Austin and Carr Chapter 13 Research PDF

# Chapter 12 ## Human Services - Morganton, NC: Developmental Disabilities Services Managers, Inc. - Reid, D. H., & Parsons, M. B. (1995b). Motivatinghuman service staff: Supervisory strategies for maximizing work effort and work enjoyment. Morganton, NC: Habilitative Man- agement Consultants, Inc....

# Chapter 12 ## Human Services - Morganton, NC: Developmental Disabilities Services Managers, Inc. - Reid, D. H., & Parsons, M. B. (1995b). Motivatinghuman service staff: Supervisory strategies for maximizing work effort and work enjoyment. Morganton, NC: Habilitative Man- agement Consultants, Inc. - Reid, D. H., & Parsons, M. B. (1995c). Comparing choice and questionnaire measures of the acceptability of a staff training procedure. Journal of Applied Behavior Analysis, 28, 95-96. - Reid, D. H., & Parsons, M. B. (1996). A comparison of staff acceptability of immediate versus delayed verbal feedback in staff training. Journal of Organizational Behavior Management, 16, 35-47. - Reid, D. H., Parsons, M. B., & Green C. W. (1989). Staff management in human services: Behavioral research and application. Springfield, IL: Charles C Thomas. - Schell, R. M. (1998). Organizational Behavior Management: Applications with profes- sional staff. Journal of Organizational Behavior Management, 18, 157-171. - Shoemaker, J., & Reid, D. H. (1980). Decreasing chronic absenteeism among institu- tional staff: Effects of a low-cost attendance program. Journal of Organizational Behavior Management, 2, 317-328. - Singer, G., Sowers, J., & Irvin, L. K. (1986). Computer-assisted video instruction for training paraprofessionals in rural special education. Journal of Special Education Technology, 8, 27-34. - Sneed, T. J., & Bible, G. H. (1979). An administrative procedure for improving staff performance in an institutional setting for retarded persons. MentalRetardation, 17, 92-94. - Stoddard, L. T., McIlvane, W. J., McDonagh, E. C., & Kledaras, J. B. (1986). The use of picture programs in teaching direct care staff. Applied Research in Mental Retardation, 7, 349-358. - Stumphauzer, J. S., & Davis, L. C. (1983). Training Mexican American mental health personnel in behavior therapy. Journal of Behavior Therapy & Experimental Psychia- try, 14, 215-217. - Templeman, T. P., Fredericks, H. D. B., Bunse, C., & Moses, C. (1983). Teaching research in-service training model. Education and Training of the Mentally Retarded, 28, 245-252. - Wallace, C. J., Davis, J. R., Liberman, R. P., & Baker, V. (1973). Modeling and staff behavior. Journal of Consulting and Clinical Psychology, 41, 422-425. - Whitman, T. L., Scibak, J. W., & Reid, D. H. (1983). Behavior modification with the severely and profoundly retarded: Research and application. New York: Academic Press. - Wilson, P. G., Reid, D. H., & Korabek-Pinkowski, C. A. (1991). Analysis of public verbal feedback as a staff management procedure. Behavioral Residential Treatment, 6,263- 277. - Wolf, M. M., Kirigin, K. A., Fixsen, D. L., Blase, K. A., & Braukmann, C. J. (1995). The teaching-family model: A case study in data-based program development and refinement (and dragon wrestling). Journal of Organizational Behavior Management, 15, 11-68. # Chapter 13 ## Basic Behavioral Research and Organizational Behavior Management - Nearly 50 years ago, Skinner (1953) predicted that behavioral principles derived from basic laboratory research would form the foundation of a useful, problem- solving science of human behavior. In the inaugural issue of the Journal of Applied Behavior Analysis (JABA), Baer, Wolf, and Risley (1968) announced the arrival of such a science, christened “applied behavior analysis." In the 30 years since that article appeared, applied behavior analysis has flourished. One of several areas evidencing remarkable growth is called organizational behavior management. - Organizational behavior management (OBM) (Frederiksen & Lovett, 1980) and performance management (PM) (Daniels, 1989) are terms used interchangeably to refer to the use of behavior analysis in business, industry, and government. OBM has proven useful for dealing with a wide range of behavioral problems in both the public (e.g., Brand, Staelin, O'Brien, & Dickinson, 1982) and the private (e.g., Merwin, Thomason, & Sanford, 1989) sectors. For example, OBM has improved the services provided by doctors and other staff at a medical facility (Gikalov, Baer, & Hannah, 1997); reduced machine set-up time in a rubber manufacturing company (Wittkopp, Rowan, & Poling, 1990); reduced application processing time in a university admissions department (Wilk & Redmon, 1990); facilitated the use of statistical process control in a machine shop (Henry & Redmon, 1990); increased innovation in a public utility company (Smith, Kaminski, & Wylie, 