ROUNDTABLE
TITLE
Integration of Science and Practice in Industrial/Organizational Psychology
ABSTRACT
The objective of the proposed roundtable is to engage interested parties in a critical review of the historical and recent literature, with the aim of identifying and analyzing studies that best exemplify the integration of rigorous science with powerfully effective practice, and subsequently to conduct further work in that vein.
PRESS PARAGRAPH
Industrial and organizational psychologists conduct scientific research on human behavior in the workplace, and they apply their discoveries to the improvement of worker productivity and satisfaction. Ideally, science is experimental, objective, quantitative, and skeptical. Practice, however, sometimes has to rely on common sense and experience when there is insufficient time or resources to support research . The objective of the proposed roundtable is to initiate a critical review of the literature, with the aim of analyzing those studies that best exemplify the integration of rigorous science with powerfully effective practice, and subsequently to conduct further work in that vein.
The objective of the proposed roundtable is to engage interested parties in a critical review of the historical and recent literature, with the aims of identifying and analyzing those studies that best exemplify the integration of rigorous science with powerfully effective practice, and subsequently conducting further work in that vein.
Taylor’s (1911) shoveling experiment is a prototype of the integration of science and practice. He described it as follows, in his Principles of Scientific Management:
For a first-class shoveler there is a given shovel load at which he will do his biggest day’s work. What is this shovel load? Will a first-class man do more work per day with a shovel load of 5 pounds, 10 pounds, 15 pounds, 20, 25, 30, or 40 pounds? Now this is a question which can be answered only through carefully made experiments. By first selecting two or three first-class shovelers, and paying them extra wages for doing trustworthy work, and then gradually varying the shovel load and having all the conditions accompanying the work carefully observed for several weeks by men who were used to experimenting, it was found that a first-class man would do his biggest day’s work with a shovel load of about 21 pounds. For instance, that this man would shovel a larger tonnage per day with a 21-pound load than with a 24-pound load or than with an 18-pound load on his shovel. It is, of course, evident that no shoveler can always take a load of exactly 21 pounds on his shovel, but nevertheless, although his load may vary 3 or 4 pounds one way or the other, either below or above the 21 pounds, he will do his biggest day’s work when his average for the day is about 21 pounds.
This excerpt illustrates five features of a practical and scientific approach to the study of to work behavior:
1. One aspect of the environment, shovel size, was manipulated over a large range of values, while other variables where held constant.
2. The employee’s performance was measured objectively, in tons, as a function of the manipulated variable, measured in pounds.
3. Taylor distinguished three sources of variation in performance:
a. Systematic variation, as a function of the work load,
b. Individual differences, between first-class and other workers,
c. Unexplained variation, in the weight of the shovel load and in the
variation in daily performance.
4. Taylor’s approach was experimental, objective, quantitative, and replicated over several weeks. Its object was to describe an objective, quantitative relationship between a manipulated environmental variable and a responding behavioral variable.
5. It was parsimonious and skeptical, in that it did not attempt to “explain” the observed relationship in terms of unobserved hypothetical constructs.
The research that best meets these criteria comprises field experiments and realistic experimental simulations, as opposed to surveys and other correlational studies. Even here, however, analyzing papers from a rigorously scientific perspective may suggest alternative formulations of published research. For example, the following reinterpretation of a paper takes a skeptical and objective view of a study that was originally presented in more theoretical and subjective terms. On the basis of equity theory, Greenberg (1988) hypothesized that employees moved to higher or lower status offices would change their performance to maintain a constant ratio of outcomes to inputs. In his experiment, underwriters in an insurance company were temporarily reassigned to the offices of higher-, lower-, or equal-status coworkers, while their own offices were refurbished. The highest status office was occupied by one person, the lowest by six persons with less space per person, and the middle status office by two persons with an intermediate amount of space per person. Employees were randomly reassigned to offices that were one or two steps higher, one or two steps lower, or of status equal to their own.
The participants completed five questionnaires: one asked whether the temporary office was usually assigned to a coworker of higher, lower, or equal status; a second asked whether the facilities in the temporary office enabled the participant to perform better, equally well, or more poorly at the job. The general satisfaction scale of the Minnesota Satisfaction Questionnaire asked how satisfied the respondent was with 20 dimensions of the job, an environmental satisfaction questionnaire asked about six aspects of the office, and a separate item asked how high or low was the overall level of rewards from the job.
Five different changes in office assignment produced five significantly different levels of performance. Satisfaction was affected less than performance: there were four significantly different levels of environmental satisfaction, four of overall rewards, and only two of general job satisfaction.
Greenberg considered the manipulated variable to be "workplace status." The changes in status were characterized as conditions of "overpayment" and "underpayment." Employees who were moved to higher-status offices presumably felt overpaid, so they increased their output to restore their previous outcome/input ratio. Conversely, underpaid employees decreased their performance. Environmental satisfaction and overall reward were interpreted as intervening variables, i.e., as mental states that caused changes in performance. Job satisfaction was judged not to be an intervening variable, because it was so little affected by the intervention. A skeptic might note that environmental satisfaction and overall rewards are also suspect as intervening variables, because they too were less affected by the intervention than performance was.
In support of his assertion that status, rather than the physical properties of the office, was the manipulated variable, Greenberg cited the participants’ responses to the status and the ability-to-perform questionnaires. Almost all participants recognized the status level of their temporary office, and most indicated that their temporary office enabled them to perform their job as well as their regular office. “These data,” concluded the author, “discount the possibility that performance increases or decreases noted while in the temporary offices were the result of opportunities provided by or thwarted by office conditions.” But could the participants simply have failed to report the effect of the office conditions on their performance?
