Research Proposal
The effect of control tightness on the organisational reputation of
pro-fessional service firms
Name: Lorenzo Hulzebos Student number: 11402687
Thesis supervisor: Dr. ir. S. P. van Triest Date: June 25th, 2018
Word count: 14.164
MSc Accountancy & Control, specialization Control
Statement of Originality
This document is written by student Lorenzo Hulzebos who declares to take full responsibil-ity for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of comple-tion of the work, not for the contents.
Abstract
This paper investigates whether control tightness has a direct effect on the organisational rep-utation of professional service firms (PSFs). Both tight and loose controls might be able to improve a firm’s reputation depending on the type of firm and the circumstances it finds itself in. Using survey data, this research measures the effect of both formal and informal controls on reputation, as well as research whether primary service professionals and staff profession-als have a significantly different effect in this regard. The results from regression analysis find support only for cultural control tightness’ effect on reputation. There is no evidence support-ing the claim that the other forms of control tightness have direct effect on reputation, nor is a significant difference between primary service professionals and staff professionals supported.
Keywords: Professional service firms; Reputation; Management Control System; Control tightness
Acknowledgements
This paper could not have existed in its current form were it not for the help of others. As such, I would like to thank Helena Kloosterman for setting up this survey project in the first place, Bianca Groen for acting as my thesis supervisor at the start of the process, and Sander van Triest for helping me get this thesis to a finished stage after that.
I would also like all the people who helped me gather enough responses for the survey project, completing the questionnaires without needing to be tricked or incessantly bullied. Without their help in this I would not have been able to participate in the survey project, and are thus deserving of my thanks.
Additional thanks go out to my family for letting me use this thesis as an excuse to get out of social obligations, as well as my friends, who helped to keep me sane while bashing my head against the wall by constantly supplying me with cute animal pictures.
Contents
Abstract ... 3
Acknowledgements ... 4
1. Introduction ... 6
2. Literature review ... 8
2.1 Professional Service Firm ... 8
2.2 Organisational reputation ... 8
2.3 Management Control System ... 10
2.4 Control tightness ... 12 3. Hypothesis development ... 13 4. Research design ... 16 4.1 Survey design ... 16 4.2 Respondents ... 18 4.3 Constructs ... 19 4.4 Validity analysis ... 21 5. Results ... 26 5.1 Descriptive statistics ... 26 5.2 Regression analysis ... 29 6. Discussion ... 33 6.1 Conclusions ... 33 6.2 Limitations ... 35 6.3 Future research ... 35 References ... 36 Appendices ... 41
1. Introduction
This paper’s aim is to examine the relationship between control tightness and organisational reputation in professional service firms (PSFs). As of this writing, there has not been an ex-traordinary amount of research done regarding PSFs, especially when compared to the re-search done regarding the manufacturing sector (e.g. Auzair & Langfield-Smith, 2005; Emp-son, Muzio, Broschak, & Hinings, 2015; Von Nordenflycht, 2010). Service firms have “emerged as one of the most rapidly growing, profitable, and significant sectors of the global economy” (Empson et al., p.1), making more research into this relatively underserved field quite valuable. It is important to note, however, that there is a difference between service firms and professional service firms. While manufacturing firms are rather easy to identify as they create tangible products and service firms don’t create tangible products, the distinction between what is a service and what is a professional service is less clear-cut.
The characteristics that make pure service firms unique are their “intangibility of ser-vices”, “inseparability of production from consumption”, where customers are involved in the creation of the service, “perishability of services”, where services not used or acted upon are lost, and “heterogeneity”, meaning the service is highly variable and will differ depending on the person or time (Auzair & Langfield-Smith, 2005, p. 400). PSFs are more ambiguous, often being defined through occupations researchers regularly state to be professional services, such as accounting firms and corporate banking. Auzair & Langfield-Smith define PSFs as “organ-isations with relatively few transactions, highly customised, process-oriented, with relatively long contact time, with most value added in the front-office and where considerable judge-ment is applied in meeting customer needs” (p. 402), while Von Nordenflycht (2010) con-cludes that PSFs have three sources of distinctiveness: knowledge intensity, low capital inten-sity, and a professionalised workforce.
Professional service firms thus deal in intangible services, making the work done dif-ficult to measure in both input and quality of output (King & Clarkson, 2015). This difdif-ficulty makes a professional service firm’s reputation “perhaps one of its most important strategic resources” (Flanagan & O’Shaughnessy, 2005, p. 445). I expect control tightness, the “degree of certainty that employees will act as the organisation wishes”, (Merchant & Van der Stede, 2007, p. 118) to have some sort of effect on the organisational reputation of PSFs, since the manner in which a PSF’s employees interact with customers and the quality they deliver go a long way towards building the firm’s reputation. In this it is assumed that employers want to specify all possible contingencies which could arise between the firm and the customer, just
like they would want to specify them in the employer-employee relationship (Weigelt & Camerer, 1988). As such, the question will be answered:
“What is the effect of control tightness on the organisational reputation of professional ser-vice firms?”
