HOMEWORK

© P

ix m

an n/

Im ag

ez oo

/ G

et ty

Im ag

es

9

Evaluating and Institutionalizing Organization Development Interventions

learning objectives

Illustrate the research design and measurement issues associated with evaluating organization development (OD) interventions.

Explain the key elements in the process of institutionalizing OD interventions.

This chapter focuses on the final stage of theorganization development cycle—evaluationand institutionalization. Evaluation is con- cerned with providing feedback to practitioners and organization members about the progress and impact of interventions. Such information may sug- gest the need for further diagnosis and modification of the change program, or it may show that the intervention is successful. Institutionalization is a process for maintaining a particular change for an appropriate period of time. It ensures that the

results of successful change programs persist over time.

Evaluation processes consider both the implementation success of the intended intervention and the long-term results it produces. Two key aspects of effective evaluation are measurement and research design. The persistence of intervention effects is examined in a framework showing the organization characteristics, intervention dimensions, and processes contributing to institutionalization of OD interventions in organizations.

9-1 Evaluating Organization Development Interventions Assessing OD interventions involves judgments about whether an intervention has been implemented as intended and, if so, whether it is having desired results. Managers investing resources in OD efforts increasingly are being held accountable for results— being asked to justify the expenditures in terms of hard, bottom-line outcomes. More and more, managers are asking for rigorous assessment of OD interventions and are using the results to make important resource allocation decisions about OD, such as whether to continue to support the change program, to modify or alter it, or to terminate it and try something else.

Traditionally, OD evaluation has been discussed as something that occurs after the intervention. Chapters 10 through 20, for example, present evaluative research about the interventions after discussions of the respective change programs. That view can be

207

misleading, however. Decisions about the measurement of relevant variables and the design of the evaluation process should be made early in the OD cycle so that evaluation choices can be integrated with intervention decisions.

There are two distinct types of OD evaluation: one intended to guide the implemen- tation of interventions and another to assess their overall impact. The key issues in eval- uation are measurement and research design.

9-1a Implementation and Evaluation Feedback Most discussions and applications of OD evaluation imply that evaluation is something done after intervention. It is typically argued that once the intervention is implemented, it should be evaluated to discover whether it is producing the intended effects. For exam- ple, it might be expected that a job enrichment program would lead to higher employee satisfaction and performance. After implementing job enrichment, evaluation would involve assessing whether these positive results indeed did occur. This after- implementation view of evaluation is only partially correct. It assumes that interventions have been implemented as intended and that the key purpose of evaluation is to assess their effects. However, in many, if not most, organization development programs, imple- menting interventions cannot be taken for granted.1 Most OD interventions require sig- nificant changes in people’s behaviors and ways of thinking about organizations, but they typically offer only broad prescriptions for how such changes are to occur. For example, job enrichment (see Chapter 14) calls for adding discretion, variety, and meaningful feedback to people’s jobs. Implementing such changes requires considerable learning and experimentation as employees and managers discover how to translate these general prescriptions into specific behaviors and procedures. This learning process involves much trial and error and needs to be guided by information about whether behaviors and procedures are being changed as intended.2 Consequently, we should expand our view of evaluation to include both during-implementation assessments about if and how well changes are actually being implemented and after-implementation evaluation of whether they are producing expected results.

Both kinds of evaluation provide organization members with feedback about interven- tions. Evaluation aimed at guiding implementation may be called implementation feedback, and assessment intended to discover intervention outcomes may be called evaluation feed- back. Figure 9.1 shows how the two kinds of feedback fit with the diagnostic and interven- tion stages of OD. The application of OD to a particular organization starts with a thorough diagnosis of the situation (Chapters 5 and 6), which helps identify particular organizational problems, areas for improvement, or strengths to leverage as well as the likely drivers underlying them. Next, from an array of possible interventions (Chapters 10 through 20), one or some set is chosen as a means of improving the organization. The choice is based on knowledge linking interventions to diagnosis (Chapter 7) and change management (Chapter 8).

In most cases, the chosen intervention provides only general guidelines for organiza- tional change, leaving managers and employees with the task of translating those guide- lines into specific behaviors and procedures. Implementation feedback informs this process by supplying data about the different features of the intervention itself, percep- tions of the people involved, and data about the immediate effects of the intervention. These data, collected repeatedly and at short intervals, provide a series of snapshots about how the intervention is progressing. Organization members can use this informa- tion, first, to gain a clearer understanding of the intervention (the kinds of behaviors and procedures required to implement it) and, second, to plan for the next implementation

208 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

steps. This feedback cycle might proceed for several rounds, with each round providing members with knowledge about the intervention and ideas for the next stage of implementation.

Once implementation feedback informs organization members that the intervention is sufficiently in place and accepted, evaluation feedback begins. In contrast to imple- mentation feedback, it is concerned with the overall impact of the intervention and with whether resources should continue to be allocated to it or to other possible inter- ventions. Evaluation feedback takes longer to gather and interpret than does implemen- tation feedback. It typically includes a broad array of outcome measures, such as performance, job satisfaction, productivity, and turnover. Negative results on these mea- sures tell members either that the initial diagnosis was seriously flawed or that the wrong intervention was chosen. Such feedback might prompt additional diagnosis and a search for a more effective intervention. Positive results, on the other hand, tell members that the intervention produced expected outcomes and might prompt a search for ways to institutionalize the changes, making them a permanent part of the organization’s normal functioning.

An example of a job enrichment intervention helps to clarify the OD stages and feedback linkages shown in Figure 9.1. Suppose the initial diagnosis reveals that employee performance and satisfaction are low and that jobs being overly structured and routinized are an underlying cause of this problem. An inspection of alternative interventions to improve productivity and satisfaction suggests that job enrichment might be applicable for this situation. Existing job enrichment theory proposes that increasing employee discretion, task variety, and feedback can lead to improvements in work quality and attitudes and that this job design and outcome linkage is especially strong for employees who have growth needs—needs for challenge, autonomy, and development. Initial diagnosis suggests that most of the employees have high growth

FIGURE 9.1

Implementation and Evaluation Feedback

© Ce

ng ag

e Le

ar ni

ng

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 209

needs and that the existing job designs prevent the fulfillment of these needs. Therefore, job enrichment seems particularly suited to this situation.

Managers and employees now start to translate the general prescriptions offered by job enrichment theory into specific behaviors and procedures. At this stage, the interven- tion is relatively broad and must be tailored to fit the specific situation. To implement the intervention, employees might decide on the following organizational changes: job discretion can be increased through more participatory styles of supervision; task variety can be enhanced by allowing employees to inspect their job outputs; and feedback can be made more meaningful by providing employees with quicker and more specific informa- tion about their performances.

