What are the hallmarks of quality? And how can a novice distinguish the truly fine from the very bad? These questions are paramount in the realm of scientific discovery, because of the importance we late-20th century folk attach to science.
In a preliminary attempt to describe bad science, expert junk science detector Steven J. Milloy offered this: "To paraphrase what Supreme Court Justice Potter Stewart said about obscenity..., 'I can't define junk science, but I know it when I see it'." Milloy has also provided this description, "'Junk science' is bad science used to further a special agenda."
Good science is much easier to define. Quality research is a dedicated search for truth that can be discovered through scientific methods. Good science begins with clear definitions of terms, discernible separations among categories under investigation, comprehensive discussion of assumptions underlying the research and theoretical constructs operationalized -- to set forth the questions the research is designed to answer. A good piece of research also places itself accurately within the context of prior literature and ideas, specifying with exacting detail the methodological procedures used to collect information relevant to the question being examined, so that subsequent researchers can replicate the study. Finally, good science presents a complete description of the information collected, and those conclusions and varying interpretations that may be derived from the data to answer the questions posed.
Good public policy follows good science, but we must be especially wary of science that seems to follow politically favored but heretofore unsubstantiated positions. The consequences are bad enough when political candidates find a parade and rush to the head of it; worse, when scientists follow suit.
Politicized science is treated as a techno-talisman, with magical powers capable of shutting down all debate, rendering opponents speechless. After all, statistics, as numbers, are nearly impossible for easy, immediate refutation or challenge, and technical procedures are usually treated as appropriately being the sole province of expert practitioners. This tends to render science(whether well constructed or of the junk variety) beyond the boundaries of democratic discourse, but it should not be so.
Around 25 years ago, cognitive psychologists Amos Tversky and Daniel Kahneman described the typical heuristic rules people use to reach judgments in conditions of uncertainty, such as in reasoning about science. Their rather astounding findings included the discovery that even scientists are poor in reasoning from statistical evidence that should override common patterns of probabilistic thinking. Indeed, like nonscientists, trained scientists often fail to use statistics correctly to draw valid inferences. Thus, nonscientists should not be daunted at the prospect of debunking dubious research claims.
The most pernicious current instances of junk science are "studies" related to the environment (particularly about global warming), gun control, sexual identity politics, and public opinion research. We have previously discussed in detail the junk science applications of poor-quality opinion surveys (on "Pollaganda," 98-40), and we will touch on the others briefly.
Claims that global warming is a settled fact are overblown. Different temperature measures have produced different indications about possible trends in global climate change. Furthermore, natural patterns of variability in world temperature may falsely suggest trends over a relatively small segment of measurements. (As Tversky and Kahneman noted, observers tend to see any research sample of any size as overrepresentative of the entire data set; thus, people would too easily believe a string of high temperatures proves warming trends, when such an observed pattern could be wholly within natural variations.) And commonsense observers rightly ask whether even cumulative effects of manmade phenomena could possibly match the scope of such frequent natural disasters as earthquakes, volcanoes, hurricanes and tornadoes.
Junk science studies of gun "violence" are prone to using shifting categories and definitions to make invalid points favoring more gun control. A typical trick is mixing the categories of young children, violently predatory teenage gang members, and even adults, to suggest that accidental shooting deaths and injuries among children are rising -- when, in fact, the opposite is true. Good science in this field is of the sort conducted by John R. Lott, Jr., who has found that increasing gun availability to law-abiding citizens decreases levels of crime (including murder and rape). The explanation is simple: Criminals and intended victims are rational actors. Criminals are less likely to attack, the more likely a possible victim could possess firearms; armed victims are more prone to use their weapons, the more threatened with serious harm they feel. (This and similar research indicates that more Americans will die and be injured than would otherwise be the case, should Congress pass the gun control legislation now under consideration.)
Even more egregious examples of junk science have appeared recently in studies related to sexual identity politics. Great fanfare accompanied the announcement of small studies that suggested genes linked to homosexual behavior (so-called "gay genes") had been observed. A larger, better study published in the journal "Science" this month stated that study "data do not support the presence of a gene of large effect influencing sexual orientation."
In the July 1998 Psychological Bulletin article, "A Meta-Analytic Examination of Assumed Properties of Child Sexual Abuse Using College Samples," published by the American Psychological Association (APA), authors Bruce Rind, Philip Tromovitch and Robert Bauserman purported to have found that sexual abuse of children is not always psychologically harmful, particularly if the child was "willing" and "consented" to the abuse. The authors suggested that pedophilia be judged wrong only if the victim considered it damaging. Among this study's many monstrous design flaws, is limitation to college students, excluding entirely those abuse victims from "clinical" and "legal" samples. These researchers removed the most severely abused victims from their study, queried assessments of lasting psychological harm from generally well-adjusted people, and concluded that child sexual abuse may not be seriously damaging. Not a surprising finding, and not at all valid.
Junk science's overreach exceeds its grasp. In the hands of postmodern deconstructionists, all science is junk science...and just another tool of combat in the political and culture wars.
Junk science follows its own myths and offers a new faith. This crucial point is rarely admitted during either lay or expert discussions of scientific findings: The ways we think about science are not themselves scientific. And that places science squarely within the bounds of suitable lay discourse.
The value of science is properly judged on extra-scientific grounds. Just consider: In the Star Wars movies, which is more compelling -- the technological wonders on display, or the fight between the forces of good and evil?