Tight controls are a way for executives and managers to try and influence the results of the organisations through the actions of its employees (Van der Stede, 2001). This sounds fairly logical; managers will likely impose tight controls upon their employees to act upon the firm’s strategy and increase its performance and reputation. For professional services, howev-er, this may be counterproductive. Professionals are presumed to prefer autonomy over bu-reaucracy (King & Clarkson, 2015; Von Nordenflycht, 2010), which is in line with Adler & Borys’ typology of organisations (1996). In it, they assert that routine, common jobs benefit from a high level of formalisation (tight controls), while the same high level of formalisation leads to negative employee attitudes when the tasks are non-routine or unique. In the latter case, tight controls would be trying to rigorously regulate tasks that will change on a case-by-case basis, which would only lead to worse results and a worse reputation instead. As such, though it seems likely that tight controls will lead to better performance and a better reputa-tion, it’s very much possible that PSFs deviate from this and benefit more from loose controls. Additionally, I also examine whether there is a significant difference in the effect of control tightness in the sort of work a professional does. This splits the responding professionals in two groups: those who perform the primary function of the organisation and those who per-form their work in a staff capacity. The expected difference in effect of control tightness on reputation originates from the amount of direct client contact these professionals have; prima-ry service professionals have more client contact than staff professionals, thus leaving a big-ger impression on clients. When recalling the organisation and service provided, primary ser-vice professionals presumably have a greater impact on reputation than staff professionals. These topics serve as the impetus for this research.
The rest of this paper is structured as follows. Prior literature on professional service firms, control tightness, and reputation is revised in the next section as a precursor to the theo-ry-building and hypothesis development of the third section. Section four describes the re-search methods. The results are examined in the fifth section, while the final section discusses the findings, the limitations, and several ideas for further research.
2. Literature review
Three terms follow from the research question, these being control tightness, organisational reputation, and professional service firms. To be able to provide a clear answer to the question asked, the first order of business is to gain insight into its separate parts. As such, these terms will be discussed before developing any hypotheses.
2.1 Professional Service Firm
Professional service firms are firms that provide a professional service, i.e. a service that is both highly customisable and has a strong client interaction component to it (Maister, 1993). Other important characteristics of PSFs are its knowledge intensity, low capital intensity, and professionalised workforce (Løwendahl, 2005; Von Nordenflycht, 2010). Knowledge intensi-ty isn’t exclusive to PSFs, however, which Løwendahl illustrates by showing that non-professional service firms and production firms can both require high knowledge intensity as well. What differentiates professional services firms from these other organisation types is that they are also labour- and capital-intensive.
Professional services are still services and share several aspects with non-professional services. These unique characteristics of services are their intangibility, perishability, and in-separability of production and consumption (Auzair & Langfield-Smith, 2005). If the service isn’t used, it can’t be stored for later use by that specific or a different consumer; instead, it becomes worthless and perishes. Furthermore, a service provided to a client can’t be ex-changed for a similar service, as they are highly variable. This also makes real standardisation of services impossible. Empson (2001) describes this by saying that “the primary activity of a PSF is the application of specialist technical knowledge to the creation of customised solu-tions to clients’ problems” (p. 842). A full list of the professions deemed to be professional services is included in the research design.
2.2 Organisational reputation
The most-frequently referenced definition for reputation puts it as “a collective representation of a firm’s past actions and results that describe the firm’s ability to deliver valuable outcomes to multiple stakeholders” (Fombrun, 1996). However, there is no consensus for a single clear definition of the term (Wæraas & Maor, 2014). This is because there are several terms and concepts that are very similar, but not necessarily the same thing as reputation. The term
‘rep-utation’ is commonly conflated with related terms like image, legitimacy, and status (Deephouse & Suchman, 2008; Rindova, Pollock, & Hayward, 2006; Washington & Zajac, 2005; Whetten & Mackey, 2002). Important differences exist between these terms, though they are subtle.
In their research on the evaluation of status, Washington & Zajac (2005) find that the definition for organisational reputation has been used to define organisational status. They argue that the two terms differ, however, as status is “fundamentally a sociological concept that captures differences in social rank that generate privilege or discrimination (not perfor-mance-based awards), while reputation is fundamentally an economic concept that captures differences in perceived or actual quality that generate earned, performance-based rewards” (Washington & Zajac, 2005, p. 283). Meanwhile, organisational legitimacy refers to a “gener-alised perception that or assumption that the actions of an entity are desirable, proper, or ap-propriate within some socially constructed system of norms, values, beliefs, and definitions” (Suchman, 1995, p. 574). Suchman’s definition finds some overlap with Fombrun’s definition for reputation, both referring to an organisation’s actions, though legitimacy’s definition doesn’t require these actions to deliver valuable future outcomes.
