johns_ob_6e_ebook_chapp

Johns_ob_6e_ebook_ch - 554 Indiviaual Behaviour Part Two Appendix Research in Organizational Behaviour Learning Objectives After reading the

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: 554 Indiviaual Behaviour Part Two Appendix Research in Organizational Behaviour Learning Objectives After reading the Appendix, you should be able to: 1 Explain what a hypothesis is and define the meaning of a variable. 2 Differentiate reliability from validity and con- vergent validity from discriminant validity. 3 Understand observational research and dis- tinguish between participant and direct observation. 4 Describe correlational research and explain why causation cannot be inferred from correlation. 5 Explain experimental research and distinguish between independent and dependent variables and explain the meaning of internal validity. 6 Discuss the relative advantages and disadvan- tages of various research techniques. 7 Describe random sampling and external validity and the role they play in the research process. 8 Explain the Hawthorne effect and how it can occur. 9 State the basic ethical concerns to which researchers must attend. Research is a way of finding out about the world through objective and systematic information gathering. The key words here are objective and systematic, and it is these characteristics that separate the outcomes of the careful study of organizational behaviour from opinion and common sense. Understanding how researchers conduct their research is important to the study of organizational behaviour for several reasons. First of all, you should be aware of how the information presented in this book was collected. This should increase your confidence in the advantages of systematic study over common sense. Second, you will likely encounter reports, in management periodicals and the popular press, of interventions to improve organizational behaviour, such as job redesign or employee development programs. A critical perspective is necessary to differentiate those interventions that are carefully designed and evaluated from useless or even damaging ones. Those backed by good research deserve the greatest confidence. Occasionally, a manager may have to evaluate a research proposal or consultant’s intervention to be carried out in his or her own organization. A brief introduction to research methodology should enable you to ask some intelligent questions about such plans. Trained behavioural scientists who have backgrounds in management, applied psychology, or applied sociology carry out research in organizational behaviour. While this introduction will not make you a trained behavioural scientist, it should provide an appreciation of the work that goes into generating accurate knowledge about organizational behaviour. Chapter 3 Appendix Perception, Attribution, and Judgment of Others 555 The Basics of Organizational Behaviour Research All research in organizational behaviour begins with a question about work or organizations. Sometimes, this question might stem from a formal theory in the field. For example, a motivation theory called equity theory (see Chapter 5) is concerned with peoples’ reactions to fairness or lack of it. Equity theory suggests the following research question: What do people do when they perceive their pay to be too low in comparison with other people? Other times, a research question might stem from an immediate organizational problem. For example, a human resource manager might ask herself: How can we reduce absenteeism among our customer service personnel? Often, research questions are expressed as hypotheses. A hypothesis is a formal statement of the expected relationship between two variables. Variables are simply measures that can take on two or more values. Temperature is a variable, but so are pay, fairness, and absenteeism. A formal hypothesis stemming from equity theory might be: The less fair people perceive their pay to be, the more likely they will be to resign their jobs. Here, a variable that can take on many values, perceived fairness, is linked to a variable made up of two values, staying or leaving. The human resource manager might develop this hypothesis: The introduction of a small attendance bonus will reduce absenteeism. Here, a variable with two values, bonus versus no bonus, is related to one that can take on many values, days of absenteeism. Good researchers carefully measure the variables they choose. For one thing, a measure should exhibit high reliability. Reliability is an index of the consistency of a research subject’s responses. For example, if we ask someone several questions about how fair his or her pay is, the person should respond roughly the same way to each question. Similarly, the person should respond roughly the same way to the same questions next week or next month if there has been no change in pay. Measures should also exhibit high validity. Validity is an index of the extent to which a measure truly reflects what it is supposed to measure. For instance, a good measure of perceived pay fairness should not be influenced by employees’ feelings of fairness about other workplace factors such as supervision. Also, a researcher would expect people who are objectively underpaid to report high pay unfairness and for them to report increased fairness if their pay were increased. Researchers are often able to choose measures with a known history of reliability and validity. Good measures should also be strongly related to other measures of the same variable and should not be related to measures of different variables. For example, a measure of job satisfaction should be highly correlated to other measures of job satisfaction. This is known as convergent validity and it exists when there is a strong relationship between different measures of the same variable. In addition, good measures should not be related to measures of different variables. For example, a measure of job satisfaction should not be strongly related to measures of job performance. This is known as discriminant validity and it exists when there is a weak relationship between measures of different variables. Good measures should have both convergent and discriminant validity. Thus, a measure of job satisfaction should be more strongly related to other measures of job satisfaction than to measures of job performance. There are three basic kinds of research techniques—observation, correlation, and experimentation. As you will see, each begins with a research question or ques- Hypothesis. A formal statement of the expected relationship between two variables. Variables. Measures that can take on two or more values. Reliability. An index of the consistency of a research subject’s responses. Validity. An index of the extent to which a measure truly reflects what it is supposed to measure. Convergent validity. When there is a strong relationship between different measures of the same variable. Discriminant validity. When there is a weak relationship between measures of different variables. 