1990); enhanced the effectiveness of performance management training in city government (Nordstrum, Lorenzi, & Hall, 1991); increased functional sales behaviors in salespersons (Luthans, Paul, & Taylor, 1985); increased on-time project completions by engineers (McCuddy & Griggs, 1984); increased the safety and productivity of roofers (Austin, Kessler, Riccobono, & Bailey, 1996), and; increased the friendliness of police officers (Wilson, Boni, & Hogg, 1997). - Several different procedures were used in these examples, and in other OBM applications. For instance, behavior has been improved by altering its antecedents and consequences through a training procedure (Brown, Malott, Dillon, & Keeps, 1980; Komaki, Barwick, & Scott, 1978); by providing prompts and task-related information (Austin, Hatfield, Grindle, & Bailey, 1993; Carter, Hansson, Holmberg, & Melin, 1979; Engerman, Austin, & Bailey, 1997; Greene & Neistat, 1983; Runnion, Watson, & McWhorter, 1978); by setting goals (Eldridge, Lemasters, & Szypot, 1978; Komaki, Barwick, & Scott, 1978); by providing written or verbal performance feedback (Henry & Redmon, 1990; Silva, Duncan, & Doudna, 1981; Wittkopp et al., 1990) or praise (Gaetani & Johnson, 1983; Komaki, Blood, & Holder, 1980); by contingent reinforcement (Foxx & Schaeffer, 1981; Newby & Robinson, 1983; Smith et al., 1990), and; by complex "package interventions" (Wilk & Redmon, 1990). - Performance feedback, which is information that is given to persons regarding the quantity or quality of their past performance (Prue & Fairbank, 1981), appears to be the OBM procedure used most often in published articles. In a review of all articles published in the Journal of Organizational Behavior Management (JOBM) from 1977-1986, Balcazar et al. (1989) reported that 50% of the studies used feedback, either alone or in combination with other procedures. Prue and Fairbank (1981) suggested that feedback is often the intervention of choice because of its low cost, simplicity, and flexibility. In addition, feedback has been demonstrated to be effective in a variety of applications (e.g., Johnson & Masotti, 1990; Ralis & O'Brien, 1986; Sulzer-Azaroff, Loafman, Merante, & Hlavacek, 1990). - Given how often and how effectively it has been used in OBM, it is rather ironic that the manner in which performance feedback affects behavior often is unknown. In 1982, Peterson contended that "feedback" is not a principle of behavior, but is instead an ambiguous term used to describe a wide range of manipulations (independent variables). Depending on how feedback is arranged, it might serve a reinforcing, punishing, or discriminative stimulus function. It might also act as an establishing operation, or engender new rule-governed behavior. Therefore, Peterson suggested that, In summary, procedures labeled feedback can be explained by operant principles and need a behavioral analysis to determine why they are effective. Such an analysis would also help design "feedback" to be maximally effective. Presumably various aspects of information about performance could be manipulated to have several behavioral effects, leading to more powerful changes in behavior. Indeed, rules could be developed telling behavioral technicians how to establish "feedback" as an effective conditioned reinforcer, or a discriminative stimulus, or an estab- lishing stimulus. Much ambiguity would be eliminated if behavior analysts no longer used the term "feedback.” It is not a new principle of behavior and does not refer to a specific procedure; it at best has simply become professional slang (p. 102). - Peterson's critical analysis of the concept of feedback did not lessen organiza- tional behavior analysts' enthusiasm for the term, or foster careful empirical analyses of how feedback affects behavior. His criticism was, however, consistent with the spirit of the times, insofar as several of Peterson's colleagues noted that, as applied behavior analysis grew, it had moved progressively away from its basic research roots in the experimental analysis of behavior, becoming ever more technological and imprecise (e.g., Branch & Malagodi, 1980; Deitz, 1978; Epling & Pierce, 1983; Hayes, Rincover, & Solnick, 1980; Michael, 1980; Pierce & Epling, 1980; Poling, Picker, Grossett, Hall-Johnson, & Holbrook, 1981). Most authors were in some regard critical of the alleged schism between applied behavior analysis and the experimental analysis of behavior, but some were not (e.g., Baer, 1981). A common criticism was that basic research findings were being ignored by applied researchers and practitioners. ## Basic and Applied Research - In general, "basic" and "applied” research are best viewed as endpoints on a continuum that scales the practical benefits of experimentation, not as clear