Figure 1: The presentation and perception of organisational identity (Whetten & Mackey, 2002, p. 401)
The concept of organisational identity is what differentiates reputation and image (Whetten & Mackey, 2002). An organisation’s identity is “that which is most central, endur-ing, and distinctive about an organisation” (p. 394), the total sum of what the organisation is. From the organisation’s self, what the organisation shows of its identity to the outside world is the organisation’s image. Reputation, then, is how the public perceives the organisation and
the feedback the organisation receives from them concerning its identity claims. This is illus-trated in figure 1. Having thus established the differences between reputation and several sim-ilar concepts by defining these concepts, it would be prudent to define reputation as well.
Fombrun (1996) calls reputation the collective representation of a firm, containing both its past actions and its possible future results, but in practice there is more to it than that. One characteristic that makes it even more different from similar concepts is that reputation is “fundamentally rival” (Deephouse & Suchman, 2008, p. 62). Reputation is used as a measure for ranking an organisation, quantifying their reputation by making it comparable to other organisations. As such, Deephouse & Suchman also argue that reputation is differentiating, encouraging organisations to distinguish themselves from its competitors by trying to appear unique or doing something wholly different. But even that isn’t necessarily enough to obtain a good reputation.
In their study of organisational reputation in the public sector, Wæraas and Maor (2014) open with their claim that when reputation is used to describe government organisa-tions, the most common connotation is ‘bad’. The evidence they find in their research claims the public sector is very much aware of the value of reputation and is acting on achieving a favourable reputation. Having a good reputation is good for business, whether that business is in the public or private sector: “A good reputation can lead to numerous strategic benefits, such as lower firm costs, enabling firms to charge premium prices, attracting applicants, in-vestors, and customers, increasing profitability, and creating competitive barriers” (Walker, 2010, p. 357). The question for any kind of firm or organisation is thus how to achieve a good reputation.
2.3 Management Control System
Management control, according to Merchant & Van der Stede (2017), includes all the devices or systems that managers use to ensure the behaviours and decisions of their employees are consistent with the organisation’s objectives and strategies. The systems mentioned are thus referred to as management control systems (MCS). The definition of MCSs has come to em-brace a much broader scope in recent years (Chenhall, 2003), from procedures and systems used to maintain or alter patterns in organisational activity (Simons, 1987) to including “eve-rything managers do to help ensure that their organisation’s strategies and plans are carried out” (Merchant & Van der Stede, 2017, p. xiii).
Companies using a management control system probably expect some kind of control problem to occur in the company’s lifetime. It is then the task of management and the MCS to deal with these control problems using control mechanisms. These control mechanisms are often called control responses, as per the survey used in this research, or management controls (Merchant & Van der Stede, 2017). Merchant and Van der Stede’s management control framework builds on a number of seminal works and theories, including Ouchi’s control mechanisms (1979), Simons’ Levers of Control (1994), and Merchant’s objects of control (1982). The last of these is where the foundation of the control response terms used in this research are first found, with Merchant presenting a framework with three objects of control, these being Control of Specific Actions, Control of Results, and Control of Personnel. Mer-chant & Van der Stede built on this in creating a framework containing action controls, results controls, personnel controls, and cultural controls.
Action controls are defined as controls that “ensure employees perform (or do not per-form) certain actions that are known to be beneficial (or harmful) to the organisation” (Mer-chant, 1982, p. 45; Merchant & Van der Stede, 2017, p. 86). This includes such things as the use of standardised processes and procedures or dictating the behaviour of employees to im-plement certain strategies (Anthony, Dearden, & Govindarajan, 1992). For this research, the term behavioural controls will be used in favour of action controls. When using results con-trols, organisations don’t dictate what actions their employees should take; instead, actions are influenced because employees are judged based on their results, such as staying inside a pre-determined budget (Merchant, 1982). These two types of controls, featuring written rules, standard procedures, and budgeting systems, are also called the formal controls (Anthony, Dearden, & Bedford, 1989; Langfield-Smith, 1997). Due to their nature they are very visible and easier to test for, which means empirical research has focused on these formal controls (Langfield-Smith, p. 208). In addition to these written rules, there are also the usually unwrit-ten policies, also known as the informal controls.
Personnel and cultural controls, sometimes taken together under the header of social controls, both involve employee monitoring. In the case of personnel controls, this tends to revolves around self-monitoring, driving employees to behave according to the organisational objectives. This kind of goal congruence between employee and organisation is facilitated through the use of mechanisms in an employee’s selection, placement, and training (Merchant & Van der Stede, 2017). Cultural controls, much like Ouchi’s clan controls (1979), rely on common agreement and shared values and beliefs. It requires a large amount of commitment
to the socially prescribed behaviours within the organisation, encouraging mutual monitoring as a form of group and peer pressure (Merchant & Van der Stede, 2017). Within organisations, these four forms of control responses can be present to a greater or lesser extent, which is what is referred to as the degree of control tightness.