After three months of trying to implement these changes, the members use imple- mentation feedback to see how the intervention is progressing. Questionnaires and interviews (similar to those used in diagnosis) are administered to measure the differ- ent features of job enrichment (discretion, variety, and feedback) and to assess employ- ees’ reactions to and understanding of the changes. Company records are analyzed to show the short-term effects on productivity of the intervention. The data reveal that productivity and satisfaction have changed very little since the initial diagnosis. Employee perceptions of job discretion and feedback also have shown negligible change and employees seem confused about the expectations of managers, but percep- tions of task variety have shown significant improvement. In-depth discussion and analysis of this first round of implementation feedback help supervisors gain a better feel for the kinds of behaviors needed to move toward a participatory leadership style. This greater clarification of one feature of the intervention leads to a decision to involve the supervisors in leadership training to develop the skills and knowledge needed to lead participatively. A decision also is made to make job feedback more meaningful by translating such data into simple bar graphs, rather than continuing to provide voluminous statistical reports.

After these modifications have been in effect for about three months, members gather a second round of implementation feedback to see how the intervention is progressing. The data now show that productivity and satisfaction have moved moderately higher than in the first round of feedback and that employee perceptions of task variety and feed- back are both high. Employee perceptions of discretion, however, remain relatively low. Members conclude that the variety and feedback dimensions of job enrichment are suffi- ciently implemented but that the discretion component needs further improvement. They decide to put more effort into supervisory training and to ask OD practitioners to provide counseling and coaching to supervisors about their leadership styles.

After four more months, a third round of implementation feedback is sought. The data now show that satisfaction and performance are significantly higher than in the first round of feedback and moderately higher than in the second round. The data also show that discretion, variety, and feedback are all high, suggesting that the job enrichment intervention has been successfully implemented. Now evaluation feedback is used to assess the overall effectiveness of the program.

The evaluation feedback includes all the data from the satisfaction and performance measures used in the implementation feedback. Because both the immediate and broader effects of the intervention are being evaluated, additional outcomes are examined, such as employee absenteeism, maintenance costs, and reactions of other organizational units not included in job enrichment. The full array of evaluation data might suggest that after one year from the start of implementation, the job enrichment program is having the expected effects and thus should be continued and made more permanent.

210 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

9-1b Measurement Providing useful implementation and evaluation feedback involves two activities: select- ing the appropriate variables and designing good measures of them.

Selecting Appropriate Variables Ideally, the variables measured in OD evaluation should derive from the theory or conceptual model underlying the intervention. The model should incorporate the key features of the intervention as well as its expected results. The general diagnostic models described in Chapter 5 meet this criterion, as do the more specific models introduced in Chapters 10 through 20. For example, the job- level diagnostic model described in Chapter 5 proposes several major features of work: task variety, feedback, and autonomy. The theory argues that high levels of these ele- ments can be expected to result in high levels of work quality and satisfaction. In addi- tion, as we shall see in Chapter 14, the strength of this relationship varies with the degree of employee growth needs: the higher the need, the more that job enrichment produces positive results.

The job-level diagnostic model suggests a number of measurement variables for implementation and evaluation feedback. Whether the intervention is being implemen- ted could be assessed by determining how many job descriptions have been rewritten to include more responsibility or how many organization members have received cross- training in other job skills. Evaluation of the immediate and long-term impact of job enrichment would include measures of employee performance and satisfaction over time. Again, these measures would likely be included in the initial diagnosis, when the company’s problems or areas for improvement are discovered.

Measuring both intervention and outcome variables is necessary for implementation and evaluation feedback. Unfortunately, there has been a tendency in OD to measure only outcome variables while neglecting intervention variables altogether.3 It generally is assumed that the intervention has been implemented, and attention, therefore, is directed to its impact on such organizational outcomes as performance, absenteeism, and satisfaction. As argued earlier, implementing OD interventions generally takes con- siderable time and learning. It must be empirically determined that the intervention has been implemented; it cannot simply be assumed. Implementation feedback serves this purpose, guiding the implementation process and helping to interpret outcome data. Outcome measures are ambiguous without knowledge of how well the intervention has been implemented. For example, a negligible change in measures of performance and satisfaction could mean that the wrong intervention has been chosen, that the correct intervention has not been implemented effectively, or that the wrong variables have been measured. Measurement of the intervention variables helps determine the correct interpretation of outcome measures.

As suggested above, the selection of intervention variables to be measured should derive from the conceptual framework underlying the OD intervention. OD research and theory increasingly have come to identify specific organizational changes needed to imple- ment particular interventions (much of that information is discussed in Chapters 10 through 20). These variables should guide not only implementation of the intervention but also choices about what change variables to measure for evaluative purposes. Addi- tional sources of knowledge about intervention variables can be found in the numerous references at the end of each of the chapters on intervention in this book and in several of the books in the Wiley Series on Organizational Assessment and Change.4

The choice of what outcome variables to measure also should be dictated by inter- vention theory, which specifies the kinds of results that can be expected from particular

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 211

change programs. Again, the material in this book and elsewhere identifies numerous outcome measures, such as job satisfaction, intrinsic motivation, organizational commit- ment, absenteeism, turnover, and productivity.

Historically, OD assessment has focused on attitudinal outcomes, such as job satisfac- tion, while neglecting hard measures, such as performance. Increasingly, however, managers and researchers are calling for development of behavioral measures of OD outcomes. Man- agers are interested primarily in applying OD to change work-related behaviors that involve joining, remaining, and producing at work, and are assessing OD more frequently in terms of such bottom-line results. Macy and Mirvis have done extensive research to develop a standardized set of behavioral outcomes for assessing and comparing intervention results.5

Table 9.1 lists 11 outcomes, including their behavioral definitions and recording categories. The outcomes are in two broad categories: participation-membership, including absentee- ism, tardiness, turnover, internal employment stability, and strikes and work stoppages; and performance on the job, including productivity, quality, grievances, accidents, unsched- uled machine downtime and repair, material and supply overuse, and inventory shrinkage. All of the outcomes should be important to most managers, and they represent generic descriptions that can be adapted to both industrial and service organizations.

Designing Good Measures Each of the measurement methods described in Chapter 6—questionnaires, interviews, observations, and unobtrusive measures—has advantages and disadvantages. Many of these characteristics are linked to the extent to which a measurement is operationally defined, reliable, and valid. These assessment characteristics are discussed below.

Operational Definition. A good measure is operationally defined; that is, it specifies the empirical data needed, how they will be collected and, most important, how they will be converted from data to information. For example, Macy and Mirvis developed operational definitions for the behavioral outcomes listed in Table 9.1 (see Table 9.2).6

They consist of specific computational rules that can be used to construct measures for each of the behaviors. Most of the behaviors are reported as rates adjusted for the num- ber of employees in the organization and for the possible incidents of behavior. These adjustments make it possible to compare the measures across different situations and time periods. These operational definitions should have wide applicability across both industrial and service organizations, although some modifications, deletions, and addi- tions may be necessary for a particular application.