556 Research in Organizational Behaviour Appendix tions. Correlation and experimentation are most likely to test specific hypotheses and devote explicit attention to measurement quality. Observational Techniques Observational research. Research that examines the natural activities of people in an organizational setting by listening to what they say and watching what they do. Observational research techniques are the most straightforward ways of finding out about behaviour in organizations and thus come closest to the ways in which we develop common-sense views about such behaviour. In this case, observation means just what it implies—the researcher proceeds to examine the natural activities of people in an organizational setting by listening to what they say and watching what they do. The difference between our everyday observations and the formal observations of the trained behavioural scientist is expressed by those key words systematic and objective. First, the researcher approaches the organizational setting with extensive training concerning the nature of human behaviour and a particular set of questions that the observation is designed to answer. These factors provide a systematic framework for the business of observing. Second, the behavioural scientist attempts to keep a careful ongoing record of the events that he or she observes, either as they occur or as soon as possible afterward. Thus, excessive reliance on memory, which may lead to inaccuracies, is unnecessary. Finally, the behavioural scientist is well informed of the dangers of influencing the behaviour of those whom he or she is observing and is trained to draw reasonable conclusions from his or her observations. These factors help ensure objectivity. The outcomes of observational research are summarized in a narrative form, sometimes called a case study. This narrative specifies the nature of the organization, people, and events studied, the particular role of and techniques used by the observer, the research questions, and the events observed. Participant Observation Participant observation. Observational research in which the researcher becomes a functioning member of the organizational unit being studied. One obvious way for a researcher to find out about organizational behaviour is to actively participate in this behaviour. In participant observation the researcher becomes a functioning member of the organizational unit he or she is studying in order to conduct the research. At this point you may wonder, “Wait a minute. What about objectivity? What about influencing the behaviour of those being studied?” These are clearly legitimate questions, and they might be answered in the following way: In adopting participant observation, the researcher is making a conscious bet that the advantages of participation outweigh these problems. It is doubtless true in some cases that “there is no substitute for experience.” For example, researcher Robert Sutton wanted to find out how employees cope with jobs that require them to express negative emotions.1 To do this, he trained and then worked as a bill collector. This is obviously a more personal experience than simply interviewing bill collectors. Another advantage to participant observation is its potential for secrecy—the subjects need not know that they are being observed. This potential for secrecy does raise some ethical issues, however. Sociologist Tom Lupton served as an industrial worker in two plants in England to study the factors that influenced productivity.2 Although he could have acted in secrecy, he was required to inform management and union officials of his presence to secure records and documents, and he thus felt it unfair not to inform his workmates of his purpose. It should be stressed that his goals were academic, and he was not working for the managements of the companies involved. Sometimes, however, secrecy seems necessary to accomplish a research goal, as the following study of “illegal” industrial behaviour shows. Joseph Bensman and Israel Gerver investigated an important organizational problem: What happens when the activities that appear to be required to get a job Appendix Research in Organizational Behaviour 557 done conflict with official organizational policy?3 Examples of such conflicts include the punch press operator who must remove the safety guard from his machine to meet productivity standards, the executive who must deliver corporate money to a political slush fund, or the police officer who cannot find time to complete an eight-page report to justify having drawn her revolver on a night patrol. The behaviour of interest to Bensman and Gerver was the unauthorized use of taps by aircraft plant workers. A tap is a hard steel hand tool used to cut threads into metal. The possession of this device by aircraft assemblers was strictly forbidden because the workers could use it to correct sloppy or difficult work such as the misalignment of bolt holes in two pieces of aircraft skin or stripped lock nuts; both these problems could lead to potential structural weaknesses or maintenance problems. Possession of a tap was a strict violation of company policy, and a worker could be fired on the spot for it. On the other hand, since supervisors were under