2.4 Control tightness
Control tightness is defined as the degree of certainty that employees will act congruent with the organisation’s strategy (Merchant, 1998; Van der Stede, 2001). Using many control mech-anisms results in greater tightness, which Merchant says results in a higher degree of certainty that employees will act as the organisation wishes. This follows as a solution for the agency problem first mentioned as such by Jensen & Meckling (1976). Agency problems arise when “the desires or goals of the principal and agent conflict and it is difficult or expensive for the principal to verify what the agent is actually doing” (Eisenhardt, 1989, p. 58). As such, com-panies use control systems and tight controls specifically to solve the agency problem, though the way in which an organisation will do this can differ because of the four types of control. Generally speaking, a control system is tight when “participation in setting objectives in low, targets are imposed and managers must consider them firm commitments”, whereas a control system is loose when “managers are conscious of the social and individual side of control, participation is high, targets are negotiated” (Amigoni, 1978, p. 285). Other ways to describe tight and loose controls have been to differentiate between the prevalence and frequency of monitoring (Anthony, Dearden, & Govindarajan, 1992; Campbell, Epstein, & Martinez-Jerez, 2011; Hunton, Maulding, & Wheeler, 2008) or the extent to which individual group and or-ganisational unit behaviour is prescribed and controlled (Whitley, 1999). In this context, tight controls are akin to the coercive formalisation described by Adler & Borys (1996). Mean-while, loose control systems are much like enabling formalisation, both putting stock in nego-tiated targets and high participation. By loosening control, giving employees more freedom to perform their duties however they please and giving them both the time and the opportunity to think about how they would change their way of working, organisations create an environ-ment in which employees will ask to be involved in e.g. the target-setting process (Wouters & Wilderom, 2008). In this sense, loose controls are a complement to commitment, whereas tight controls are more of a substitute.
It’s impossible to say this is true for every firm, however, and this is supported by con-tingency theory: depending on such things as the internal and external environment, the size
of the firm, and the strategy it follows, firms will behave differently when presented with pos-sible scenarios (Donaldson, 2001). As such, one company may benefit from utilising tight controls, while it is anathema to another. In a contingency view of formalisation, having high levels of formalisation (tight controls) can result in both positive and negative attitudes of employees. When the tasks performed are routine, a high level of formalisation is useful and generally appreciated, whereas with more unique tasks that same high level of formalisation is a hated hurdle to productivity. Agarwal (1999) found that in the case of salespersons, formali-sation leads to reduced organiformali-sational commitment. Lower levels of formaliformali-sation are met with more positive attitudes, as evidenced in Conradi & Dybå (2001). In this study, a subset of professional service employees state that “10% of the routines are useful, while the remain-ing 90% is nonsense” (p. 271), although other responses are more nuanced. The conclusion does state that “formal routines must be supplemented by collaborative, social processes” (p.268), further boosting the importance of formalisation being enabling, not coercive. In ad-dition, Kasim, Amiruddin, and Auzair find that “when task variability is low (high), a MCS with a narrow (broad) scope of information will be positively related [to] performance” (2012, p. 52). It has been noted before that it mostly depends on the rules in place: good rules are taken for granted, while bad rules are despised (Perrow, 1986, p. 24). Following this line of thought, the high variety work present in PSFs should function better under loose controls, keeping attitudes and morale positive, which ought to lead to a better reputation.
3. Hypothesis development
I expect control tightness to have an effect on the organisational reputation of all firms, in-cluding professional service firms. Whether it is tight controls or loose controls that have a positive effect on reputation is debatable, and in any case contingency theory instructs us that the appropriate control system will depend on the case-specific circumstances and aspects of the organisation (Otley, 1980). The same is true for control tightness. However, control tight-ness and exactly how tightly to apply management controls has received little attention in prior research (Merchant & Van der Stede, 2017). Tight controls, being considered more bu-reaucratic than loose controls, impose more restrictions and stricter procedures on employees (Auzair & Langfield-Smith, 2005) and are thus more suited for routine, “mechanistic” work (Adler & Borys, 1996). Another key aspect of tight controls Auzair & Langfield-Smith high-light is their bias towards strong incentives, a method used by executives and managers to try and influence the results of the organisations for the better. As such, when control tightness
does get discussed in the literature, the primary focus is on control tightness in a results con-trol context (Van der Stede, 2001). It would be surprising if only results concon-trol tightness had an effect on reputation, leading me to think that tightness in all four types of controls have an impact on reputation to a greater or lesser degree.
