Observations and interviews can provide a great deal of descriptive data, but they fall short in the task of explanation. The same is true of information obtained from tests, questionnaires, inventories, rating scales, and other psychometric devices administered in person-to-person or mailed surveys. With proper care, a survey questionnaire consisting of many different demographic and personal-opinion items can be administered in person, by mail, or over the telephone to a sample of persons representative of the population of interest. The obtained responses to the various questions and scores computed from them can also be correlated with measures on other variables or criteria, and appropriate conclusions can be drawn from the results. But because possible extraneous variables that may affect the responses have not been controlled for, it is not appropriate to draw cause—effect conclusions. If the correlation between two of the variables, say A and B, is significant, we can predict the occurrence or score on one variable from the other variable, but we are not justified in saying that variable A causes variable B or vice versa. The latter conclusion requires conducting an experiment in which A is the independent variable, B is the dependent variable (or vice versa), and, by means of random assignment or matching, other (extraneous, confounded, concomitant) variables are controlled. It is possible to conduct experiments in certain kinds of developmental studies, say, in a study of the effects of a drug or special training on memory in four different chronological age groups, but experimentation is definitely limited as a method for studying human development. For this reason, special developmental research methods and quasi-experimental procedures have been devised.
Was this article helpful?