A similar experiment provides additional information about the effects of the work environment on performance, and about the relationship between objective performance and self-rated performance. Larsen, Adams, Deal, Kweon, and Tyler (1998) investigated the effects of office plants on productivity, attitudes, and perceptions. They measured performance on a letter-identification task in an office with no plants, 10 plants, or 20 plants. Different participants were observed in each condition. The participants also completed questionnaires that asked them to rate office attractiveness, comfort, pleasantness of the work, subjective mood, and the level of their performance. The authors’ theory was simpler than Greenberg’s (1988). It did not invoke the mechanisms of equity theory. It merely proposed that plants would increase workers’ satisfaction, which would in turn improve their performance. However, contrary to the authors expectations, the several questionnaire measures were affected differently or not at all, and performance decreased rather than increased. Three of the six measures were significantly influenced by plant density, but in different ways. As the number of plants increased, (1) task performance decreased, while (2) office attractiveness increased. (3) Pleasantness was highest with ten plants, and (4) self-reported performance was not affected by the number of plants, nor was (5) comfort or (6) mood. Relevant to Greenberg’s (1988) experiment, the participants reported the same level of performance in all condition, even though their performance decreased significantly as the number of plants increased. Larsen et al. (1998) entertained the possibility that perhaps satisfaction decreases performance of boring tasks, but concluded that “more research is needed to confirm this explanation before discounting the simpler theory that productivity may have decreased in the presence of more plants because of visual distraction.”
The conflicting relationships between satisfaction and performance in the preceding studies can be resolved by regarding the different satisfaction scores as responses to different questions, rather than as measures of a mental state that intervenes between the environment and behavior. In this more skeptical, parsimonious, and objective view, satisfaction scores and performance scores are measures of different responses to different stimuli, and there is no necessary relationship between them. This interpretation is supported by the work of Cherrington, Reitz, and Scott (1971), who demonstrated that they could create a positive correlation between satisfaction and performance by paying a bonus to high performers and not to low performers. Similarly, they created a negative correlation between satisfaction and performance by selectively rewarding poor performers. Random rewards produced no correlation between satisfaction and performance. They quoted Skinner’s (1969) opinion that feelings are, at best, accompaniments rather than causes of behavior and that both are the products of common environmental variables. They also cited Bandura’s (1969) view that self-reports of satisfaction are simply another class of behavior, rather than indexes of an underlying state endowed with special causal powers.
Much of the research and practice in I/O psychology assumes a cognitive model in which subjective events considered to be as controllable and measurable as are environmental and behavioral variables. This model is illustrated in Figure 1, below. The preceding discussion points out the limitations of this model, which are even more serious for surveys and correlational studies than for experiments. In contrast, Figure 2 illustrates an alternative model, which is more objective, more powerful, and a better framework for the integration of science and practice.
In sum, the objectives of the proposed roundtable are threefold:
1. To develop criteria for the identification and conduct of experimental investigations that best exemplify the integration of science and practice.
2. To develop theoretical framework to organize and summarize the relationships found in those investigations.
3. To initiate a collaboration aimed reviewing the relevant literature and conducting research that is both rigorously scientific and eminently practical.
Questions
Satisfaction
Environment Attitudes, Performance
Etc.
Task
Figure 1. A commonly accepted model that explains relationships between environmental and behavioral variables by positing intervening mental processes, which are measured by responses to questionnaires.
Questions Answers to
Questions
Environment (Subjective
processes)
Task Performance
Figure 2. A model that describes objective, quantitative relationships between variables that can be manipulated and responses that can measured. These relationships are the principles that constitute the science and practice of an applied organizational psychology. Subjective events are outside the scope of this discipline.
The host, Dr. Ira T. Kaplan, has worked for over forty years as an applied experimental psychologist in business, industry, and the military. He received his Ph.D. in experimental psychology from Columbia University in 1959. Since 1957, he has done basic research in visual perception, memory, and problem solving, and applied research in human factors, training, performance assessment, and customer satisfaction. He has worked for the System Development Corporation, Dunlap and Associates, and the U. S. Army, and has consulted for IBM, Xerox, and AT&T, among others. He taught in the graduate program in Experimental Cognition at the City University of New York, and is now a Professor in the Psychology Department at Hofstra University, where he has taught graduate and undergraduate courses in industrial and organizational psychology since 1985.
References
Bandura, A. (1969). Self-efficacy mechanism in human agency. American Psychologist, 37, 122 – 147.
Cherrington, D. J., Reitz, H. J., & Scott, W. E., Jr. (1971). Effects of contingent and non-contingent reward on relationship between satisfaction and task performance. Journal of Applied Psycholog, 55, 531 – 536.
Greenberg, J. (1988). Equity and workplace status: A filed experiment. Journal of Applied Psychology, 73, 606 – 613.
Larsen, L., Adams, J., Deal, B., Kweon, B-S., & Tyler, E. (1998). Plants in the workplace: The effects of plant density on productivity, attitudes, and perceptions. Environment and Behavior, 30, 261 – 281.
Skinner, B. F. (1969). Contingencies of reinforcement: A theoretical analysis. New York: Appleton-Century Crofts.
Taylor, F W. (1911). Scientific management, New York: Harper.