extreme pressure to maintain a high quota of completed work, the occasional use of a tap to correct a problem could save hours of disassembly and realignment time. How was this conflict resolved? The answer was provided by one of the authors, who served as a participant observer while functioning as an assembler. Put simply, the supervisors and inspectors worked together to encourage the cautious and appropriate use of taps. New workers were gradually introduced to the mysteries of tapping by experienced workers, and the supervisors provided refinement of skills and signals as to when a tap might be used. Taps were not to be used in front of inspectors or to correct chronic sloppy work. If “caught,” promiscuous tappers were expected to act truly penitent in response to a chewing out by the supervisors, even if the supervisors themselves had suggested the use of the tap. In short, a social ritual was developed to teach and control the use of the tap to facilitate getting the work out without endangering the continued presence of the crucial tool. Clearly, this is the kind of information about organizational behaviour that would be extremely difficult to obtain except by participant observation. Direct Observation In direct observation the researcher observes organizational behaviour without participation in the activity being observed. There are a number of reasons why one might choose direct observation over participant observation. First, there are many situations in which the introduction of a new person into an existing work setting would severely disrupt and change the nature of the activities in that setting. These are cases in which the “influence” criticism of participant observation is especially true. Second, there are many job tasks that a trained behavioural scientist could not be expected to learn for research purposes. For example, it seems unreasonable to expect a researcher to spend years acquiring the skills of a pilot or banker in order to be able to investigate what happens in the cockpit of an airliner or in a boardroom. Finally, participant observation places rather severe limitations on the observers’ opportunity to record information. Existence of these conditions suggests the use of direct observation. In theory, the researcher could carry out such observation covertly, but there are few studies of organizational behaviour in which the presence of the direct observer was not known and explained to those being observed. Henry Mintzberg’s study of the work performed by chief executives of two manufacturing companies, a hospital, a school system, and a consulting firm provides an excellent example of the use of direct observation.4 At first glance, this might appear to be an inane thing to investigate. After all, everybody knows that managers plan, organize, lead, and control, or some similar combination of words. In fact, Mintzberg argues that we actually know very little about the routine, everyday behaviour managers use to achieve these vague goals. Furthermore, if we ask managers what they do (in an interview or questionnaire), they usually respond with a variation of the plan-organize-lead-control theme. Direct observation. Observational research in which the researcher observes organizational behaviour without taking part in the studied activity. 558 Research in Organizational Behaviour Appendix Mintzberg spent a week with each of his five executives, watching them at their desks, attending meetings with them, listening to their phone calls, and inspecting their mail. He kept detailed records of these activities and gradually developed a classification scheme to make sense of them. What Mintzberg found counters the common-sense view that some hold of managers—sitting behind a large desk, reflecting on their organization’s performance, and affixing their signatures to impressive documents all day. In fact, Mintzberg found that his managers actually performed a terrific amount of work and had little time for reflection. On an average day, they examined 36 pieces of mail, engaged in five telephone conversations, attended eight meetings, and made one tour of their facilities. Work-related reading encroached on home lives. These activities were varied, unpatterned, and of short duration. Half the activities lasted less than nine minutes, and 90 percent less than one hour. Furthermore, these activities tended to be directed toward current, specific issues rather than past, general issues. Finally, the managers revealed a clear preference for verbal communications, by either telephone or unscheduled face-toface meetings; in fact, two-thirds of their contacts were of this nature. In contrast, they generated an average of only one piece of mail a day. In summary, both participant and direct observation capture the depth, breadth, richness, spontaneity, and realism of organizational behaviour. However, they also share some weaknesses. One of these weaknesses is a lack of control over the environment in which the study is being conducted. Thus, Mintzberg could not ensure that unusual events would not affect the executives’ behaviour. Also, the small number of observers and situations in the typical observational study is problematic. With only one observer there is a strong potential for selective perceptions and interpretations of observed events. Since only a few situations are analyzed, the extent to which the observed behaviours can be generalized to other settings is limited. (Do most executives behave like the five that Mintzberg studied?) It is probably safe to say that observational techniques are best used to make an initial examination of some organizational event on which little information is available and to generate ideas for further investigation with more refined techniques. Correlational Techniques Correlational research. Research that attempts to measure variables precisely and examine relationships among these variables without introducing change into the research setting. Surveys. The use of questionnaires to gather data from participants who answer questions on the relevant variables. Interview. A technique in which the researcher asks