In the first place, I expect the two types of formal control tightness to have a negative effect on the organisational reputation of PSFs. A lot of studies have found that setting specif-ic goals and other such controls increase performance (Chow, 1983; Locke et al., 1981), act-ing as an antecedent to performance. Similarly, reputation can be viewed as an antecedent of performance as well, acting as an asset of the organisation that could lead to superior perfor-mance, which may increase reputation further (Boyd, Bergh, & Ketchen Jr., 2010; Roberts & Dowling, 2002). Performance can then lead to further increases in reputation, making reputa-tion both an antecedent and a consequence to the control system. As performance’s anteced-ent, I expect control tightness to affect reputation as well. For the formal controls, I theorise that giving professionals looser controls and more freedom in the way they can approach their work increases reputation, considering they value autonomy (Barley, 2005; Hall, 1968; Stamps, Piedmont, Slavitt, & Haase, 1978; Raelin, 1989). As such, I have constructed the following hypotheses:
H1a: Professional service firms with low behavioural control tightness have a higher reputation compared to professional service firms with high behavioural control tight-ness.
H1b: Professional service firms with low results control tightness have a higher repu-tation compared to professional service firms with high results control tightness. Many researchers argue that trust and formal controls are inversely related (e.g. Dekker, 2004; Dyer & Singh, 1998; Poppo & Zenger, 2002). In this case, trust would lead employers to em-ploy fewer formal controls, hoping it will lead to better performance (Zaheer, McEvily, & Perrone, 1998). Looking instead at informal controls, I expect tighter social controls to lead to a higher reputation. Reputation as a consequence of the informal controls is based not on the service delivered or any sort of direct experience between public and organisation, but on re-calling what they have heard of an organisation (Bromley, 1993). Grunig and Hung-Baesecke (2015) find that the recall of several organisations’ conduct had the greatest effect on the pub-lic’s evaluation of the organisations’ reputation. As such, compliance with informal controls is a matter of great importance for an organisation’s reputation, which leads me to believe tighter social controls lead to a better reputation. This results in the following hypotheses:
H2a: Professional service firms with high personnel control tightness have a higher reputation compared to professional service firms with low personnel control tightness. H2b: Professional service firms with high cultural control tightness have a higher reputation compared to professional service firms with low cultural control tightness. Up to this point, I have made no distinction between professionals who perform the primary service of the PSF and professionals who perform staff functions, services other than the pri-mary service of the PSF. However, they do not only differ in terms of the function they have within the organisation, but also in their contact with clients. In general, primary service pro-fessionals will have more contact with clients and influence what Bromley (1993) calls the primary reputation, the reputation based on direct experiences with the organisation. Although this may differ between the various occupations considered professional services, this differ-ence in client contact holds true for most occupations; an accountant performing the primary function of an accounting firm will presumably have more client contact than an accountant performing a staff function at a non-accounting firm. If this is the case, it would mean that the effect of primary service professionals on reputation is greater than that of staff function pro-fessionals, at least for an organisation’s primary reputation. In the case of formal control tightness, the actions performed by staff personnel because of set goals or restrictions do not directly translate into the kind of performance that would influence reputation, leading to a less negative effect on reputation:
H3a: The relationship between behavioural control tightness and reputation is less negative for staff function professionals than it is for primary service professionals. H3b: The relationship between results control tightness and reputation is less negative for staff function professionals than it is for primary service professionals.
In the case of informal control tightness, this effect would be in the opposite direction, though similarly subdued. As I expect tighter informal controls to lead to an increase in reputation and primary service professionals experience society’s spotlights more than staff personnel, my expectation is that primary service professionals have a greater effect on reputation through informal controls as well. Put differently, this would mean that informal control tightness has a less positive effect on reputation for staff:
H3c: The relationship between personnel control tightness and reputation is less posi-tive for staff function professionals than it is for primary service professionals.
H3d: The relationship between cultural control tightness and reputation is less positive for staff function professionals than it is for primary service professionals.
As mentioned before, Grunig and Hung-Baesecke (2015) find that the greatest effect on repu-tation is caused by the public’s recall of the organisation’s behaviour and conduct. This af-fects what they call ‘reputational reputation’ and what Bromley (1993) calls ‘secondary repu-tation’ – reputation based on “hearsay” (Bromley, p. 42). Though it is reportedly a greater part of reputation than primary reputation, secondary reputation is also called more superficial by Bromley, making it more difficult to directly influence. The parts of the organisation that can work towards influencing secondary reputation are e.g. management, the Human Resources department, and the Public Relations department, and management. These departments are just a subset of professionals, too specific to research with the available data. Instead I put more focus on primary reputation, which is supported better by the research design and avail-able data.
4. Research design
This chapter explains the way this research is designed. To test my hypotheses, I was in need of a sufficient amount of data. As such, I joined the Professional Service Firm Survey Project (hereafter called the PSF project), which has been running for a number of years already at the University of Amsterdam under the supervision of Helena Kloosterman. The topics highlight-ed in this chapter are the survey design, the respondents, the variables, and analyses of the variables.
4.1 Survey design
The survey used to obtain the data for this research is part of the PSF project and was devel-oped by Helena Kloosterman. Containing questions relevant to control variables, control re-sponses, control tightness, antecedents, and consequences, the data obtained from the survey allows for diverse research possibilities.
