3 POLICY-CAPTURING STUDIES
4 CORPORATE GOVERNANCE AND GOVERNANCE DECISIONS
5 STEPS OF POLICY-CAPTURING
5.1 STEP 1: STUDY DESIGN
5.2 STEP 2: STUDY EXECUTION
5.3 STEP 3: ANALYSIS
5.4 STEP 4: INTERPRETING RESULTS
5.5 STEP 5: REPORTING RESULTS
6.1 FINDING ARTICLES / DATA COLLECTION
6.2 ARTICLES FROM GOVERNANCE RESEARCH
6.2.1 Reasons for PC
6.2.2 Noteworthy Characteristics
6.3 ARTICLES FROM OTHER AREAS OF RESEARCH
6.3.1 Reasons for PC
6.3.2 Noteworthy Characteristics
7 OBSTACLES AND GUIDELINES
7.1 STEP 1: STUDY DESIGN
7.2 STEP 2: STUDY EXECUTION
7.3 STEP 3: ANALYSIS
7.4 STEP 4: INTERPRETING RESULTS
7.5 STEP 5: REPORTING RESULTS
8 CONJOINT ANALYSIS
8.1 STEPS OF CONJOINT ANALYSIS
8.2 QUESTIONS IN CONJOINT ANALYSIS
figure 1: “Options for Study Design” 21
table 1: „Review of Studies from Governance Research“ A
table 2: „Review of Articles from Other Areas of Research“ E
illustration not visible in this excerpt
A great number of scientists from various areas of research (e.g. psychology or human resource management) seek to achieve a deeper understanding of managerial decision-making processes. For this purpose, several methods has been developed in order to assess the underlying determinants, their implementation, and their impact on managers’ decisions, e.g. a traditional survey (which collects quantitative information focused on either opinions or factual information, depending on its purpose) or self- assessment (which means individuals assessing their own work and which can help them to criticize their own work, and form evaluations about its strengths and weaknesses).
A considerably adjuvant approach, belonging to the class of experimental studies, is the Policy-Capturing (PC) approach. It has been largely applied in different branches of research, such as human resource management and psychology, ranging, for instance, from compensation decisions or job evaluation to expatriate selection.
Recently, there are also some scholars in the field of corporate governance research who made use of the advantages PC has to offer, like e.g. the studies by Kale and Puranam (2005) or Sutcliffe and Zaheer (1998) which I will refer to in greater detail in the course of this thesis (see chapter 6 in particular). They deal with topics like acquisitions, integration or collaboration which clearly can be attributed to the realm of governance research (see also chapter 4).
This thesis aims at trying to apply PC to potential new areas, as it is a method used to gain an insight into individuals’ decision-making processes, and discusses its methodological advantages as well as typical obstacles researchers have to cope with.
Its purpose is to broaden the application of PC in governance research and provide guidelines for inexperienced researchers in order to improve the study’s design, execution and analysis thereby also enhancing the results’ validity and generalizability.
In the following, I will give a short outline of the content of the thesis at hand. As my topic deals with the application of Policy-Capturing (PC) studies in governance research, I will first address both terms - the PC technique and a definition of governance decisions - separately (see chapters 3 and 4) and then combine the explanations into a superordinate context.
Chapter 5 attends to the typical steps a PC study consists of and describes representative questions and decisions researchers have to consider at every stage of their research, namely study design (see chapter 5.1), study execution (see chapter 5.2), analysis (see chapter 5.3), interpreting results (see chapter 5.4) and reporting results (see chapter 5.5). Subsequently, I will describe my process of finding suitable articles (see chapter 6.1) and give a brief survey of the PC studies dealing with governance decisions (see chapter 6.2) as well as those from other areas of research (see chapter 6.3) which I reviewed and analyzed.