Operational definitions are extremely important in measurement because they pro- vide precise guidelines about what characteristics of the situation are to be observed and how they are to be used. They tell OD practitioners and organization members exactly how diagnostic, intervention, and outcome variables will be measured.

Reliability. Reliability concerns the extent to which a measure represents the “true” value of a variable—that is, how accurately the operational definition translates data into information. For example, there is little doubt about the accuracy of the number of cars leaving an assembly line as a measure of plant productivity. Although it is possible to miscount, there can be a high degree of confidence in the measurement. On the other hand, when people are asked to rate their level of job satisfaction on a scale of 1 to 5, there is considerable room for variation in their response. They may just have had an argument with their supervisor, suffered an accident on the job, been rewarded for high levels of productivity, or been given new responsibilities. Each of these events can sway the response to the question on any given day. The individuals’ “true” satisfaction score is difficult to discern from this one question and the measure lacks reliability.7

212 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

TABLE 9.1

Behavioral Outcomes for Measuring OD Interventions: Definitions and Recording Categories

Behavioral Definitions Recording Categories

Absenteeism: each absence or illness over four hours

Voluntary: short-term illness (less than three consecutive days), personal business, family illness

Involuntary: long-term illness (more than three consecutive days), funerals, out-of-plant accidents, lack of work (temporary layoff), presanctioned days off

Leaves: medical, personal, maternity, military, and other (e.g., jury duty)

Tardiness: each absence or illness under four hours

Voluntary: same as absenteeism Involuntary: same as absenteeism

Turnover: each movement beyond the organizational boundary

Voluntary: resignation Involuntary: termination, disqualification, requested resignation,

permanent layoff, retirement, disability, death

Internal employment stability: each move- ment within the organizational boundary

Internal movement: transfer, promotion, promotion with transfer Internal stability: new hires, layoffs, rehires

Strikes and work stoppages: each day lost as a result of strike or work stoppage

Sanctioned: union-authorized strike, company-authorized lockout Unsanctioned: work slowdown, walkout, sitdown

Accidents and work-related illness: each recordable injury, illness, or death from a work-related accident or from exposure to the work environment

Major: OSHA accident, illness, or death which results in medical treatment by a physician or registered professional person understanding orders from a physician

Minor: non-OSHA accident or illness which results in one-time treat- ment and subsequent observation not requiring professional care

Revisits: OSHA and non-OSHA accident or illness which requires subsequent treatment and observation

Grievances: written grievance in accordance with labor–management contract

Stage: recorded by step (first through arbitration)

Productivity:* resources used in production of acceptable outputs (comparison of inputs with outputs)

Output: product or service quantity (units or $) Input: direct and/or indirect (labor in hours or $)

Production quality: resources used in production of unacceptable outputs

Resource utilized: scrap (unacceptable in-plant products in units or $); customer returns (unacceptable out-of-plant products in units or $); recoveries (salvageable products in units or $); rework (additional direct and/or indirect labor in hours or $)

Downtime: unscheduled breakdown of machinery

Downtime: duration of breakdown (hours or $) Machine repair: nonpreventive maintenance ($)

Inventory, material, and supply variance: unscheduled resource utilization

Variance: over- or under-utilization of supplies, materials, inventory (resulting from theft, inefficiency, and so on)

*Reports only labor inputs.

SOURCE: B. Macy and P. Mirvis, “Organizational Change Efforts: Methodologies for Assessing Organizational Effectiveness and Program Costs Versus Benefits,” Evaluation Review 6, pp. 306–10. © 1982 by Sage Publications, Inc. Reprinted by permission of Sage Publications, Inc.

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 213

TABLE 9.2

Behavioral Outcomes for Measuring OD Interventions: Measures and Computational Formula

Behavioral Measure* Computational Formula

Absenteeism rate** (monthly) ∑ Absence days Average workforce size Working days

Tardiness rate** (monthly) ∑ Tardiness incidents Average workforce size Working days

Turnover rate (monthly) ∑ Turnover incidents Average workforce size

Internal stability rate (monthly) ∑ Internal movement incidents Average workforce size

Strike rate (yearly) ∑ Striking Workers Strike days Average workforce size Working days

Accident rate (yearly) ∑ of Accidents illnesses Total yearly hours worked

200,000***

Grievance rate (yearly) ∑ Grievance incidents Plant: Average workforce size

Individual: ∑ Aggrieved individuals Average workforce size

Productivity:**** Total Below standard Below budget Variance Per employee

Output of goods or services units or $ Direct and or indirect labor hours or $ Actual versus engineered standard Actual versus budgeted standard Actual versus budgeted variance Output/average workforce size

Quality:**** Total Below standard Below budget Variance Per employee

Scrap Customer returns Rework Recoveries ($, units, or hours) Actual versus engineered standard Actual versus budgeted standard Actual versus budgeted variance Total/average workforce size

Downtime Labor ($) Repair costs or dollar value of replaced equipment ($)

Inventory, supply, and material usage Variance (actual versus standard utilization) ($)

*All measures reflect the number of incidents divided by an exposure factor that represents the number of employees in the organization and the possible incidents of behavior (e.g., for absenteeism, the average workforce size × the number of working days). Mean monthly rates (i.e., absences per workday) are computed and averaged for absenteeism, leaves, and tardiness for a yearly figure and summed for turnover, grievances, and internal employment stability for a yearly figure. The term rate refers to the number of incidents per unit of employee exposure to the risk of such incidences during the analysis interval.

**Sometimes combined as number of hours missing/average workforce size × working days. ***Base for 100 full-time equivalent workers (40 hours × 50 weeks). ****Monetary valuations can be expressed in labor dollars, actual dollar costs, sales dollars; overtime dollar valuations can be

adjusted to base year dollars to control for salary, raw material, and price increases.

SOURCE: B. Macy and P. Mirvis, “Organizational Change Efforts: Methodologies for Assessing Organizational Effectiveness and Program Costs Versus Benefits,” Evaluation Review 6, pp. 306–10. © 1982 by Sage Publications, Inc. Reprinted by permission of Sage Publications, Inc.

214 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

OD practitioners can improve the reliability of their measures in four ways. First, rigorously and operationally define the chosen variables. Clearly specified operational definitions contribute to reliability by explicitly describing how collected data will be converted into information about a variable. An explicit description helps to allay the organization’s concerns about how the information was collected and coded.