and discrete categories. As we use the terms, basic behavioral research is designed to provide information about behavior and the variables that control it. A basic research study does not directly attempt to solve a problem, and need not benefit participants. Historically, research in this vein has been categorized as the experi- mental analysis of behavior (Skinner, 1953). Applied research, in contrast, is intended to remedy a behavioral problem, and usually benefits participants directly. Such research typically would be categorized as applied behavior analysis (Baer et al., 1968). Basic research can, of course, have significant implications for application, even though it was not designed to meet this end. For instance, as we discuss later, several scientists and practitioners have suggested that the matching equation, which is based on the experimental analysis of behavior, can be put to use in predicting and controlling problem behaviors in the workplace and elsewhere. Others disagree. - In some cases, basic research is intended specifically to provide information that suggests strategies for improving human behavior. An example that we consider later involves the laboratory studies of incentive pay conducted by Alyce Dickinson and her colleagues. The primary justification for conducting such studies is to garner information relevant to the selection of pay systems in the everyday world. As we shall see, however, making the leap from laboratory simulations to practical implications is fraught with difficulty. ## Quantifying the Relation between Basic and Applied Areas - One problem in considering the relation between basic and applied behavior analysis is in quantifying the extent to which the fields interact. In an early attempt to generate relevant data, Poling et al. (1981) looked at cross-citation rates in the Journal of the Experimental Analysis of Behavior (JEAB) and the Journal of Applied Behavior Analysis (JABA). They also surveyed members of the editorial boards of JEAB and JABA. The former journal is probably the most prestigious outlet for basic behavior analytic research, the latter for applied behavioral research. Poling and colleagues found that, from 1968 through 1979, articles published in JEAB were cited increasingly infrequently in JABA. Moreover, a survey of editorial board members revealed that few read both journals, published in both journals, or found both journals useful in their research. Most editors of JABA reported that JEAB research had decreased in value to applied researchers across time, and the majority of editors of both journals felt that the basic and applied research areas were growing apart. They did not, however, agree as to whether this emerging schism was harmful. - In a follow-up to the Poling et al. study, Poling, Alling, and Fuqua (1994) determined cross-citation rates for JABA and JEAB during 1983-1992. Across years, 0.3 to 4.8% of JABA citations were JEAB articles, with a mean of 2.4%. For JAВА citations in JEAB, the mean was 0.6% and the range across years was 0.1 to 1.7%. For both journals, there were no obvious trends in the percentage of cross-citations across time. The JEAB article most often cited in JABA was Herrnstein (1970), which was cited 12 times. That article deals with quantitative analysis of behavior and the matching equation (or law). - When the data for JEAB citations in JABA from 1983-1992 (Poling et al., 1994) are compared to findings from 1968 through 1979 (Poling et al., 1981), it is apparent that the downward trend evident in the first survey did not continue. In fact, the percentage of total references in JABA articles that came from JEAB was higher for each year from 1983 through 1992 than for any year during the late 1970s. Perhaps the oft-repeated suggestion that applied behavior analysts should attend more carefully to basic research findings (e.g., Epling & Pierce, 1983; Pierce & Epling, 1980) has exercised some behavioral control. Or it may be that basic researchers are attending increasingly to topics that potentially are of applied significance, a tack some authors have suggested to be gainful (e.g., Epling & Pierce, 1986; Poling et al., 1981). If they are, however, they are not commonly citing JABA articles for verification: From 1983 through 1992, very few references in JEAB were from JABA (Poling et al., 1994). - Although OBM articles are published in JABA, many of the articles come from other applied areas. Studies of JEAB citations in JABA have not reported data separately for OBM and other areas, thus, the extent to which basic research is cited in OBM articles is unknown. ## Basic-research Citations in JOBM Articles: 1992-1997 - As an initial step towards assessing the influence of basic behavioral research on OBM, we determined how frequently