The following section (chapter 7) covers obstacles that researchers conducting a PC study may frequently be confronted with, offers solutions on how to overcome those hurdles in case of occurrence and provides the reader with guidelines that should be adhered to in order to enhance the study’s design and execution as well as the results’ analysis, interpretation and reporting (according to the five steps dealt with in chapter 5). Having covered PC in extenso, the next chapter (chapter 8) discusses conjoint analysis as a closely related technique highlighting conspicuous parallels to PC (see chapter 8.1 in particular) and giving some advice scholars employing PC could also benefit from.
I end my thesis with a conclusion (chapter 9), summing up the most important and most interesting findings and giving a brief outlook on the future of the application of PC studies in governance research.
3 Policy-Capturing Studies
As already outlined previously, I will begin with a compendious introduction to PC studies, explaining its purpose, its fundamentals and its procedure. PC is a field-experimental technique widely used in decision-making research in order to understand and improve individuals’ decision processes.1 It is an indirect method of attitude measurement, as opposed to e.g. traditional surveys or questionnaires, which allows for an assessment of the participants’ information-processing strategies.2
PC experimentally manipulates the level of a number of theoretically determined criteria (so-called cues; the independent variables) which are then combined into various scenarios the decision-markers are asked to review and evaluate according to their personal judgment3. The participants rate combinations rather than directly revealing their preferences which would demand far more self-insight.
Thus, PC allows for analyzing both the average behavior of groups of people making independent decisions (nomothetic4 ), like in traditional between-subjects designs, and the decision-making behavior of individuals (idiographic5 ).
The researchers’ aim is to capture individual decision policies and, eventually, arrive at a linear model of the judgment process6, such as a regression equation, i.e. a mathematical description of the model that was used which shows how decision makers weight and combine different pieces of information7.
Thereby, PC identifies relevant factors and provides instructive information about an individual’s information processing and judgment structure. The regression weights (so-called beta weights) can be interpreted as showing the cues’ respective relative importance if the cues are orthogonal.8 9
As the respondents are never asked directly, PC enables the researcher to determine hidden attitudes and mitigate social desirability effects when inferring the policy through statistical analysis.10 The results obtained can be of considerable help for researchers and respondents to predict and improve future decision-making11 and lead to a common understanding of problems.
Since individuals are generally unable to accurately describe their own decisionmaking methodology and, consequently, intended and captured policies often differ considerably, the best objective means of obtaining an individual’s model is to construct it by observing and recording the person’s behavior over a range of specific, controlled situations. This captures the policy of the individual decision-maker; the results show what is most important from their point of view.12
PC describes how people utilize external information in arriving at overall judgments. It is a relatively straightforward technique that has been utilized in decision- making studies for many years (see also Appendix II “Review of Studies from Other Areas of Research”), e.g. with regard to job search (Aiman-Smith, Bauer, Cable, 2001; Cable, Judge, 1994), compensation (Beatty, McCune, Beatty, 1988), performance appraisal (Hobson, Gibson, 1983), general personnel decisions (Dineen, Noe, Wang, 2004; Donnelly, Bownas, 1984; Ones, Viswesvaran, 1999) and differences in decision- making across cultures (Beatty, McCune, Beatty, 1988; Zhou, Martocchio, 2001). It enables researchers to decompose problems into separate parts so that the decision maker’s judgment can be focused on each relatively simple part.13
The technique is characterized by its high internal validity14 and therefore adds to traditional survey studies which generally possess high external validity.15 Nevertheless, PC obtains the advantages of statistical inference and greater objectivity at the expense of richness.16
PC is suitable for research aimed at describing and explaining decision-making processes or rules, whereas non-linear models are to apply to prediction-oriented research.
On the one hand, the researchers’ work when designing the scenarios to be evaluated is simplified due to the use of hypothetical cue levels. On the other hand, however, this incident may cause doubtful generalizability of the results in the end.17
4 Corporate Governance and Governance Decisions
Having explained the first part of this thesis addressing the application of PC studies in governance research, namely the PC methodology, I will now focus on the second part of the topic and give a compendium of corporate governance and governance decisions.
“Governance” refers to the processes and systems by which an organization operates. The term stems from the Latin word for “steering” and describes the way in which executives govern a corporation.