To ensure that the questions in the survey measure what they are meant to measure and the quality and effectiveness is maximised, two separate pre-tests were performed. In the first pre-test, a group of twenty professionals was asked to match 52 statements to eight control constructs, these being both explicit and implicit behavioural, results, personnel, and cultural control. Through the completion of the task by fourteen of these professionals, 32 statements
were included in the survey for their low number of incorrect matches, the highest being six incorrect matches for implicit behavioural control. The second-pre-test assessed the quality of the survey as a whole, calling in an additional twenty professionals. Their comments resulted in minor changes in wording and additional answers for a few multiple choice questions.
Every student participating in the PSF project was to submit at least ten completed questionnaires from suitable respondents. To be deemed suitable, a respondent needs to have worked in the field for more than 3 years, be a mid-level employee (not an owner or board member), work in an organisation with more than 50 employees, speak and understand Eng-lish at a business level, and, most importantly, hold a job that is considered a professional service. Because there may be some ambiguity on what is a PSF, for the purpose of the PSF project, professional service firms include the following occupations:
Accounting Actuarial services Advertising Architecture Biotechnology Consulting Engineering Consulting IT Consulting HR Consulting Management/Strategic Consulting Technology Engineering Fashion design Financial advising Graphic design Insurance brokerage Investment banking Investment management Law Marketing/public relations Media production Medicine/Physician practices Pharmaceutical Project management Real estate Recruiting – Executive Research / R&D
Risk management services
Software development
Talent management/agencies
Other
Though the 29 occupations offer a wide range of PSFs, there is also an ‘other’ catego-ry for respondents who did not associate their professional service position with any of the options given. Further information on how the responses of these respondents were dealt with is given in the next paragraph, which handles the respondents.
4.2 Respondents
In the complete database, the number of respondents was 730. For a number of reasons more accurately explained in table 1, the sample for this research ultimately ended up with 491 eli-gible respondents.
Table 1: Respondent selection process
Total respondents 730
Respondents who did not complete the survey 92 -
Respondents who did not read the definition and remarks 21 -
Respondents who did not answer all relevant questions 71 -
Respondents who did not meet the PSF criteria 55 -
Total eligible respondents 491
At this point, 34 respondents who had listed their occupation as ‘Other’ remained. In some cases, the occupations for these respondents were recoded to best fit one of the 29 listed occu-pations, leading to increases in value for these occupations as shown in table 2. For example, people in audit and control functions were recoded to accounting, while occupations like physiotherapy are now part of medicine/physician practices. Accounting is the industry most represented at 17.1% of all respondents, followed by Medicine and Physician practices (11.2%), Consulting IT (7.1%), and Consulting Management and Strategic (6.9%). A full list-ing of the frequency of each occupation is included in the appendices.
Table 2: Recoded respondents' occupation
Total respondents with 'Other' occupation 30
Accounting 11 Consulting IT 2 Consulting Management/Strategic 1 Financial advising 3 Insurance brokerage 1 Law 1 Medicine/Physician practices 9 Pharmaceutical 1
Risk management services 1
Other demographic information gleaned from the dataset tells us the majority of respondents are Dutch, representing 72.3% of the total sample. Further information of note is included in table 3. Of particular interest is the split in whether the respondents have an occupation that
represents the primary service of the PSF or act in a more supportive role, coming close to 60/40 in favour of primary service professionals.
Table 3: Respondent demographics
Percentage N Min Max Mean Std. Dev.
Age 488 23 66 37.23 8.87 Gender* 489 1 2 1.35 0.48 Male 65.85% Female 34.15% Education** 488 1 3 1.75 0.68
Bachelor degree or lower 38.52%
Master degree 47.95%
PhD or other doctorate degree 13.52%
Work representative for organisation*** 491 1 2 1.42 0.49
Represents primary service 57.84%
Represents staff role 42.16%
Years of experience in field**** 477 1 11 7.28 3.05
Years of experience in organisation**** 490 1 11 5.99 3.19 * Male = 1; Female = 2
** Bachelor degree or lower = 1; Master degree = 2; PhD or other doctorate degree = 3 *** Represents primary service = 1; Represents supportive role = 2
**** Less than 1 year = 1; 1 year = 2; …; 9 years = 10; 10 years or more = 11
4.3 Constructs
As evidenced by the research question and literature overview, the variables of greatest im-portance for this research are organisational reputation and the four types of control tightness. Organisational reputation is the dependent variable, while the control tightness constructs are the independent variables. These five constructs are included in the survey as such, built from a number of questions and statements. Respondents were asked to rate the statements relating to these variables on a 5-point Likert scale that scales from Strongly Disagree (1) to Strongly Agree (5).
Not all of the statements were asked in the same direction, however. To increase the reliability of the survey questions and the respondents’ answers, some questions were formu-lated negatively. For example, the statement “In my organisation, we have rules for every-thing” indicates a high measure of control when a respondent strongly agrees with it. When a respondent strongly agrees with the statement “Employees in my organization are encouraged
to adjust procedures to suit the situation”, it indicates a low measure of control. A full list of the questions asked that are relevant to this research are included in the appendices, along with a note whether they were reverse coded or not. In addition to the statements rated on the Likert scale, the results control construct also included two other questions: how many per-formance targets were used in their evaluation, and how often the results of perper-formance measures get discussed with a supervisor. These questions were not used in answering the hypotheses, however.