respondents a series of questions to gather data on the variables of interest. Existing data. Data that is obtained from organizational records such as productivity, absence, and demographic information. Correlational research attempts to measure variables precisely and examine relationships among these variables without introducing change into the research setting. Correlational research sacrifices some of the breadth and richness of the observational techniques for more precision of measurement and greater control. It necessarily involves some abstraction of the real event that is the focus of observation in order to accomplish this precision and control. More specifically, correlational approaches differ from observational approaches in terms of the nature of the data researchers collect and the issues they investigate. The data of observational studies are most frequently observer notes. We hope that these data exhibit reliability and validity. Unfortunately, because observations are generally the products of a single individual viewing a unique event, we have very little basis on which to judge their reliability and validity. The data of correlational studies involve surveys and interviews, as well as existing data. Surveys involve the use of questionnaires to gather data from participants who answer questions on the relevant variables. The interview is a technique in which the researcher asks respondents a series of questions to gather data on the variables of interest. Interview data can be quantitative and similar to that obtained from a survey or it can be more qualitative and descriptive. The type of data obtained will depend on the purpose of the interview and the nature of the questions asked. Existing data come from organizational records and include productivity, absence, and demographic information (e.g., age, gender). Variables often measured by surveys and interviews include: Appendix ■ ■ ■ Research in Organizational Behaviour employees’ perceptions of how their managers behave on the job, the extent to which employees are satisfied with their jobs, and employees’ reports about how much autonomy they have on their jobs. It is possible to determine in advance of doing research the extent to which such measures are reliable and valid. Thus, when constructing a questionnaire to measure job satisfaction, the researcher can check its reliability by repeatedly administering it to a group of workers over a period of time. If individual responses remain fairly stable, there is evidence of reliability. Evidence of the validity of a questionnaire might come from its ability to predict which employees would quit the organization for work elsewhere. It seems reasonable that dissatisfied employees would be more likely to quit, and such an effect is partial evidence of the validity of a satisfaction measure. In addition to the nature of the data collected, it was pointed out above that correlational studies differ from observational studies in terms of the kinds of events they investigate. Although the questions investigated by observational research appear fairly specific (What maintains an “illegal” behaviour such as tapping? What do executives do?), virtually any event relevant to the question is fair game for observation. Thus, such studies are extremely broad based. Correlational research sacrifices this broadness to investigate the relationship (correlation) between specific, well-defined variables. The relationship between the variables of interest is usually stated as a hypothesis. Using the variables mentioned above, we can construct three sample hypotheses and describe how they would be tested: ■ ■ ■ Employees who are satisfied with their jobs will tend to be more productive than those who are less satisfied. To test this, a researcher might administer a reliable, valid questionnaire concerning satisfaction and obtain production data from company records. Employees who perceive their supervisor as friendly and considerate will be more satisfied with their jobs than those who do not. To test this, a researcher might use reliable, valid questionnaires or interview measures of both variables. Older employees will be absent less than younger employees. To test this, a researcher might obtain data concerning the age of employees and their absenteeism from organizational records. In each case, the researcher is interested in a very specific set of variables, and he or she devotes effort to measuring them precisely. A good example of a correlational study is that of Belle Rose Ragins and John Cotton, who studied employees’ willingness to serve as mentors to newer organizational members.5 Mentorship was defined as helping a junior person with career support and upward mobility. The major focus of the study was the relationship between gender and willingness to mentor. The authors reviewed literature that hypothesizes that women may face more barriers to becoming mentors than men because they are in a minority in many employment settings. The authors were also interested in the relationships between age, organizational rank, length of employment, and prior mentorship experience and willingness to mentor. These variables were measured with questionnaires completed by over 500 employees in three research and development organizations. The researchers found that men and women were equally willing to serve as mentors, although the women perceived more barriers (e.g., lack of qualifications and time) to being a mentor. They also found that higher rank and prior experience as a mentor or a protégé were associated with greater willingness to mentor. Notice that a study such as this could also incorporate existing data from records. For example, we might hypothesize that those with better performance evaluations would be more confident about serving as mentors. 