher is saying to conjure up a mental image, but not nearly enough to carve thing-space at the joints.

In my experience, this is a very serious problem in social and human sciences -- there are whole subfields where the authors do not know how little they know, and proceed to wax eloquently about all of the empty concepts they have coined. There are other subfields where the researchers suspect that they might not have done very good research, and they cover their tracks with advanced statistics and jargon. After you dig through a few of these booby-trapped caves of wonder, you start to lose, if not respect for scholarship, at least some of the urge to do the moderately hard work of digesting literature reviews yourself on a regular basis. It is dangerous to assume that casually studying the leading textbook in a soft field will usually make you smarter.