Second, use multiple methods to measure a particular variable. As discussed in Chapter 6, the use of questionnaires, interviews, observations, and unobtrusive measures can improve reliability and result in a more comprehensive understanding of the organi- zation. Because each method contains inherent biases, several different methods can be used to triangulate on dimensions of organizational issues. If the independent measures converge or show consistent results, the dimensions or problems likely have been diag- nosed accurately.8

Third, use multiple items to measure the same variable on a questionnaire. For exam- ple, in Hackman and Oldham’s Job Diagnostic Survey for measuring job characteristics (Chapter 14), the intervention variable “autonomy” is operationally defined by the average of respondents’ answers to the following three questions (measured on a 7-point scale):9

1. The job permits me to decide on my own how to go about doing the work. 2. The job denies me any chance to use my personal initiative or judgment in carrying

out the work. [reverse scored] 3. The job gives me considerable opportunity for independence and freedom in how

I do the work.

By asking more than one question about “autonomy,” the survey increases the accuracy of its measurement of this variable. Statistical analyses (called psychometric tests) are readily available for assessing the reliability of perceptual measures, and OD practitioners should apply these methods or seek assistance from those who can apply them.10 Similarly, there are methods for analyzing the content of interview and observational data, and OD evaluators can use these methods to categorize such information so that it can be understood and replicated.11

Fourth, use standardized instruments. A growing number of standardized question- naires are available for measuring OD intervention and outcome variables. For example, the Center for Effective Organizations at the University of Southern California (http://ceo .usc.edu) and the Institute for Social Research at the University of Michigan (http://home .isr.umich.edu) have developed comprehensive survey instruments to measure the features of many of the OD interventions described in this book, as well as their attitudinal out- comes.12 Considerable research and testing have gone into establishing measures that are reliable and valid. These survey instruments can be used for initial diagnosis, for guiding implementation of interventions, and for evaluating immediate and long-term outcomes.

Validity. Validity concerns the extent to which a measure actually reflects the variable it is intended to measure. For example, the number of cars leaving an assembly line might be a reliable measure of plant productivity, but it may not be a valid measure. The number of cars is only one aspect of productivity; they may have been produced at an unacceptably high cost or at exceptionally low quality. Because the number of cars does not account for cost and quality, it is not a completely valid measure of plant productivity.

OD practitioners can increase the validity of their measures in several ways. First, ask colleagues and organization members if a proposed measure actually represents a particular variable. This is called face validity or content validity. If experts and members agree that the measure reflects the variable of interest, then there is increased confidence in the measure’s validity. Second, use multiple measures of the same variable, as

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 215

described in the section about reliability, to make preliminary assessments of the mea- sure’s criterion or convergent validity. That is, if several different measures of the same variable correlate highly with each other, especially if one or more of the other measures have been validated in prior research, then there is increased confidence in the measure’s validity. A special case of criterion validity, called discriminant validity, exists when the proposed measure does not correlate with measures that it is not supposed to correlate with. For example, there is no good reason for daily measures of assembly line produc- tivity to correlate with daily air temperature. The lack of a correlation would be one indi- cator that the number of cars is measuring productivity and not some other variable. Finally, predictive validity is demonstrated when the variable of interest accurately fore- casts another variable over time. For example, a measure of team cohesion can be said to be valid if it accurately predicts improvements in team performance in the future.

It is difficult, however, to establish the validity of a measure until it has been used. To address this concern, OD practitioners should make heavy use of content validity processes and use measures that already have been validated. For example, presenting proposed measures to colleagues and organization members for evaluation prior to mea- surement has several positive effects: It builds ownership and commitment to the data collection process and improves the likelihood that the client system will find the data meaningful. Using measures that have been validated through prior research improves confidence in the results and provides a standard that can be used to validate any new measures used in collecting the data.

9-1c Research Design In addition to measurement, OD practitioners must make choices about how to design the evaluation to achieve valid results. The key issue is how to design the assessment to show whether the intervention did in fact produce the observed results. This is called internal validity. The secondary question of whether the intervention would work simi- larly in other situations is referred to as external validity. External validity is irrelevant without first establishing an intervention’s primary effectiveness, so internal validity is the essential minimum requirement for assessing OD interventions. Unless managers can have confidence that the outcomes are the result of the intervention, they have no rational basis for making decisions about accountability and resource allocation.

Assessing the internal validity of an intervention is, in effect, testing a hypothesis— namely, that specific organizational changes lead to certain outcomes. Moreover, testing the validity of an intervention hypothesis means that alternative hypotheses or explana- tions of the results must be rejected. That is, to claim that an intervention is successful, it is necessary to demonstrate that other explanations—in the form of rival hypotheses—do not account for the observed results. For example, if a job enrichment program appears to increase employee performance, such other possible explanations as new technology, improved raw materials, or new employees must be eliminated.

Accounting for rival explanations is not a precise, controlled, experimental process such as might be found in a research laboratory.13 OD interventions often have a number of features that make it difficult to determine whether they produced the observed results. They are complex and often involve several interrelated changes that obscure whether indi- vidual features or combinations of features are accounting for the results. Many OD inter- ventions are long-term projects and take considerable time to produce desired outcomes. The longer the time period of the change program, the greater are the chances that other factors, such as technology improvements, will emerge to affect the results. Finally, OD interventions usually are applied to existing work units rather than to randomized groups

216 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

of organization members. Ruling out alternative explanations associated with randomly selected intervention and comparison groups is, therefore, difficult.

Given the problems inherent in assessing OD interventions, practitioners have turned to quasi-experimental research designs.14 These designs are not as rigorous and controlled as are randomized experimental designs, but they allow evaluators to rule out many rival explanations for OD results other than the intervention itself. Although several quasi-experimental designs are available, those with the following three features are particularly powerful for assessing changes:

1. Longitudinal measurement. This involves measuring results repeatedly over rela- tively long time periods. Ideally, the data collection should start before the change program is implemented and continue for a period considered reasonable for pro- ducing expected results.

2. Comparison unit. It is always desirable to compare results in the intervention situ- ation with those in another situation where no such change has taken place. Although it is never possible to get a matching group identical to the intervention group, most organizations include a number of similar work units that can be used for comparison purposes.

3. Statistical analysis. Whenever possible, statistical methods should be used to rule out the possibility that the results are caused by random error or chance. Various statistical techniques are applicable to quasi-experimental designs, and OD practi- tioners should apply these methods or seek help from those who can apply them.

Table 9.3 provides an example of a quasi-experimental design having these three fea- tures. The intervention is intended to reduce employee absenteeism. Measures of absentee- ism are taken from company monthly records for both the intervention and comparison groups. The two groups are similar yet geographically separate subsidiaries of a multiplant company. Table 9.3 shows each plant’s monthly absenteeism rate for four consecutive months both before and after the start of the intervention. The plant receiving the inter- vention shows a marked decrease in absenteeism in the months following the intervention, whereas the control plant shows comparable levels of absenteeism in both time periods. Statistical analyses of these data suggest that the abrupt downward shift in absenteeism fol- lowing the intervention was not attributable to chance variation. This research design and the data provide relatively strong evidence that the intervention was successful.

