Some Particularly Vexing Problems in Experimentation on Reading

Since human behaviors are multiply determined and the best reading researchers can do is to determine "contingently necessary causes" for observable phenomena (and that only probabilistically), it is imperative that they employ sound research strategies. Of the many pitfalls in research, three have been selected for special consideration: the frequent use of statistical analyses which at worst are incorrect and at best fail to utilize the full power of the experimental design employed; the failure to delineate clearly just what experimental treatments have been employed; and the use of control groups which do not "control" in any meaningful way. Some consequences of these errors are indicated, and some ways to minimize them are suggested.