JEAB and other basic research sources were cited in articles published from 1992-1997 in JOBM, which is devoted entirely to OBM. To do so, a single observer rated each reference in every article as to whether the reference came from JEAB or another basic-research outlet (e.g., Journal of Experimental Psychology, Learning and Motivation). - Figure 1 shows for each year and for all years together the percentage of JOBM citations that came from basic-research sources. There were no trends across years, although considerable variability was evident. In total, 98 of 1269 references (7.7%) were rated as basic-research citations; 87 of those references (6.8%) came from JEAB. This value is substantially higher than that reported by Poling et al. (1994) for JEAB citations in JABA articles. Although analysis is complicated by the different time bases involved in the two surveys, comparing the present data to those reported by Poling et al. (1994) suggests that people active in OBM may be more inclined to cite basic-research sources than are those active in other areas of applied behavior analysis. Figure 2 shows for each year and for all years together the percentage of JOBM articles that cited at least one basic-research reference. Here again, data were variable across years, but no trends were evident. Across the rated years, 40 of 72 JOBM pub- lications (55%) contained one or more basic-research reference. We made no attempt to ascer- tain how those references were used, however, and the extent to which they contributed to de- veloping and conceptualizing behavior-change procedures is unknown In all, 72 different JEAB articles were cited in JOBM publications from 1992- 1997. One article (Baum, 1973) was cited four times, one article was cited three times (Rao & Mawhinney, 1991), and 10 articles were cited twice. The most-cited article (Baum, 1973) is concerned with the quantitative analysis of behavior, specifically, the matching equation (law). As noted previously, matching was also the topic of the JEAB article most-often cited in JABA publications from 1983-1992, although the specific article was written by Herrnstein (1970). The third most-cited JEAB article also was written by Herrnstein (1961) and dealt with match- ing. Given that basic research in the quantitative analysis of behavior appears to have had a considerable influence of ap- plied behavior analysis in gen- eral, and OBM in particular, we address the topic at some length later in this chapter. ## Basic Research Areas with Potential Applied Implications - Interest in real and ideal relations between basic and ap- plied behavior analysis has not abated over time (e.g., Baer, 1987; Davey, 1988; Epling & Pierce, 1986; Hayes, 1991; Perone, Galizio, & Baron, 1988). It fact, interest appears to have grown over the past few years, during which JABA has published special articles dealing with developments in basic research and their potential applications. Among the topics that have been considered are schedules of reinforcement (Lattal & Neef, 1996), stimulus class formation (Stromer, Mackay, & Remington, 1996), response effort (Friman & Poling, 1995), convergence (Hineline & Wacker, 1993), relational responding (Hayes & Hayes, 1993), language as behavior (Cataldo & Brady, 1994); matching (Pierce & Epling, 1995), behavioral pharmacology (Kirby & Bickel, 1995), discriminated operants (Nevin & Mace, 1994), collateral effects of behavioral interventions (Shull & Fuqua, 1993), and the nature of reinforcement (Iwata & Michael, 1994). In addition, abstracts from JEAB articles have appeared recently in JABA. - It is beyond our scope to examine the possible implications for OBM of the various areas of basic research discussed in JABA. Readers interested in the practical implications of basic research findings are well advised to read the articles cited in the preceding paragraph. Although coverage rarely deals specifically with OBM applications, the implications of basic research for OBM are not fundamentally different than for other areas of applied behavior analysis, and we recommend the articles highly. Most were written by two people, an expert in applied behavior analysis and a specialist in the experimental analysis of behavior, and they charac- teristically offer impressive arguments for the practical implications of basic research findings. - Unfortunately, it is difficult to evaluate the veracity of these arguments. We have argued elsewhere (e.g., Dickinson & Poling, 1996; Poling & Foster, 1993; Poling, Schlinger, Starin, & Blakely, 1990), and remain convinced, that humans are similar to other animals in their sensitivity to environmental arrangements. But we have also argued that it is unwise to extrapolate findings from basic research settings to very different applied settings. For