In addition to the independent variables, there may be other variables that can explain the dependent variable. The first one of these is the organisation size. Defined here as the number of employees working in the organisation, organisational size is often a control varia-ble for this kind of research. Larger firms tend to receive a lot of attention and are more well-known and more attractive to people looking at the market (Stanwick & Stanwick, 1998; Tur-ban & Greening, 1997; TurTur-ban & Keon, 1993). As such, it is expected that larger firms have a higher reputation, or at the very least more potential to garner a favourable reputation. Simi-larly, an organisation’s performance is also closely related to their reputation and the way they are viewed, as Brown and Perry assert by saying “annual ratings of America’s largest corpora-tions are shown to be heavily influenced by previous financial results” (1994, p. 1347). It is thus imperative to control for performance’s influence on reputation as well.
The cause of both positive and negative performance can often be attributed to a firm’s strategies (Hitt et al., 2001). The correct execution of a firm’s strategy at the right time can greatly affect the performance, and subsequently, the reputation of a firm. Many strategy scholars have also researched the notion of reputation as a resource, finding it is linked with strategy and competitive advantage (Clark & Montgomery, 1998; Deephouse, 2000; Mahon, 2002). As such, strategy is another variable with major consequences for reputation, meaning it is a valuable control variable as well. The final control variable I consider to be important to include in this research is that of environmental uncertainty. Environmental uncertainty is the (un)predictability and stability in an organisation’s environment (Gordon & Narayanan, 1984), requiring that organisations adapt to the happenings of their environment to remain a viable business (Duncan, 1972). This is corroborated upon by Swamidass and Newell (1987), who concluded that environmental uncertainty influenced strategic variables which, in turn, influ-ence business performance and consequently reputation.
4.4 Validity analysis
Before starting to do the actual research with the variables mentioned earlier, it is important to make sure they are both reliable and valid in use. To ascertain this, the variables were subject-ed to both factor analysis and reliability analysis using Cronbach’s alpha (α). Factor analysis looks at the input variable to see if they measure different aspects of the same underlying var-iable (Field, 2014). For example, the survey questionnaire has several questions for the four types of control tightness. These questions are used to measure tightness, but more specifical-ly, measure the explicit and implicit tightness. As a result, when using factor analysis with the control tightness variables included, it is expected that there will be two components for each type of tightness, one for explicit tightness and one for implicit tightness. As an extension of this, we test the reliability of a variable with Cronbach’s alpha. If alpha is high, the questions included measure the same underlying thing. As for what constitutes as a ‘high alpha’, scores higher than 0.7 are acceptable (Nunnally, 1978), while others argue that 0.49 can still be ac-ceptable if it has other ‘desirable properties’ (Schmitt, 1996). The benchmark used in this analysis accepts alphas of 0.6 and higher, which Churchill Jr. (1979) suggests is sufficient.
Two tests were performed to measure how well-suited the data is for factor analysis, these being the Kaiser-Meyer-Olkin (KMO) Test and Bartlett’s test of sphericity. KMO scores between 0.5 and 1.0 indicate the data is suitable for factor analysis, showing that the patterns of correlations are relatively compact. Bartlett’s test tests whether the correlation matrix used is actually an identity matrix, meaning the variables do not correlate at all. This score is sig-nificant, thus indicating that factor analysis is possible (0.000 < 0.05). Similarly, the KMO measure equals 0.797 and also indicates that the data used is suitable for factor analysis.
For reputation (REP), the 4 questions resulted in a single eigenvalue greater than 1. As shown by Cronbach’s alpha in table 4, the reliability of the factor is also more than sufficient.
Table 4: Factor analysis REP
Reputation Component
Factor REP
1 Well-respected 0.880
2 Good value for price 0.712
3 Consistent quality and service 0.860
4 Brand name recognition 0.715
Initial Eigenvalues 2.530
Variance explained (%) 63.26%
Cronbach's α 0.798
The first part of the formal controls has a problematic variable; respondents seem to have mis-interpreted question six. This question will be omitted from the next stage of research. As shown in table 5, Behavioural Control Tightness (BCT) gives us two eigenvalues greater than 1, which correspond to explicit and implicit tightness (EBCT, IBCT). The questions for ex-plicit and imex-plicit move in different directions, meaning a high score on the first four ques-tions would logically lead to low scores on the latter four quesques-tions. To be able to use all questions in a single variable, the latter questions were reverse coded to move in the same direction as the first four. Even so, question #6 seems to have been misinterpreted by re-spondents, and will be omitted from future analyses. This raises the alpha from 0.548 to a more reliable 0.703.
