The dangers of generalizing from comparative studies of how different groups of people learn, remember, conceptualize (think, that is), much less the dangers of making educational or social policy based on such data, are documented in detail by Professors Cole (Communications, Univ. of California, San Diego) and Means (Educational Psychology, Univ. of Maryland). They first review the assumptions underlying traditional experimental research in psychology, then show how comparative studies typically violate the logic of such experimental designs. Most importantly, they point out that ""all other things cannot be equal"" when groups are compared, no matter how carefully the researcher thinks he has matched them. Educational variables, cultural differences, physiological variation, even historical events may contribute to variation in subjects' responses to instructions, or to the task itself, and certainly can influence their ultimate performance. The authors therefore suggest alternate research strategies--among them scrupulous analysis of the processes involved in the tasks being studied, with comparisons of patterns of performance (not the usual search for ""better"" scores); or comparisons of groups and tasks, with systematic variation to eliminate a possible explanation of varying performance levels. In addition, they describe four approaches in which subjects' performances are compared with a theoretical model which has predicted particular outcomes. The final chapter will be of particular interest to the lay reader concerned about the leap from laboratory to life--notably, for its discussion of the dubious uses of research findings to explain (or attempt to change) spontaneous and varied real-world behaviors. A valuable study for educational researchers and users of educational research--which one need not have a statistical background to comprehend.