example, Latham and Huber (1992) reviewed a series of eight articles that examined the effects of scheduling incentive pay in various ways. They used the results of those studies to support the contention that "the operant principles developed through basic research can be utilized in applied settings to design reinforcement schedules that evoke desired behavior from individual em- ployees or employee groups" (Latham & Huber, 1992, p. 146). - Dickinson and Poling (1996) contended that, in fact, the studies reviewed by Latham and Huber (1992) did not involve manipulations of like-named reinforce- ment schedules from the basic laboratory, and that their results cannot be compared meaningfully to basic-research findings. For example, in the eight studies of concern, conditioned reinforcers were employed, complex response sequences (e.g., planting 1,000 trees) were arranged as operants, the possibility existed that instructions (or rules) affected sensitivity to programmed contingencies, and delays to reinforcement were substantial (hours or days). Moreover, an hourly pay schedule also was in effect in six of the eight studies. These conditions are very different than those character- istically arranged in basic research on schedules. - Nonetheless, Latham and Huber (1992) discussed these conditions as if they involved simple (e.g., fixed-ratio, variable-ratio) schedules of reinforcement, and attempted to relate workers' performance under these schedules to patterns charac- teristically observed in basic research involving like-named simple schedules. Dickinson and Poling (1996) acknowledged that the manipulations examined by Latham and Huber (1992) were in many cases effective in improving productivity. But, they claimed, these results cannot validly be explained in terms of alterations in simple reinforcement schedules. Although such an analysis appears to tie important applied findings to well established behavioral phenomena, the tie is a weak one, perhaps best viewed as a metaphorical extension, and with caution. Of course, some will disagree with Dickinson and Poling (1996), and find Latham and Huber's analysis elegantly simple and utterly compelling. There is no gold standard for evaluating the validity, or practical value, of attempts to integrate basic and applied research findings. Mace (1994) has, however, made the case that the applied relevance of basic research is likely to be enhanced when a three-step approach is followed: 1. Animal models of important human problems are developed. 2. Findings generated by these models are replicated with humans in controlled settings. 3. Findings based on animal models and controlled human experiments are used to generate interventions that are tested with humans exhib- iting the behavior problem of concern in the natural environment. - As Pierce and Epling (1995) point out, there are very few examples of the actual application of this model. One is their work concerning activity anorexia (Pierce & Epling, 1994). A second, they claim, is research involving the matching law. As indicated previously, the basic-research article most often cited in JOBM in recent years deals with this topic. Therefore, it merits consideration. ## Matching and OBM¹ - In brief, the term “matching” refers to how organisms distribute their behavior (or time) between two (or more) concurrent schedules of reinforcement. Concurrent schedules operate simultaneously and independently for two or more response classes (Catania, 1991). If those schedules are variable-interval (VI) schedules arranging equal amounts of food delivery, the proportion of responses allocated to each alternative often is equal to (i.e., “matches") the proportion of reinforcers obtained under that alternative. - Matching was first demonstrated by Herrnstein (1961), who exposed pigeons to several different concurrent VI schedules of food delivery. Each VI schedule specified that food became available aperiodically and was delivered dependent on a keypeck, with the average time between successive food availabilities equal to the schedule value. Herrnstein recorded the number of pecks emitted and the number 1. A substantial portion of our discussion of matching is based on Poling and Foster (1993). of food deliveries obtained under each alternative. He found that the relative proportion of responses emitted under a schedule matched the relative proportion of reinforcers (food deliveries) under that alternative, a relation that is expressed in equation 1, which is the first version of the matching equation, or "matching law." B₁/(B₁ + B₂) = R₁/(R₁ + R2 (Equation 1) where B₁ is behavior (i.e., total responses) allocated to alternative one, B₂ is behavior allocated to alternative two, R, is the number of reinforcers received under alternative