In lieu of an abstract, here is a brief excerpt of the content:

The Review of Higher Education Summer 1982, Volume 5, No. 4 Pages 197-231 Copyright® 1982 Association for the Study of Higher Education All Rights Reserved PERSPECTIVES ON QUANTITATIVE ANALYSIS FOR RESEARCH IN POSTSECONDARY EDUCATION Ernest T. Pascarella The “ grievances” (some justified, others not) against quantification in the social sciences are legion (e.g., Rist, 1980; Guba, 1978; Hamilton, et al., 1977; Hamilton, 1976; Popper, 1968). Most of these criticisms of quantifi­ cation, however, can be clustered into three general areas. First, the process of quantification often results in the potential distortion of “ truth.” Consider, for example, the use of the arithmetic mean versus the median to indicate the typical percent of white students in each school in a large “ integrated” metropolitan school district. Because of the distorting characteristics ot skewed distributions, a few schools with 90-95% white students in a population where nearly all the remaining schools have less than 15% whites, will tend to inflate the arithmetic mean inordinately. Thus, the mean or average percentage of whites in each school in our hypothetical school district might be 30 percent, suggesting reasonable progress in racial integration. However, the median, or 50th percentile (a more appropriate measure of central tendency in severely skewed distributions), might be only 10 percent. This obviously suggests a somewhat different picture. Of course both measures of central tendency (mean and median) are “ ob­ jective” in the sense that they are calculated by standardized and verifiable methods. What is not necessarily objective, however, are interpretive issues that develop out of the various uses or misuses to which the data may be put. The bottom line appears to be that, while statistics don’t lie, people who use them are not immune to stretching the truth on occasion. A second general criticism is that the process of quantification is often overly reductionistic. In attempting to quantitatively portray some important phenomenon (e.g., effective undergraduate teaching), we must often reduce Ernest T. Pasearella is professor of education. University of Illinois at Chicago. 197 198 The Review of Higher Education it to measurable components or variables. Indeed, many of the more sophis­ ticated multivariate statistical routines, such as principal components analysis, factor analysis, and canonical correlation can be viewed as data reduction techniques (Cooley and Lohnes, 1976). The danger, of course, is that in putting data in a form which is amenable to statistical analysis, or indeed choosing only those variables which can be measured statistically, we may radically alter the very nature of the phenomenon we are attempting to explain. Consequently, while we may statistically account for an impressive percentage of the variance in some variable, the variable we are predicting may have little significance in terms of theoretical or policy implications. Thus, as Fincher (1978) suggests, the use of sophisticated statistical routines yields little unless we have valid and comprehensive measures of the social/behavioral constructs we are attempting to explain. Of course validity/reliability issues are an important consideration in all forms of social and behavioral inquiry, qualitative as well as quantitative. One important strength of quantitative methods is that they provide a sound basis for objectively estimating the amount of error in the data. The third general criticism of quantification in social science inquiry is related to the second. Ironically, this criticism seems to stem form the apparent status that quantification lends to a discipline. Just as mathematics was a key element in the paradigm development of the natural sciences (Kuhn, 1962), so, too, it would appear that advancements in psychometric and statistical knowledge paralleled the development and acceptance of psychology and sociology as “ scientific” disciplines. Unfortunately, the very stature which quantification lends to a scientific discipline can often itself be mistaken for the act of “ doing science.” Thus, the third and perhaps most subtle criticism of quantification in the social sciences is that the very act of quantifying often lends a false veneer of scientific respectability to social or behavioral inquiry that may be of little or no scientific merit in other respects. This sense of respectability derived from the quantification of information is suggested by the extent to which the educational research community in general is perhaps unduly impressed by large sample...

pdf

Share