559 560 Research in Organizational Behaviour Appendix Correlation and Causation A final important point should be made about correlational studies. Consider a hypothesis that friendly, considerate supervisors will have more productive employees than unfriendly, inconsiderate supervisors. In this case, a researcher might have some employees describe the friendliness of their supervisors on a reliable, valid questionnaire designed to measure this variable and obtain employees’ productivity levels from company records. The results of this hypothetical study are plotted in Exhibit A.1, where each dot represents an employee’s response to the questionnaire in conjunction with his or her productivity. In general, it would appear that the hypothesis is confirmed—that is, employees who describe their supervisor as friendly tend to be more productive than those who describe him or her as unfriendly. As a result of this study, should an organization attempt to select friendly supervisors or even train existing supervisors to be more friendly to obtain higher productivity? The answer is no. The training and selection proposal assumes that friendly supervisors cause their employees to be productive, and this might not be the case. Put simply, supervisors might be friendly if their employees are productive. This is a possible interpretation of the data, and it does not suggest that selection or training to make supervisors friendly will achieve higher productivity. This line of argument should not be unfamiliar to you. Heavy smokers and cigarette company lobbyists like to claim that smoking is related to the incidence of lung cancer because cancer proneness prompts smoking, rather than vice versa. The point here is that correlation does not imply causation. How can we find out which factors cause certain organizational behaviours? The answer is to perform an experiment. Independent variable. The variable that is manipulated or changed in an experiment. Dependent variable. In an experiment, the variable that is expected to vary as a result of the manipulation of the independent variable. Exhibit A.1 Hypothetical data from a correlational study of the relationship between supervisory friendliness and employee productivity. Experimental Techniques If observational research involves observing nature, and correlational research involves measuring nature, experimental research manipulates nature. In an experiment, a variable is manipulated or changed under controlled conditions, and the consequence of this manipulation for some other variable is measured. If all other conditions are truly controlled, and a change in the second variable follows the change that was introduced in the first variable, we can infer that the first change has caused the second change. In experimental language, the variable that the researcher manipulates or changes is called the independent variable. The variable that the independent variable is expected to affect is called the dependent variable. Consider the following hypothesis: The introduction of recorded music into the work setting will lead to increased productivity. In this hypothesis, the independent variable is music, which High Productivity of Employees Experimental research. Research which changes or manipulates a variable under controlled conditions and examines the consequences of this manipulation for some other variable. Employee Y Employee X Low Unfriendly Friendly Employees’ Perceptions of Supervisors Appendix Research in Organizational Behaviour is expected to affect productivity, the dependent variable. Consider another hypothesis: Stimulating, challenging jobs will increase the satisfaction of the workforce. Here, the design of the job is the independent variable and satisfaction is the dependent variable. Let us return to our hypothesis that friendly, considerate supervisors will tend to have more productive employees. If we wish to determine whether friendly supervision contributes to employee productivity, the style of supervision becomes the independent variable, and productivity becomes the dependent variable. This means that the researcher must manipulate or change the friendliness of some supervisors and observe what happens to the productivity of their employees. In practice, this would probably be accomplished by exposing the bosses to some form of human relations training designed to teach them to be more considerate and personable toward their workers. Exhibit A.2 shows the results of this hypothetical experiment. The line on the graph represents the average productivity of a number of employees whose supervisors have received our training. We see that this productivity increased and remained higher following the introduction of the training. Does this mean that friendliness indeed increases productivity and that we should proceed to train all of our supervisors in this manner? The answer is again no. We cannot be sure that something else did not occur at the time of the training to influence productivity, such as a change in equipment or job insecurity prompted by rumoured layoffs. To control this possibility, we need a control group of supervisors who are not exposed to the training, and we need productivity data for their employees. A control group is a group of research subjects who have not been exposed to the experimental treatment, in this case not exposed to the training. Ideally, these supervisors should be as similar as possible in experience and background to those who receive the training, and their employees should be performing at the same level. The results of our improved experiment are shown in Exhibit A.3. Here, we see that the productivity of the employees whose supervisors were trained increases following training, while that of the control supervisors remains constant. We can, thus, infer that the human relations training affected employee productivity. The extent to which a researcher can be confident that changes in a dependent variable are due to the independent variable is known as internal validity. Note that this is different from the validity of a measure which was discussed earlier in the chapter. Internal validity has to do with the validity of an experimental design. To return to the example above, if a control was not included in the design, then the internal validity would be low because other factors might explain the improvement in productivity. However, with a control group, one can have much more confidence that