37. Diffusion and Adoption of Educational Technology: A Critique of Research Design

Robert E. Holloway

Northern Arizona University

The body of research reviewed here is divided into three logical categories: (1) descriptive surveys of equipment diffusion, (2) correlational studies of technology and demographic variables, and (3) constructivist and action research on adapting technology. It is assumed that the diffusion process is similar to a market model based on a knowledge cycle (Rich, 1981). This cycle, somewhat simplified, moves through invention, commercial production, and eventual sale to secondary markets. Technology is developed for other markets first. Education is a secondary or tertiary market where technology is adapted. Technology, used in this context, is a combination of equipment and its use. Learning, except for activities such as in-service training, is not a dependent variable in these"studies. Neither is the research driven by theory. Taking one area, the literature on microcomputer adoption, Gillman (1988, p. 192) concluded that:

It is notable that few of the studies in the 8 years of research under review were theoretically based or utilized a methodology based on multivariate analysis. The breakdown of the types of research pursued in the 70 studies reviewed was as follows:

Case studies, 11
Exploratory field studies, 21
Surveys, 21
Ex post facto (quasi-experimental), 12
Experimental (controlled), 2
Reviews, 3

An unevenly distributed but full range of research methods exists among studies of diffusion and adoption. Survey studies are over represented, likely because they are inexpensive and simple compared to experimental studies. Most federal longitudinal studies of all kinds are of short duration because of political and economic factors. Those commercial studies that are available must be compared with caution because there are significant differences in methods used for sampling techniques, instrumentation, and analysis. Kinnaman (1993) suggested using multiple studies for decision making, since reported research represents a wide range of competence, care, and sources. From this and other sources, an overview list of issues can be constructed.

• Research methods in studies of diffusion and adoption reflect the same strengths and weakness as other research foci in educational technology.
• Multimethod studies, needed to provide holistic understanding of diffusion and, especially, adoption, are gradually evolving.
• Results of diffusion and adoption studies are more likely to be used in marketing and journalism in the popular press than results from research on learning.
• Baseline data and tracking of simple indicators of diffusion are sporadic, focused on success, and often do not meet the cannons of research.

The most important distinction among the three categories used in this review is the intent of the research. Survey or diffusion studies are to find what is happening: Who is buying how much of which kind of technology? The most common reporting statistic is descriptive. Demographic and trend generalizability are important outcomes. The correlational studies in this section are designed to find relationships, usually using survey data. The differences are in the additional data collected, such as personality differences among adopters, shifting to measures of beliefs and attitudes, and in generating more testable hypotheses. The last category, action research, has the potential to help us understand why diffusion and adoption occur as they do.

Some of the earliest studies of innovation in education included such variables as school wealth (see Mort, 1938). More sophisticated than this simple description, correlational studies share the goal of generalization about factors associated with purchase and use. The last, action research, centers on the user's perception of the world and is usually less concerned with generalizations. The three categories form a progression. The survey literature tells us what is and perhaps what has quantitatively changed; the correlational studies tell us about the factors associated with change; and the constructivist and action research seeks to understand and bring about change at qualitative and quantitative levels. Of the three areas, the correlational studies are currently the richest in grounded speculation about why technology is in schools and how it is used. Factors affecting adoption, such as district wealth, help explain the presence of technology. Such factors do not address use or adaptation. In contrast, models, such as Morehouse & Stockdill's (1991) five-stage process-41) front-end analysis, (2) prototype development, (3) small-scale implementation, (4) organizational adoption, and (5) institutionalization--or Stages of Concern (Hall, George & Rutherford, 1977) are common ways of describing innovation and planning for intentional change in educational technology. The stages are still descriptive and do not directly deal with attitudes and beliefs effectively. It is likely, as Cuban suggests, that education has a value system that is positive toward personal contact and views technology as impersonal. Dealing with such subjective beliefs is difficult for survey and correlational designs.

37.1 Category Overview

37.1.1 Survey Research

37.1.2 Correlational Studies

37.1.3 Action Research

Action research is designed to bring about change. Constructivist and qualitative work help us understand why adopters do what they do. (It is acknowledged that this category is somewhat contrived.) Bogdan and Bilken (1992), while specifying it is the systematic collection of information, limit action research to social change. Kernmis (1988) defines the term as reflection on praxis. A formal requirement is that truth is determined by the way it relates to practice (p. 43). As used in this review, action research is a synonym for applied research (see 42.2). The social-order concerns remain, but utility moves into a more central position. The perception of the participants remains as a guide to interpretation. Early studies in this category simply reported the effects of training or impact of the technology in the workplace. They were often one-shot case study, workshoplike efforts to introduce change. Earlier work on describing change in a variety of fields led to several models focused on stages of change. The first models of prescriptive change (see, for example, Havelock, 1973) used a process approach and were based on low-level generalizations from research. There was little model-specific empirical evidence. Current work on the process of change has been parsed out in prescriptive stages in a series of studies that has established its own research tradition. Most of these studies are based on Hall and Hord's (1987) concems-based approach to change. Some of these are specific to technology, such as Hall and Win's 1981 study of implementation of computers in classrooms. More studies are being done with the intent to modify practice through an iterative intervention focus on in-service training using variations of Hall's "Concems-Based Adoption Moder' (C-BAM) as a guide. (See, for example, Winner's 1982 dissertation "Introducing the Microcomputer into the Elementary Classroom: An Inservice Program for Teachers.")

Policy, gender, and equity studies are also included in this category. (See, for example, Turkle & Papert's "Epistemological Pluralism and the Revaluation of the Concrete," 1992.) An illuminating subcategory, placed here because few studies are available, is qualitative research focused on perceptions of adopters. For example, Mehan's 1989 study uses qualitative data to speculate on the weight of social practice in the use of microcomputers in classrooms. A more basic study is Nordenbo's (1990) description of how computer novices perceive information technol' Often ogy. these studies emphasize a constructivist perspective and grounded research in a context that is naturalistic.