Quasi-experimental research designs using longitudinal data, comparison groups, and statistical analysis permit reasonable assessments of intervention effectiveness. Repeated measures often can be collected from company records without directly involv- ing members of the experimental and comparison groups. These unobtrusive measures are especially useful in OD assessment because they do not interact with the intervention

TABLE 9.3

Quasi-Experimental Research Design

Monthly Absenteeism (%)

Sept. Oct. Nov. Dec. Jan. Feb. Mar. Apr.

Intervention group 5.1 5.3 5.0 5.1 Start of intervention 4.6 4.0 3.9 3.5

Comparison group 2.5 2.6 2.4 2.5 2.6 2.4 2.5 2.5

© Ce

ng ag

e Le

ar ni

ng

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 217

and affect the results. Measures that are more obtrusive, such as questionnaires and interviews, are reactive and can sensitize people to the intervention. When this happens, it is difficult to know whether the observed findings are the result of the intervention, the measuring methods, or some combination of both.

Multiple measures of intervention and outcome variables should be applied to mini- mize measurement and intervention interactions. For example, obtrusive measures such as questionnaires could be used sparingly, perhaps once before and once after the inter- vention. Unobtrusive measures, such as the behavioral outcomes shown in Tables 9.1 and 9.2, could be used repeatedly, thus providing a more extensive time series than the questionnaires. When used together, the two kinds of measures should produce accurate and nonreactive evaluations of the intervention.

The use of multiple measures also is important in assessing perceptual changes result- ing from interventions. Considerable research has identified three types of change—alpha, beta, and gamma—that occur when using self-report, perceptual measures.15

Alpha change refers to movement along a measure that reflects stable dimensions of reality. For example, comparative measures of perceived employee discretion might show an increase after a job enrichment program. If this increase represents alpha change, it can be assumed that the job enrichment program actually increased employee percep- tions of discretion.

Beta change involves the recalibration of the intervals along some constant measure of reality. For example, before-and-after measures of perceived employee discretion can decrease after a job enrichment program. If beta change is involved, it can explain this apparent failure of the intervention to increase discretion. The first measure of discretion may accurately reflect the individual’s belief about the ability to move around and talk to fellow workers in the immediate work area. During implementation of the job enrich- ment intervention, however, the employee may learn that the ability to move around is not limited to the immediate work area. At a second measurement of discretion, the employee, using this new and recalibrated understanding, may rate the current level of discretion as lower than before.

Gamma change involves fundamentally redefining the measure as a result of an OD intervention. In essence, the framework within which a phenomenon is viewed changes. For example, the presence of gamma change would make it difficult to compare mea- sures of employee discretion taken before and after a job enrichment program. The mea- sure taken after the intervention might use the same words, but they represent an entirely different concept. As described above, the term “discretion” may originally refer to the ability to move about the department and interact with other workers. After the intervention, discretion might be defined in terms of the ability to make decisions about work rules, work schedules, and productivity levels. In sum, the job enrichment interven- tion changed the way discretion is perceived and how it is evaluated.

These three types of change apply to perceptual measures. When changes other than alpha ones occur, interpreting measurement changes becomes far more difficult. Potent OD interventions may produce both beta and gamma changes, and this severely compli- cates interpretations of findings reporting change or no change. Further, the distinctions among the three different types of change suggest that the heavy reliance on question- naires, so often cited in the literature, should be balanced by using other measures, such as interviews and unobtrusive records. Analytical methods have been developed to assess the three kinds of change, and OD practitioners should gain familiarity with these recent techniques.16

Application 9.1 describes the implementation and evaluation feedback that were developed for the Alegent Health project. It is a good example of how data can be used

218 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

a p

p li

ca ti

o n

9 1 EVALUATING CHANGE AT ALEGENT HEALTH

I n July and August of 2005, Alegent Health (AH) conducted six large group interventions or “decision accelerators” (DAs) to generate innovative strategies for its six clinical ser-

vice areas. Researchers at USC’s Center for Effective Organizations contracted with AH to assess the impact of the interventions and to help the organization learn how to leverage fur- ther change. The applications in Chapter 4 described how the researchers entered and contracted with the organization, and Applica- tion 6.1 described the data collection and anal- ysis process. In this application, we describe the implementation and evaluation feedback the researchers generated.

In terms of implementation feedback, the collected data described perceptions of change progress and employee involvement. For exam- ple, executive interviews and surveys from peo- ple who participated in “review DAs” to reflect on implementation progress and plan future changes supported a positive assessment of overall prog- ress. People generally agreed that the implemen- tation of the clinical strategies was going well. They were positive about the social capital that had been created by the DAs, especially between administrators and physicians, as well as the evi- dence of culture change. Nearly everyone in the organization believed that the clinical strategies were the “right thing to do.” In contrast to these positive findings, there was some concern about feeling involved in the change.

That is, the DAs were a great energizer for the organization, generated comprehensive strategies, and catalyzed important changes. However, the data also contained some reserva- tions about the organization’s ability to leverage the changes. As described in Application 6.1, the implementation approaches were mostly informal; the organization was afraid that too many dedicated change processes and systems might slow down the change process. The data revealed a more complex set of issues.

First, executives and some physicians under- stood the roles, decisions, and processes related to implementation more than operational man- agers and other physicians. The six clinical ser- vices areas studied described an intense period

of business planning following the DAs. Senior management, all of whom had participated in the original six DAs, drove this process and were quite clear about the resulting priorities and initiatives. This clarity, however, was not widely shared by the hospital COOs, many physicians, and many operational managers. This resulted in a percep- tion of a strong connection between strategy for- mulation and implementation at the top of the organization, but a weak perception in the middle of the organization. Managers and nurses felt overwhelmed—they heard about change with lit- tle context, and believed that the speed of change was slower rather than faster because “every- thing was important.” Similarly, many physicians who were energized by the DAs wanted to know “where’s the change?” While exceptions to this observation did exist, there was a general sense that senior managers were more involved and saw more change than others saw.

Second, the absence of formal change- management processes made important resource allocation decisions, trade-offs on tech- nology, and coordination of quality processes across the system more difficult. There was no visible mechanism, for example, to decide how quality programs should be rolled out or where to pilot electronic medical record systems. The lack of formal change-management systems (action plans, governance mechanisms, learning practices) following the DAs was related to some feelings among all stakeholder groups that change was slow in coming and overwhelming when it did come. Most people correctly viewed the strategies created by the DAs as high-level plans providing general direction. However, the process for developing action plans and imple- mentation activities was not visible to many peo- ple. Ad hoc change systems emerged based on the nature of the strategy implementation activi- ties and these helped to focus attention and resources. Interestingly, these systems all started to emerge about one year after the original DAs.

