What are some typical errors that a researcher should look for when reviewing journals?

One error I encountered recently was in a magazine article, “Are Fathers Necessary?” in Atlantic Magazine.  (http://www.theatlantic.com/magazine/archive/2010/07/are-fathers-necessary/8136/)

The article appears to be about an idea that is gaining more widespread acceptance: that a having a male present while raising children is statistically linked to better outcomes for those children.  I will not rehash the arguments.  Obviously, there is some inherent appeal to those of us who believe that the average male and the average female learn and, to a surprising extent, think and behave differently.  Not better or worse, but differently.  This idea that fathers are necessary is very important because fewer and fewer homes with children actually have fathers or adult males present.  Of course, it is simply not possible in many cases to have males present, but that is not the discussion.  Like so many difficult questions in science, the “problem statement” is about what happens when fathers are not present, for better or worse.

Anyway, this is very threatening for a lot of people, particularly some single mothers who feel this disrespects the herculean effort they make every day to be the best parent to their child they possibly can be.  Apparently, it also bothers some radical feminists and lesbians who believe in the old Wellesley College motto: “A woman without a man is like a fish without a bicycle” and do not like to see themselves contradicted.

Therefore, this brings us back to the Atlantic article.  It purports to question whether the absence of fathers does statistically lead to poorer expected results for children, but what it actually argues is mostly that two lesbians can raise a baby successfully.  That is great info, but largely independent of the question the article claimed to address and claimed to debunk.  Unfortunately, in spite of the logical argument that one cannot be true if the other one is, in real life the two questions are largely independent from each other.  Proving one does not disprove the body of evidence in support of the other.

Therefore, one error to look for is articles that purport to disprove something but which instead simply prove a different point, leaving the inference but not the proof that the initial argument is wrong.