“Corporate governance” integrates all processes and institutions related to directing, administering and controlling an organization. It contains rules and procedures for organizational decision-making in order to optimize results. This way, the corporation should act on behalf of all its stakeholders18 who contribute to the company’s success in manifold ways, as e.g. by providing expertise, labor or capital, and who expect the company to let them share its revenues in one way or the other.
Governance decisions include a wide range of questions the supervisory board of a corporation has to address. The variety of possible choices regarding the form of governance may best be described with the aid of a continuum of governance structures ranging from market exchange, subsuming e.g. subcontracting or licensing, on the one hand and hierarchical governance, embracing wholly-owned subsidiaries, for instance, on the other hand. Any governance structure in between these two extremes is considered as a hybrid form, an umbrella term for a couple of forms like joint ventures or franchising.
A prominent motivation for those kinds of governance decisions is the organization’s perception that the whole is more than the sum of its parts, i.e. that they can make use of possible synergy and system effects. Therefore, an organization will exist so long as it can offer its members incentives which surpass the contributions it asks of them19.
A famous question of governance choice is finding the appropriate form of organization. There are various so-called business entities possible with inherent advantages and disadvantages a company-owner must take into concern. They must include aspects such as liability20 or taxation regulations in their considerations for the best-suited form of organization, with a sole proprietorship21, a partnership22 or a corporation23 commonly used.
Another problem today’s managers may be confronted with is the decision whether to form an alliance with another firm or not and, if yes, how to find a suitable partner. An alliance is an “association to further the common interests of the members”.24
Mergers and acquisitions (M&A) is an equally popular aspect in this context dealing with a corporate finance strategy that combines two or more companies. Merger is a legal process which takes place between equal companies, whereas acquisitions usually involve a large firm buying a small one, thus the participating firms are “un- equals”. British “Vodafone” buying the German communications company “Mannesmann” is a famous example of an acquisition which took place in February 2000 (also called hostile take-over). M&A can have miscellaneous motives, as there are e.g. economies of scale25, synergy effects26 or the possibility of cross selling27 to be mentioned.
Supervisory board members may also face the decision whether or not to follow a diversification strategy, which is a risk-management technique that mixes a wide variety of investments within a portfolio.
Related to the trend of lean production, which means the concentration on an organization’s core activities and, thus, de-diversification, is the question of whether to make or buy a product, as choosing between in-house production and supply by third parties. Companies increasingly focus on business units with potential for a competitive advantage and try to bring down the proportion of in-plant production in order to lower their vertical range of manufacture. By buying vendor parts from specialized suppliers, firms can remain competitive in today’s increasingly tough times, bearing in mind that on the other hand a company may become dependent on a supplier and cannot influence the development of single spare parts.
Outsourcing constitutes a further example of governance decisions. It signifies the delegation of non-core business units to external specialized subcontractors. A third party is responsible for the management of a whole business unit. This requires trust and a reliable flow of information in both directions in order to guarantee a successful business partnership. Reasons for outsourcing are similar to those of the make-or-buy decision mentioned above, as it is made to lower costs or to focus on core competencies.
Taking all these circumstances into account, it can be concluded that organizations will be confronted with governance decisions under very different circumstances and in various stages of their development, as there is not only the choice of the right form of organization in the beginning but also possible diversification or dediversification measures in the company’s further development.
Hence, it is important to take a closer look at governance research, as every organization was, is and will be concerned.
5 Steps of Policy-Capturing
After having explained and illustrated the two basic components this thesis intends to combine and unite and, thus, having provided the basics and the preconditions needed to comprehend the following considerations, I will now continue with introducing a scheme of the typical five steps a PC study should follow.
Aiman-Smith, Scullen and Barr (2002) define these steps (accompanied by several useful guidelines I will refer to later in this thesis; see chapter 7).
5.1 Step 1: Study Design
The first step of a PC study must address three basic thoughts.
(a) The nature of the research must be defined, which means that it must be decided whether the objective is to concentrate on specific cases and unique traits or functioning of individuals (a so-called “idiographic” approach which emphasizes individuals), or to discover general and universal laws (a so-called “nomothetic” approach which aggregates across many individuals).