Based on these implementation data, and the data presented in Application 6.1 about how the DA needed to evolve, the researchers recom- mended (1) creating different versions of the DA to address different issues and (2) formalizing

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 219

some change processes so that the resourcing, exe- cution, and communication of change were more coordinated. However, their overall recommendation was to continue using the DA for strategy formulation and visioning, as well as tactical and implementation- oriented issues.

In terms of evaluation feedback, the analysis of the activities described in the DA reports provided some important conclusions. For example, the com- position of the DAs, or the mix of AH managers and staff, physicians, community members, and other stakeholders, affected the processes and out- comes. First, in DAs where there was a higher pro- portion of physicians, there was a narrower range of stakeholder participation and an increased likelihood that the group would deviate from the agenda. In addition, there was a weak relationship between higher percentages of community participants and all DA processes. That is, when the DA had more community participants, there was broader partici- pation in the discussions, the debates were more intense, and the DA stayed on track.

Second, the composition of the DA had differ- ential impacts on the outcomes of meeting. When the DA had a high percentage of physicians, the resulting vision was less comprehensive. On the

other hand, when the DA had a high concentration of community participants, the vision was more comprehensive. These results were reflected in the survey data as well. The percentage of commu- nity participants was positively related to percep- tions that the strategy was more innovative but less aggressive and business oriented. The concen- tration of physicians in the DA tended to have oppo- site relationships with the strategy dimensions.

When these data were fed back to the organi- zation, the researchers specifically pointed out that these findings did not suggest that it was wrong to involve physicians or that a higher percentage of community members was better. To the contrary, the fundamental assumption of DA interventions was that a broader mix of stakeholders contributes to a better solution. These data did suggest that not all stakeholder groups are created equal. Too many of any type of stakeholder group may lead to lopsided discussions and sway the agenda. In several of the DAs, for example, almost half of the participants were physicians, making it likely that this constituency would disproportionately impact the flow of the meeting.

The table below summarizes many of the find- ings from the Alegent project.

Evaluation Question Data

• Does Alegent’s strategy, purpose, and organization support change?

– Yes—Many powerful internal and external forces are pushing for and supporting change

• How effective were the original six DAs in achieving intended outcomes?

– Very effective—The DAs generated a lot of energy for change, healed physician relationships, and utilized good thinking

• What DA characteristics made a difference?

– The DA’s composition was an important influence on its processes and outcomes

• How do executives and managers char- acterize the service-line strategies?

– Comprehensive, somewhat innovative, and business- oriented

• Are they similar or different? – Managers are more positive than executives

• How is the implementation process being orchestrated?

– Informally—As a result, people feel overwhelmed by change

• What processes, structure, and roles have been put in place to make the strategies a reality?

– Few—People agree there is change capacity but want more involvement and action

• How is the implementation going? – Generally positive attributions

• Is there evidence that implementation is likely to produce desired outcomes?

– While uncertainty exists, there are many shared sug- gestions for moving forward and commitment is high

220 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

to guide current implementation and evaluate the effectiveness of an intervention. But the evaluation is not perfect. What are the strengths and weaknesses of the assessment? How could it have been improved? How much confidence do you have in the lessons learned from this organization?

9-2 Institutionalizing Organizational Changes Once it is determined that changes have been implemented and are effective, attention is directed at institutionalizing the changes—maintaining them as a normal part of the organization’s functioning for an appropriate period of time.17 In complex and uncertain environments, some changes are only part of a long journey of organization adaptation. Innovating new products is not a one-time change but a continuous process that must be implemented over and over again. Other changes, such as the process for appraising per- formance, need to persist. For example, there is little to be gained from making front-line supervisors learn a new performance rating system every year.

Lewin described change as occurring in three stages: unfreezing, moving, and refreezing. Institutionalizing an OD intervention concerns refreezing. It involves the long-term persistence of organizational changes: To the extent that changes persist, they can be said to be institutionalized. Such changes are not dependent on any one person but exist as a part of the culture of an organization. This means that numerous others share norms about the appropriateness of the changes.

How planned changes become institutionalized has not received much attention in the OD literature. Rapidly changing environments have led to admonitions from consul- tants and practitioners to “change constantly,” to “change before you have to,” and “if it’s not broke, fix it anyway.” Such a context has challenged the utility of the institution- alization concept. Why endeavor to make any change permanent given that it may require changing again soon? However, the admonitions also have resulted in institution- alization concepts being applied in new ways. Change itself has become the focus of institutionalization. Dynamic strategy making, self-design, organization learning, and built-to-change interventions described in Chapter 19 all are aimed at enhancing the

Overall, the researchers concluded that: 1. There was a demonstrable and palpable

change in a variety of organization features that if not directly tied to the DA were certainly hastened by it. A large number of specific stra- tegic, operational, and practice-oriented changes connected with each clinical area had been implemented relatively quickly. In addition, there was substantial agreement that the culture was changing, as evidenced by new language, regular and extensive use of DAs, collaborative decision making, open- ness to innovation, confidence in leadership, and openness to joint ventures with the

physicians. Finally, there was broad agreement that the DA process represented a visible and tangible effort to address physician relation- ships and clearly moved those relationships in a positive direction.

2. The organization’s initial use of the DA process as a strategic visioning intervention persists in the minds of most organization members. Alegent Health can productively apply the technology and principles to other, more implementation-oriented issues. On the other hand, DAs cannot do everything, and comple- mentary governance and implementation pro- cesses are necessary.

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 221

organization’s change capability.18 In this vein, processes of institutionalization take on increased utility. This section presents a framework for identifying factors and processes that contribute to the institutionalization of OD interventions, including the process of change itself.

9-2a Institutionalization Framework Figure 9.2 presents a framework that identifies organization and intervention character- istics and institutionalization processes affecting the degree to which change programs are institutionalized.19 The model shows that two key antecedents—organization and intervention characteristics—affect different institutionalization processes operating in organizations. These processes, in turn, affect various indicators of institutionalization. The model also shows that organization characteristics can influence intervention char- acteristics. For example, organizations having powerful unions may have trouble gaining internal support for OD interventions.

9-2b Organization Characteristics Figure 9.2 shows that the following three dimensions of an organization can affect inter- vention characteristics and institutionalization processes:

1. Congruence. This is the degree to which an intervention is perceived as being in har- mony with the organization’s managerial philosophy, strategy, and structure; its cur- rent environment; and other changes taking place.20 When an intervention is congruent with these dimensions, the probability is improved that it will be supported and sustained. Congruence can facilitate persistence by making it easier to gain mem- ber commitment to the intervention and to diffuse it to wider segments of the organi- zation. The converse also is true: Many OD interventions promote employee

FIGURE 9.2

Institutionalization Framework

© Ce

ng ag

e Le

ar ni

ng

222 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

participation and growth. When applied in highly bureaucratic organizations with for- malized structures and autocratic managerial styles, participative interventions are not perceived as congruent with the organization’s managerial philosophy.