one, and R₂ is the number of reinforcers received under alternative two. - Because Equation 1 did not always provide a good description of performance under concurrent schedules, other versions of the matching equation were devel- oped. One, the generalized matching equation (Baum, 1974), is prominent in the basic research literature. It appears as Equation 2: log (B₁/B₂) = a log (r₁/r₂) + log c (Equation 2) where B is the measure of behavior (responses or time) under a component schedule, r is the obtained reinforcement rate under a component schedule, and the two alternative component schedules are designated by the subscripts. The slope of the line fitted to the data, a, is a measure of the degree to which behavior is influenced by differences in reinforcement rates (i.e., the sensitivity of behavior to reinforce- ment). Log c, the intercept, is an estimate of bias in behavior for one or the other schedule. - A third well-known version of the matching equation was developed by Herrnstein to describe behavior under what appear to be single-alternative situa- tions, such as multiple schedules (Herrnstein, 1970). This equation is based on the premise that an organism always is faced with alternative responses, each reinforced under some schedule. It takes the form: B₁ = k(R)/(R + R) (Equation 3) where B₁ is behavior allocated to alternative one (the scheduled, or known, operant), R₁ is reinforcement received under alternative one, k is the asymptotic rate of B₁, and R is the sum of all reinforcement other than that produced under alternative one. - If it is assumed that R and k are constant across conditions, Equation 3 predicts hyperbolic relation between rate of responding and rate of reinforcement under a given simple (or multiple) schedule (Herrnstein, 1970; McDowell, 1988). - Many studies demonstrate that the generalized matching equation provides a good description of relations between behavioral outputs and environmental inputs under a range of conditions (e.g., Baum, 1979; Davison & McCarthy, 1988; Mazur, 1991; Williams, 1988), although there are limits (e.g., Elsmore & McBride, 1994; Sevastano & Fantino, 1994). Put simply, the relations codified by the matching equations are real behavioral phenomena. - They are not, however, phenomena revealed through research designed to develop an animal model of a human behavior problem. In fact, no mention is made of behavior problems in the early literature on matching. Thus, research in matching does not meet Mace's (1994) first criterion for a line of basic research most likely to have applied implications. That this is so, however, does not mean that the research has no implications. Certainly the second two steps in Mace's model can be taken absent the first. - With matching, the second step - which can be construed as showing that data produced with nonhumans in controlled settings holds with humans in controlled settings - has indeed been taken. Several studies have shown that the matching equation adequately describes humans' performance under conditions similar to those where it so describes nonhumans' performance, as indicated in reviews by Pierce and Epling (1983, 1995) and McDowell (1988). There is no doubt that matching can be demonstrated as readily in humans as in nonhumans tested under comparable, controlled conditions. Thus, the second step in Mace's model is sound with regard to matching analyses. - Although many authors have addressed the topic (e.g., Fuqua, 1984; Mace, McCurdy, & Quigley, 1990; McDowell, 1981, 1982, 1988; Myerson & Hale, 1984; Pierce & Epling, 1983, 1995; Rachlin, 1989), there is no agreement as to whether matching can be profitably used to develop useful interventions, or to explain behavior problems in everyday situations.. Over two decades ago, Mawhinney suggested that the matching law is relevant to organizational behavior management (Mawhinney, 1975; Mawhinney & Ford, 1977), and further arguments in favor of this position have appeared more recently (e.g., Mawhinney & Gowan, 1991; Redmon & Lockwood, 1987). - The natural sciences are fundamentally quantitative, and it is proper that anyone who favors a natural-science approach to the study of behavior finds matching analyses fascinating. But that is not to say that all who favor such an approach believe that matching analyses are of value in OBM. Among those who have argued that matching is largely irrelevant to OBM are Poling and Foster (1992). The details of their argument will not be repeated here, but their primary points are: 1. There are no demonstrations that the matching equation adequately describes organizational behavior, and there are no effective proce- dures based on the matching equation. 