(b) Furthermore it is important in this early stage of a PC study to consider its representativeness. Researchers must develop realistic scenarios and find suitable members of the target group in order to obtain an appropriate sample of respondents.
(c) The third consideration refers to the experimental design that includes questions about the number of cues, scenarios and the related issue of potential start-up effects and respondent fatigue or boredom due to the scenarios’ length and complexity.
5.2 Step 2: Study Execution
Step 2 covers the instruments and procedures of data collection with recommendations like, e.g. providing all relevant cues and making sure that the respondents understand the task.
Furthermore, the scenarios have to be presented in an understandable manner and under appropriate testing conditions.
5.3 Step 3: Analysis
This section attends to coding independent variables and selecting the appropriate regression model.
(a) The former issue is usually addressed by choosing dummy variables28, allocated coding29 or effect coding30.
(b) The latter issue depends on the fact whether the dependent variable is categorical or numeric, which refers to the variable’s values identifying class or group membership ( logistic regression may be most adequate), or rather describing a measured value ( least square regression techniques may best be used).
5.4 Step 4: Interpreting Results
The fourth step of a PC study is concerned with topics such as, for example, multicollinearity and standardization. Standardized weights are measured in units of their standard deviation (SD or σ ), following the pattern of x − µ σ 31. The application of this procedure is recommended in order to permit the regression coefficients’ comparability.
5.5 Step 5: Reporting Results
The final step gives attention to including enough detail and accurately describing the whole procedure. Extensive reporting should allow the interested reader to replicate or amplify the current research if intended; therefore it is the researcher’s duty to provide the necessary information about the process of scenario construction, the variables incorporated and their values, the judgment scale employed, as well as the criteria for and the process of selecting appropriate respondents.
6.1 Finding Articles / Data Collection
Using „EBSCO Host Research Databases“ I searched for suitable articles with the help of the keywords “policy capturing” and “policy capture”. The studies obtained were predominantly dealing with topics from human resource management, behavior in organizations and personnel psychology, since the PC approach has been widely applied in these areas of research. The most prevalent journals were “Personnel Psychology”, “Strategic Management Journal”, “Academy of Management Journal” and “Organizational Research Methods” (for more details see also Appendix II: “Review of Studies from Other Areas of Research”).
Nevertheless, I also found a number of studies attending to the realm of governance research (see Appendix I: “Review of Studies from Governance Research”). In order to further broaden my collection, I reviewed their references and was able to obtain more articles from the desired area.
It was not easy to find governance studies applying the PC approach which may be due to the fact that this method is not yet very well-established and currently lacks scientific attention and, hence, popularity.
With this thesis I want to allude to this situation by reviewing suitable articles and identifying characteristic obstacles of the PC technique as well as recommending guidelines on how to deal with them from perusing the PC studies from other scholars mentioned above.
6.2 Articles from Governance Research
I analyzed a total of seven studies explicitly dedicated to governance research (see Appendix I: “Review of Studies from Governance Research” for a more detailed listing) regarding their application of the five steps of a PC study (see chapter 5). They are concerned with acquisitions32, levels of equity ownership33, various levels of integration34 (with special reference to vertical integration35 ) and technological collaboration36.
6.2.1 Reasons for PC
I reviewed the articles which apply the PC technique in order to find out the researchers’ motivations to do so.
Hitt and Tyler (1991) point out that PC has been used in the past to examine managerial decision-making and refer to Argyris and Schon who addressed the discrepancy between “theories in use” and “espoused theories of action” (see also chapter 3) in “Theory in practice: increasing professional effectiveness”37. Furthermore, they rely on some previous studies that found support for the procedure’s external validity.
In an analogous manner, Kale and Puranam (2005) support their application of PC through a listing of previously conducted studies, mentioning predominantly the ones I will also deal with in this chapter. They also emphasize the possibility of detecting differences between the cues which participants say they incorporate and the ones they actually do include in their decision process.