2. Stability of environment and technology. This involves the degree to which the organization’s environment and technology are changing. The persistence of change is favored when environments are stable. Under these conditions, it makes sense to embed the change in an organization’s culture and organization design processes. On the other hand, volatile demand for the firm’s products or services can lead to reductions in personnel that may change the composition of the groups involved in the intervention or bring new members on board at a rate faster than they can be socialized effectively.

3. Unionization. Diffusion of interventions may be more difficult in unionized set- tings, especially if the changes affect union contract issues, such as salary and fringe benefits, job design, and employee flexibility. For example, a rigid union contract can make it difficult to merge several job classifications into one, as might be required to increase task variety in a job enrichment program. It is important to emphasize, however, that unions can be a powerful force for promoting change, par- ticularly when a good relationship exists between union and management.

9-2c Intervention Characteristics Figure 9.2 shows that the following five features of OD interventions can affect institu- tionalization processes:

1. Goal specificity. This involves the extent to which intervention goals are specific rather than broad. Specificity of goals helps direct socializing activities (for example, training and orienting new members) to particular behaviors required to implement the intervention. It also helps operationalize the new behaviors so that rewards can be linked clearly to them. For example, an intervention aimed only at increasing product quality is likely to be more focused and readily put into operation than a change program intended to improve quality, quantity, safety, absenteeism, and employee development.

2. Programmability. This involves the degree to which the changes can be pro- grammed or the extent to which the different intervention characteristics can be specified clearly in advance to enable socialization, commitment, and reward alloca- tion. For example, job enrichment specifies three targets of change: employee discre- tion, task variety, and feedback. The change program can be planned and designed to promote those specific features.

3. Level of change target. This concerns the extent to which the change target is the total organization, rather than a department or small work group. Each level of the organization has facilitators and inhibitors of persistence. Department and group change are susceptible to countervailing forces from others in the organization. These can reduce the diffusion of the intervention and lower its ability to impact orga- nization effectiveness. However, this does not preclude institutionalizing the change within a department that successfully insulates itself from the rest of the organization. Such insulation often manifests itself as a subculture within the organization.21

Targeting the intervention toward wider segments of the organization, on the other hand, can also help or hinder change persistence. A shared belief about the intervention’s value can be a powerful incentive to maintain the change, and

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 223

promoting a consensus across organization departments exposed to the change can facilitate institutionalization. However, targeting the larger system also can inhibit institutionalization. The intervention can become mired in political resistance because of the “not invented here” syndrome or because powerful constituencies oppose it.

4. Internal support. This refers to the degree to which there is an internal support sys- tem to guide the change process. Internal support, typically provided by an internal OD practitioner, can gain commitment for the changes and help organization mem- bers implement them. External consultants also can provide support, especially on a temporary basis during the early stages of implementation. For example, in many interventions aimed at implementing high-involvement organizations (see Chapter 13), both external and internal OD practitioners provide change support. The external consultant typically brings expertise on organizational design and trains members to implement the design. The internal consultant generally helps members relate to other organizational units, resolve conflicts, and legitimize the change activities within the organization.

5. Sponsorship. This concerns the presence of a powerful sponsor who can initiate, allocate, and legitimize resources for the intervention. Sponsors must come from levels in the organization high enough to control appropriate resources, and they must have the visibility and power to nurture the intervention and see that it remains viable. There are many examples of OD interventions that persisted for sev- eral years and then collapsed abruptly when the sponsor, usually a top administra- tor, left the organization. There also are numerous examples of middle managers withdrawing support for interventions because top management did not include them in the change program.

9-2d Institutionalization Processes The framework depicted in Figure 9.2 shows the following five institutionalization pro- cesses that can directly affect the degree to which OD interventions are institutionalized:

1. Socialization. This concerns the transmission of information about beliefs, prefer- ences, norms, and values with respect to the intervention. Because implementation of OD interventions generally involves considerable learning and experimentation, a continual process of socialization is necessary to promote persistence of the change program. Organization members must focus attention on the evolving nature of the intervention and its ongoing meaning. They must communicate this information to other employees, especially new members of the organization. Transmission of information about the intervention helps bring new members onboard and allows participants to reaffirm the beliefs, norms, and values underly- ing the intervention.22 For example, employee involvement programs often include initial transmission of information about the intervention, as well as retraining of existing participants and training of new members. Such processes are intended to promote persistence of the program as new behaviors are learned and new mem- bers introduced.

2. Commitment. This binds people to behaviors associated with the intervention. It includes initial commitment to the program, as well as recommitment over time. Opportunities for commitment should allow people to select the necessary behaviors freely, explicitly, and publicly. These conditions favor high commitment and can

224 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

promote stability of the new behaviors. Commitment should derive from several orga- nizational levels, including the employees directly involved and the middle and upper managers who can support or thwart the intervention. In many early employee involve- ment programs, for example, attention was directed at gaining workers’ commitment to such programs. Unfortunately, middle managers were often ignored and considerable management resistance to the interventions resulted.

3. Reward allocation. This involves linking rewards to the new behaviors required by an intervention. Organizational rewards can enhance the persistence of changes in at least two ways. First, a combination of intrinsic and extrinsic rewards can reinforce new behaviors. Intrinsic rewards are internal to people and derive from the oppor- tunities for challenge, development, and accomplishment found in the work. When interventions provide these opportunities, motivation to perform should persist. This behavior can be further reinforced by providing extrinsic rewards, such as money, for increased contributions. Because the value of extrinsic rewards tends to diminish over time, it may be necessary to revise the reward system to maintain high levels of desired behaviors.

Second, new behaviors will persist to the extent that rewards are perceived as equitable by employees. When new behaviors are fairly compensated, people are likely to develop preferences for those behaviors. Over time, those preferences should lead to normative and value consensus about the appropriateness of the intervention. For example, many employee involvement programs fail to persist because employees feel that their increased contributions to organizational improve- ments are unfairly rewarded. This is especially true for interventions relying exclu- sively on intrinsic rewards. People argue that an intervention that provides opportunities for intrinsic rewards also should provide greater pay or extrinsic rewards for higher levels of contribution to the organization.

4. Diffusion. This refers to the process of transferring changes from one system to another. Diffusion facilitates institutionalization by providing a wider organizational base to support the new behaviors. Many interventions fail to persist because they run counter to the values, purpose, or identity of the larger organization. Rather than support the intervention, the larger organization rejects the changes and often puts pressure on the change target to revert to old behaviors. Diffusion of a change to other organizational units reduces this counter-implementation force. It tends to lock in behaviors by providing normative consensus from other parts of the organi- zation. Moreover, the act of transmitting institutionalized behaviors to other systems reinforces commitment to the changes.