the improvement was due to the training program. Thus, internal validity increases the confidence that one has in concluding that the training program was the cause of the improvement in productivity. Control group. A group of research subjects who have not been exposed to the experimental treatment. Internal validity. The extent to which a researcher has confidence that changes in a dependent variable are due to the independent variable. Exhibit A.2 Hypothetical data from an experiment concerning human relations training. Supervisory Training Employee Productivity High 561 Low TIME Exhibit A.3 Hypothetical data from an improved experiment concerning human relations training. Research in Organizational Behaviour High Appendix Supervisory Training Trained Supervisors Employee Productivity 562 No Training (Control Group) Low TIME John Ivancevich and Herbert Lyon conducted an interesting experiment that examined the effects of a shortened workweek on the employees of a company that manufactures food-packaging equipment.6 The independent variable was the length of the workweek (4 days, 40 hours versus 5 days, 40 hours). Two of the company’s divisions were converted to a 4–40 week from a 5–40 week. A third division, remaining on the 5–40 schedule, served as a control group. Workers in the control division were similar to those in the other divisions in terms of age, seniority, education, and salary. The dependent variables (measured one month before the conversion and several times after) included the workers’ responses to a questionnaire concerning job satisfaction and stress, absence data from company records, and performance appraisals conducted by supervisors. After 12 months, several aspects of satisfaction and performance showed a marked improvement for the 4–40 workers, when compared with the 5–40 workers. However, at 25 months this edge existed for only one aspect of satisfaction, satisfaction with personal worth. The authors concluded that benefits that had been proposed for the 4–40 workweek were of short-term duration. A Continuum of Research Techniques You might reasonably wonder which of the research techniques just discussed is most effective. As shown in Exhibit A.4, these methods can be placed on a continuum ranging from rich, broad-based, and loosely controlled (observation) to specific, precise, and rigorous (experimentation). The method that researchers use to investigate organizational behaviour is dictated by the nature of the problem that interests us. In the writing of this section of the chapter, special pains were taken to choose examples of problems that were well suited to the research techniques employed to investigate them. Bensman and Gerver, as well as Mintzberg, were interested in variables that were not well defined. The variables were thus not easy to isolate and measure precisely, and observation was the appropriate technique. Furthermore, “tapping” was a controversial issue, and the researchers would have had to develop considerable trust to investigate it with questionnaires or formal interviews. Similarly, Mintzberg insists that questionnaires and interviews have failed to tell us what executives actually do. Ragins and Cotton, who studied mentoring, were interested in specific variables that were relatively easily measured. On the other hand, they were not in a position to manipulate the causes of intention to mentor. Ivancevich and Lyon were also interested in a specific set of variables, and they conducted their research on the short workweek in a situation where it was Appendix 563 Research in Organizational Behaviour Observation Correlation Experimentation Rich Specific Broad-Based Precise Loosely Controlled Rigorous both possible and ethical to manipulate the workweek. In all these cases, the research technique the researchers chose was substantially better than dependence on common sense or opinion. Combining Research Techniques Robert Sutton and Anat Rafaeli tested what might seem to be an obvious hypothesis—that friendly, pleasant behaviour on the part of sales clerks would be positively associated with store sales.7 As obvious as this might seem, it would be a good idea to confirm it before spending thousands of dollars on human relations training for clerks. The study combined correlational and observational methods. In the quantitative correlational part of the study, teams of researchers entered a large North American chain’s 576 convenience stores and, posing as shoppers, evaluated the friendliness of the sales clerks on rating scales. They also recorded other factors, such as the length of the line at the register. Existing data from company records provided the total annual sales each store recorded. When the researchers analyzed the data, the results were surprising—the “unfriendly” stores tended to chalk up higher sales! To understand this unexpected result, the authors resorted to qualitative, observational research techniques. Specifically, each author spent extensive time in many of the convenience stores directly observing transactions between customers and clerks. In addition, each spent time as a participant-observer, actually doing the sales clerk’s job. This observation resolved the mystery. The researchers found that when the stores were busy, the sales clerks tended to stop the small talk, concentrate on their work, and process customers as quickly as possible. This behaviour corresponded to customers’ expectations for fast service in a convenience store. When business was slow, clerks tended to be friendly and engage in small talk to relieve boredom. Since the busier stores generated higher sales, it is not surprising that their clerks were less friendly. In fact, further analysis of the correlational data showed that clerks were less friendly when the lines were longer. This study illustrates how two research techniques can complement each other. It also shows that correlation does not imply causation. Although sales were negatively correlated with friendliness, volume of sales affected the expression of friendliness, not the other way around. Of course, these results would probably not generalize to sales settings in which customers expect more personal attention. Issues and Concerns in Organizational Behaviour Research As in every field of study, particular issues confront researchers in organizational behaviour. Three of these issues include sampling, Hawthorne effects, and ethical concerns. Exhibit A.4 Continuum of research techniques. 