In his study focusing on the integration level of acquisitions, Pablo (1994) justifies the application of PC by underscoring the opportunity to infer the respondents’ tacit policies from the statistical models.
Dealing with “Uncertainty in the transaction environment”, Sutcliffe and Zaheer (1998) give the necessity of controlling the information-processing context as a reason for making use of the PC technique and allude to previous research showing the profitableness and the benefits of an experimental decision-based point of view.
Tyler and Steensma (1995; 1998) conducted two PC studies covering governance decisions and refer to the method’s ability to get insight into the participants’ hidden decision policies which they often do not understand exactly or are unable to determine. Thereby any unwanted effects which may falsify the results in the
1 cf. Hobson, Gibson, 1983; Kale, Puranam, 2005
2 cf. Dineen, Noe, Wang, 2004; Madden, 1981
3 cf. Aiman-Smith, Scullen, Barr, 2002; Kale, Puranam, 2005; Karren, Barringer, 2002
4 relating to discovery of general and universal laws
5 emphasizing individuals: concentrating on specific cases and the unique traits or functioning of individuals, rather than on broad generalizations about human behavior
6 cf. Klaas, Wheeler, 1990; Waller, Novack, 1995
7 cf. Donnelly, Bownas, 1984
8 cf. Miller, Wiseman, 2001; Ones, Viswesvaran, 1999
9 Orthogonal cues hold the statistical advantage of independent estimates of importance as well as the psychological convenience that all possible cue combinations are to be considered.
10 cf. Ones, Viswesvaran, 1999; Stumpf, London, 1981
11 cf. Taylor, Wilsted, 1974
12 cf. Aiman-Smith, Bauer, Cable, 2001
13 cf. Wright, Goodwin, 1999
14 Internal validity refers to the extent to which it can be accurately stated that the independent variable produced the observed effect.
15 External validity relates to generalizing the findings to or across target populations, tasks and environments.
16 cf. Kale, Puranam, 2005
17 cf. Zhou, Martocchio, 2001
18 all persons who represent a group’s interest and/or are eligible for benefits from an organization, such as proprietors, creditors, employees, managers, suppliers, shareholders, customers, and also the community
19 cf. Ouchi, 1980
20 Liability describes the owner’s pecuniary obligation, i.e. the duty to pay for the company’s debts, also with unrelated assets such as personal residence, if necessary.
21 A sole proprietorship offers the owner complete control and the convenience of no double taxation at the expense of personal liability.
22 Partnerships sum up general and limited partnerships.
23 The corporation, one of the most widely used business entity forms, provides limited liability to its shareholders who own the corporation, as the corporation is a separate legal entity and their private assets cannot be reached by corporate creditors.
25 If you increase the quantity of all input factors and the costs increase by a lesser amount than they would do proportionately, there are positive economies of scales.
26 Synergy effects refer to the financial benefit a company expects when merging with or acquiring another one, e.g. by widening the customer base or reducing costs when eliminating duplicate jobs.
27 Cross selling describes the process of the combined companies A and B now also selling to their partner’s customers.
28 Dummy variables can take the values 0 or 1, depending on whether a feature is present or not. For example gender is often coded “male = 0” and “female = 1”, thus practically measuring the presence of femaleness.
29 Allocated coding of variables can represent, for instance, low, medium, and high levels through 1, 2, and 3.
30 Effect coding assigns the categories to the values -1, 0, and 1.
31 with x representing the respective value and µ representing the means of all values obtained
32 cf. Citera, Stuhlmacher, 2001; Hitt, Tyler, 1991
33 cf. Kale, Puranam, 2005
34 cf. Pablo, 1994
35 cf. Sutcliffe, Zaheer, 1998
36 cf. Tyler, Steensma, 1995; Tyler, Steensma, 1998
37 Jossey-Bass, San Francisco, CA, 1974
- Quote paper
- Martina Lütkewitte (Author), 2006, Policy-Capturing - State of the art and implications for governance research, Munich, GRIN Verlag, https://www.grin.com/document/86282