5. Sensing and calibration. This involves detecting deviations from desired interven- tion behaviors and taking corrective action. Institutionalized behaviors invariably encounter destabilizing forces, such as changes in the environment, new technolo- gies, and pressures from other departments to nullify changes. These factors cause some variation in performances, preferences, norms, and values. To detect this variation and take corrective actions, organizations must have some sensing mech- anism. Sensing mechanisms, such as implementation feedback, provide informa- tion about the occurrence of deviations. This knowledge can then initiate corrective actions to ensure that behaviors are more in line with the intervention. For example, if a high level of job discretion associated with a job enrichment intervention does not persist, information about this problem might initiate cor- rective actions, such as renewed attempts to socialize people or to gain commit- ment to the intervention.

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 225

9-2e Indicators of Institutionalization Institutionalization is not an all-or-nothing concept but reflects degrees of persistence in a change. Figure 9.2 shows five indicators of the extent of an intervention’s persistence. The extent to which the following factors are present or absent indicates the degree of institutionalization:

1. Knowledge. This involves the extent to which organization members have knowl- edge of the behaviors associated with an intervention. It is concerned with whether members know enough to perform the behaviors and to recognize the consequences of that performance. For example, job enrichment includes a number of new beha- viors, such as performing a greater variety of tasks, analyzing information about task performance, and making decisions about work methods and plans.

2. Performance. This is concerned with the degree to which intervention behaviors are actually performed. It may be measured by counting the proportion of relevant peo- ple performing the behaviors. For example, 60% of the employees in a particular work unit might be performing the job enrichment behaviors described above. Another measure of performance is the frequency with which the new behaviors are performed. In assessing frequency, it is important to account for different varia- tions of the same essential behavior, as well as highly institutionalized behaviors that need to be performed only infrequently.

3. Preferences. This involves the degree to which organization members privately accept the organizational changes. This contrasts with acceptance based primarily on organizational sanctions or group pressures. Private acceptance usually is reflected in people’s positive attitudes toward the changes and can be measured by the direction and intensity of those attitudes across the members of the work unit receiving the intervention. For example, a questionnaire assessing members’ percep- tions of a job enrichment program might show that most employees have a strong positive attitude toward making decisions, analyzing feedback, and performing a variety of tasks.

4. Normative consensus. This focuses on the extent to which people agree about the appropriateness of the organizational changes. This indicator of institutionalization reflects how fully changes have become part of the normative structure of the organi- zation. Changes persist to the degree members feel that they should support them. For example, a job enrichment program would become institutionalized to the extent that employees support it and see it as appropriate to organizational functioning.

5. Value consensus. This is concerned with social consensus on values relevant to the organizational changes. Values are beliefs about how people ought or ought not to behave. They are abstractions from more specific norms. Job enrichment, for exam- ple, is based on values promoting employee self-control and responsibility. Different behaviors associated with job enrichment, such as making decisions and performing a variety of tasks, would persist to the extent that employees widely share values of self-control and responsibility.

These five indicators can be used to assess the level of change persistence. The more the indicators are present in a situation, the higher will be the degree of institutionaliza- tion. Further, these factors seem to follow a specific development order: knowledge, perfor- mance, preferences, norms, and values. People must first understand new behaviors or changes before they can perform them effectively. Such performance generates rewards and punishments, which in time affect people’s preferences. As many individuals come to prefer the changes, normative consensus about their appropriateness develops. Finally, if

226 PART 2 THE PROCESS OF ORGANIZATION DEVELOPMENT

a p

p li

ca ti

o n

9 2 INSTITUTIONALIZING STRUCTURAL CHANGEAT HEWLETT-PACKARD

I n May 2002, the hotly contested acquisition of Compaq by Hewlett-Packard (www.hp.com) was finalized. Unlike the major organization changes before it, the acquisition challenged

the abilities of this perennial “most admired com- pany” to execute a complex structural change. The success of the integration process described in Application 8.4 is partly due to a store of institu- tionalized knowledge and capability within the HP organization. This application describes a number of large-scale structural changes at HP. The com- pany’s repeated ability to carry out such change speaks to its institutionalized capability to manage change.

Since its founding in 1939, HP has implemen- ted successfully no fewer than a dozen major orga- nizational changes, including the transition from a high-tech entrepreneurial start-up to a profes- sionally managed company; from a small instru- ments business to a leading computer company; from a company oriented around complex- instruction-set computing technology to reduced- instruction-set computing technology; from a technology/engineering-based company to a market/brand-driven company; and, from a “pure products” company to a services company.

HP’s electronics and computer business was characterized by highly volatile technological and market change. It had to quickly adopt, inno- vate, and implement a variety of technological and organizational changes just to survive. HP’s traditional and current strategies were built on innovation, differentiation, and high quality. Another important feature of HP, and one of its more enduring characteristics, is the “HP Way”—a cultural artifact that supports a partici- pative management style and emphasizes com- monness of purpose and teamwork on one hand and individual freedom and initiative on the other. Over time, however, the HP Way has been both a constraint to and a facilitator of change.

For example, the HP Way has been at the root of the company’s difficulties in institutional- izing structural and behavioral changes to bring about more cooperation among the computer divisions. The initial structural change occurred

in 1982 when HP transformed itself from a pro- ducer of high-quality electronic measuring instruments into a computer company. At the time, computers and computer-related equip- ment accounted for only about one-third of rev- enues and HP was structured into more than 50 highly autonomous and decentralized product divisions focused on specialized niche markets. Individual engineers came up with innovative ideas and “bootstrapped” new products any way they could. Organization members were encouraged to work with other engineers in other departments within the same division, but there was little incentive to coordinate the development of technologies across divisions. This focus on the individual was supported by a performance management system that mea- sured and rewarded “sustained contributions;” the key to success for an individual was work- ing with many people in the division. HP pros- pered by maximizing each of its parts.

Former CEO John Young’s decision to focus on computers fundamentally shifted the keys to success. Computer production required a coordinated effort among the different compo- nent divisions and market shares large enough to encourage software vendors to write pro- grams for their machines. In a culture that sup- ported individual contributions over divisional cooperation, Young placed all the instruments divisions into one group and all the computer divisions into another group, a basic design that persisted until the spin-off of the Agilent instru- ments business in 1999. In addition, he central- ized research, marketing, and manufacturing, which had previously been assigned to the divi- sions. Problems quickly arose. In one case, the company’s new and highly touted graphics printer would not work with its HP3000 mini- computer. The operating software, made by a third HP division, would not allow the two pieces of hardware to interface.

In response, the computer group formed committees to figure out what new technolo- gies to pursue, which to ignore, which of HP’s products should be saved, and which would be

CHAPTER 9 EVALUATING AND INSTITUTIONALIZING ORGANIZATION DEVELOPMENT INTERVENTIONS 227