564 Research in Organizational Behaviour Appendix Sampling External validity. The extent to which the results of a study generalize to other samples and settings. Random sampling. The research participants are randomly chosen from the population of interest. Researchers are usually interested in generalizing the results of their research beyond their study. The extent to which the results of a study generalize to other samples and settings is known as external validity. External validity will be greater when the results of a study are based on large, random samples. Large samples ensure that the results they obtain are truly representative of the individuals, groups, or organizations being studied and not merely the product of an extreme case or two. Random sampling means that the research participants have been randomly chosen from the population of interest. Random samples ensure that all relevant individuals, groups, or organizations have an equal probability of being studied and give confidence in the generalizability of the findings. As was noted earlier, observational studies usually involve small samples, and they are seldom randomized. Thus, generalizing from such studies is a problem. However, a well-designed observational study that answers important questions is surely superior to a large-sample, randomized correlational study that enables one to generalize about a trivial hypothesis. In experimental research, randomization means randomly assigning subjects to experimental and control conditions. To illustrate the importance of this, we can reconsider the hypothetical study on human relations training. Suppose that instead of randomly assigning supervisors to the experimental and control groups, managers nominate supervisors for training. Suppose further that to “reward” them for their long service, more-experienced supervisors are nominated for the training. This results in an experimental group containing more-experienced supervisors and a control group containing less-experienced supervisors. If supervisory experience promotes employee productivity, we might erroneously conclude that it was the human relations training that led to any improved results, and that our hypothesis is confirmed. Poor sampling due to a lack of randomization has biased the results in favour of our hypothesis. To achieve randomization, it would be a good idea to ascertain that the employees of the experimental and control supervisors were equally productive before the training began. Hawthorne Effects Hawthorne effect. A favourable response by subjects in an organizational experiment that is the result of a factor other than the independent variable that is formally being manipulated. The Hawthorne effect was discovered as a result of a series of studies conducted at the Hawthorne plant of the Western Electric Company near Chicago many years ago. As explained in Chapter 1, these studies examined the effects of independent variables, such as rest pauses, lighting intensity, and pay incentives, on the productivity of assemblers of electrical components.8 In a couple of these loosely controlled experiments, unusual results occurred. In the illumination study, both experimental and control workers improved their productivity. In another study, productivity increased and remained high despite the introduction and withdrawal of factors such as rest pauses, shortened workdays, and so on. These results gave rise to the Hawthorne effect, which might be defined as a favourable response of subjects in an organizational experiment to a factor other than the independent variable that is formally being manipulated. Researchers have concluded that this “other factor” is psychological in nature, although it is not well understood.9 Likely candidates include subjects’ reactions to special attention, including feelings of prestige, and heightened morale, and so on. The point is that researchers might misinterpret the true reason for any observed change in behaviour because research subjects can have unmeasured feelings about their role in the research. To return to the human relations training experiment, a Hawthorne effect might occur if the experimental subjects are grateful to management for selecting them for this special training and resolve to work harder back on the job. The supervisors might put in longer hours thinking up ways to improve productivity that have nothing to do with the training they received. However, the researcher could easily conclude that the human relations training improved productivity. Appendix Research in Organizational Behaviour 565 It is very difficult to prevent Hawthorne effects. However, it is possible, if expensive, to see whether they have occurred. To do so, investigators establish a second experimental group that receives special treatment and attention but is not exposed to the key independent variable. In the human relations experiment, this could involve training that is not expected to increase productivity. If the productivity of the supervisors’ employees in both experimental groups increases equally, the Hawthorne effect is probably present. If productivity increases only in the human relations training condition, it is unlikely to be due to the Hawthorne effect. Ethics Researchers in organizational behaviour, no matter who employs them, have an ethical obligation to do rigorous research and to report that research accurately.10 In all cases, the psychological and physical well being of the research subjects is of prime importance. In general, ethical researchers avoid unnecessary deception, inform participants about the general purpose of their research, and protect the anonymity of research subjects. For example, in a correlational study involving the use of questionnaires, investigators should explain the general reason