2. Behavior in organizations characteristically is controlled by multiple environmental variables that are difficult to isolate and quantify environments in which people live and work cannot accurately be described as concurrent schedules. Unless environmental inputs (en- vironmental variables) can be quantified and related to behavioral outputs, matching analyses are impossible. Matching analyses are of value first and foremost because they are quantitative; qualitative extensions (e.g., Redmon & Lockwood, 1987) are of little or no value in predicting or controlling behavior. 3. Organizational settings differ substantially from the basic-research settings in which the matching equation adequately describe behavior. In particular, behavior in the workplace often appears to be, in part, rule-governed. No one has extended the application of matching analyses to rule-governed behavior. 4. Even if matching analyses could be applied to organizational behav- ior, no benefit would derive from the practice. In OBM, describing precisely the relation between environmental inputs and behavioral outputs is far less important than altering environmental inputs so that behavior changes in desired ways. OBM researchers and practitioners have developed a sizable number of rough-and-ready, but generally effective, strategies for behavior change. Analyzing these interventions in terms of matching is much like slicing bologna with a microtome. - In fairness, the points outlined above are open to debate, and time may well prove Poling and Foster (1992) wrong. Even if it does not, applied researchers certainly suffer no harm from learning about matching and trying to extend what they learn to everyday human activities. There is, as discussed later, general value in learning as much as possible about behavioral relations, of almost any sort. As Mace (1994) suggested, however, the practical significance of basic research findings should be easiest to see in those situations where the basic research was designed to explicate a human behavior problem. One such line of basic research is discussed in the following section. ## Dickinson and Associates' Studies of Incentive Pay - As indicated earlier, Dickinson and her colleagues have undertaken basic research to examine the effects of individual monetary incentives on worker productivity. Although this research is indeed basic, it was generated by questions that arose from the application of monetary incentive systems in business. In 1974, with guidance from William Abernathy, a Memphis-based behavioral consultant, Union National Bank in Little Rock, AR, began implementing behaviorally-based monetary incentive systems. By the early 1980s, 75 individualized monetary incentive programs had been installed, covering 70% of the bank's 485 employees (Dierks & McNally, 1987). During that time, the bank hired Kathleen McNally, a behaviorally-trained psychologist, to oversee and continue that work. In 1987, Dierks, the Senior Vice President and Personnel Director, and McNally described the success of their pay systems: In 1985, $1 million was paid in incentive payments on a $9 million annual payroll. But it's more than worth it. Using these principles, we have increased productivity 200-300 percent. Our net profit per employee is $11,000 per year while other Little Rock banks show $5,700 and $4,200 (p. 61). - In spite of this success, McNally wanted to refine the pay systems. While they were certainly working well, several features lacked empirical justification. Dickinson and her colleagues began investigating a number of questions posed by McNally. Two of those have been the focus of much of this work: (1) Is there a functional relationship between the percentage of total pay that is incentive-based and worker productivity; and (2) Does the absolute amount of money earned in incentives influence worker productivity? The broader utility of this research, along with its rationale, have been discussed in detail elsewhere (Dickinson & Gillette, 1993; Frisch & Dickinson, 1990; LaMere, Dickinson, Henry, Henry, & Poling, 1996; Oah & Dickinson, 1992) and will not be considered here. Instead, the present focus will be on the relevance of such basic research to application, and on the difficulties of balancing controlled experimentation with the production of useful data. - Frisch and Dickinson (1990) conducted the first study in the series, examining the effects of five different percentages of incentive pay on worker performance. In a between-groups study, 75 college students were randomly assigned to one of five pay conditions. Subjects in the 0% incentive pay group received a guaranteed base pay but were not given the opportunity to earn incentives. Subjects in the remaining four groups also received a guaranteed base wage but in addition were able to earn 9, 23, 38, or 50% of their total pay in incentives. Each subject