for the research and afford potential subjects the opportunity to decline participation. If names or company identification numbers are required to match responses with data in personnel files (e.g., absenteeism or subsequent turnover), investigators must guarantee that they will not make individual responses public. In some observation studies and experiments, subjects may be unaware that their behaviour is under formal study. In these cases, researchers have special obligations to prevent negative consequences for subjects. Ethical research has a practical side as well as a moral side. Good cooperation from research subjects is necessary to do good research. Such cooperation is easier to obtain when people are confident that ethical procedures are the rule, not the exception. Learning Objectives Checklist 1. All research in organizational behaviour begins with a basic question about work or organizations. Frequently, researchers express the question as a hypothesis, a formal statement of the expected relationship between two variables. Variables are simply measures that can take on two or more values. 2. Careful measurement of variables is important in research. Reliability is an index of the consistency of a research subject’s responses. Validity is an index of the extent to which a measure truly reflects what it is supposed to measure. Convergent validity exists when there is a strong relationship between different measures of the same variable. Discriminant validity exists when there is a weak relationship between measures of different variables. 3. In observational research, one or a few observers assess one or a few instances of organizational behaviour in its natural setting. In participant observation, the observer actually takes part in the activity being observed. In direct observation, the assessment occurs without the active participation of the researcher. 4. Compared with observation, correlational research techniques attempt to measure the variables in question more precisely by using questionnaires, interviews, and existing data. No change is introduced into the research setting. One problem with correlational research is its inability to reveal which variables cause other variables. Researchers use experiments to overcome this problem. 5. In experimental research, the investigator actually changes or manipulates some factor in the organizational setting and measures the effect that this 566 Research in Organizational Behaviour manipulation has on behaviour. In experimental language, the variable that the researcher manipulates or changes is called the independent variable and the variable that the independent variable is expected to affect is called the dependent variable. Causation can be inferred from a carefully designed experiment that has high internal validity. Internal validity refers to the confidence that the researcher has in being able to conclude that changes in the dependent variable are due to the independent variable. 6. The method that researchers use to investigate organizational behaviour is dictated by the nature of the problem under investigation. When variables are not well defined and not easy to isolate and measure precisely, observation is an appropriate technique. Some of the weaknesses of observational research include a lack of control over the environment in which the study is being conducted and the small number of observers and situations in the typical observational study. Observational techniques are best used to make an initial examination of some organizational event on which little information is available and to generate ideas for further investigation with more refined techniques. When the researcher is interested in specific variables that are well defined and relatively easy to measure but cannot be manipulated, correlational research is an appropriate technique. Correlational research provides more precision and greater control than observational techniques; however, it cannot be used to study causation. When the researcher is interested in causation and the effect of an independent variable on a dependent variable and it is both possible and ethical to manipulate the independent variable, experimental research is an appropriate technique. Experimental research provides the greatest amount Appendix of rigour but sacrifices the breadth and richness of less rigorous techniques such as observational research. 7. External validity refers to the extent to which the results of a study generalize to other samples and settings. External validity will be greater when the results of a study are based on large, random samples. A random sample means that the research participants have been randomly chosen from the population of interest. This ensures that all relevant individuals, groups, or organizations have an equal probability of being studied and give confidence in the generalizability of the findings. 8. The Hawthorne effect refers to a favourable response of subjects in an organizational experiment to a factor other than the independent variable that is formally being manipulated. Researchers have concluded that this “other factor” is psychological in nature, although it is not well understood. Likely candidates include subjects’ reactions to special attention, including feelings of prestige, and heightened morale, and so on. The point is that researchers might misinterpret the true reason for any observed change in behaviour because research subjects can have unmeasured feelings about their role in the research. 9. Researchers in organizational behaviour have an ethical obligation to do rigorous research and to report that research accurately. In all cases, the psychological and physical well being of the research subjects is of prime importance. In general, ethical researchers avoid unnecessary deception, inform participants about the general purpose of their research, and protect the anonymity of research subjects. ...
View Full Document

This note was uploaded on 10/01/2010 for the course FGT mba12ehtp taught by Professor Angwi during the Spring '10 term at Télécom Paris.

Ask a homework question - tutors are online