Table 5: Factor analysis BCT
Behavioural Control Tightness Component
Factor EBCT IBCT
1 Existing procedures and rules for whatever situation 0.757 0.103
2 Established procedures cover job tasks 0.765 -0.022
3 Rules for everything 0.755 0.270
4 Frequently monitored 0.686 -0.109
5 Allows deciding how to adjust rules 0.066 0.787
6 Procedures as broad guidelines -0.680 -0.080
7 Encouraged to use procedures flexibly 0.110 0.841
8 Encouraged to adjust procedures per situation -0.015 0.825
Initial Eigenvalues 2.865 1.921
Variance explained (%) 35.81% 24.01%
Cronbach's α 0.758 0.530
Cronbach's α (total) 0.548
Cronbach's α - 1 0.703
Much like BCT, Results Control Tightness (RCT) has two eigenvalues exceeding 1, resulting in explicit and implicit RCT (ERCT, IRCT). Table 6 further shows the Cronbach’s alpha for implicit RCT is under the minimum of 0.5 mentioned by Nunnally (1978), but seeing as how RCT as a whole is used in future analyses, RCT’s alpha of 0.577 is more relevant. While an acceptable score, omitting question #6 results in an alpha of 0.636 – a more comfortable score for reliability.
Table 6: Factor analysis RCT
Results Control Tightness Component
Factor ERCT IRCT
1 Measure for everything 0.770 0.004
2 Large number of targets 0.720 -0.002
3 Goal attainment is checked constantly 0.779 -0.169
4 Frequent performance target checks 0.739 -0.212
5 Targets are a guideline 0.036 0.680
6 Considerate of deviations from targets -0.179 0.630 7 Opportunities more important than achieving targets 0.010 0.709
8 Meet target with no exceptions 0.656 0.089
Initial Eigenvalues 2.797 1.372
Variance explained (%) 34.97% 17.15%
Cronbach's α 0.789 0.422
Cronbach's α (total) 0.577
Cronbach's α - 1 0.636
Surprisingly, the first type of informal control delivers three variables with an eigenvalue over 1. Table 7 shows a split between explicit tightness (EPCT) and implicit tightness, although the questions concerning implicit tightness lead to two components (IPCT1, IPCT2). The first focuses on pre-organisation knowledge, while the other has a focus on competency consisten-cy. PCT’s alpha stands at a solid 0.652 and omitting any questions would only lead to lower-ing the score.
Table 7: Factor analysis PCT
Personnel Control Tightness Component
Factor EPCT IPCT1 IPCT2
1 Hiring process is extensive 0.827 0.026 -0.039
2 Many steps to get hired 0.841 0.109 -0.101
3 Interviewed with several people 0.714 -0.048 0.108
4 Hiring evaluates knowledge and skills 0.558 0.126 0.192
5 Colleagues same experience before hiring 0.071 0.859 0.103
6 Little consistency in hiring 0.102 0.000 0.858
7 Colleagues same education before hiring 0.066 0.846 0.129
8 Competence of employees varies greatly -0.002 0.258 0.734
Initial Eigenvalues 2.411 1.685 1.051
Variance explained (%) 30.14% 21.06% 13.14%
Cronbach's α 0.725 0.680 0.512
Cronbach's α (total) 0.652
Similarly to PCT, Cultural Control Tightness (CCT) also comes with three eigenvalues ex-ceeding 1. Table 8 shows that the variable is split three components, one explicit and two of implicit types of personnel control tightness (ECCT, ICCT1, ICCT2). The first type of implic-it CCT focuses on the relationship employee to employer, while the second is centred on col-leagues. Cronbach’s alpha for CCT stands at a reliable 0.759 and omitting a question would lead to a negligible increase that’s not worth the loss of explanatory power.
Table 8: Factor analysis CCT
Cultural Control Tightness Component
Factor ECCT ICCT1 ICCT2
1 Socialise with colleagues outside work 0.178 0.146 0.804
2 Personal and organisational values converge 0.162 0.793 0.146
3 Regular social events 0.804 0.140 0.153
4 Not friends with any colleagues 0.058 0.025 0.861
5 Sense of "ownership" for organisation 0.110 0.839 0.077
6 Organisation communicates core values 0.542 0.508 -0.092
7 Organisation plans team-building events 0.793 0.210 0.192
8 Organisation sponsors events 0.755 0.059 0.055
Initial Eigenvalues 3.059 1.263 1.058
Variance explained (%) 38.24% 15.79% 13.23%
Cronbach's α 0.764 0.660 0.623
Cronbach's α (total) 0.759
Cronbach's α - 1 0.765
Moving to the first of the control variables, Performance (PERF) has two eigenvalues with a score greater than 1. As shown in table 9, its score for Cronbach’s alpha is also sufficiently good that no further alterations of the variable are necessary.
Table 9: Factor analysis PERF
Performance Component
Factor 1 2
1 More competitive 0.649 0.387
2 Greater market share 0.249 0.867
3 Growing faster 0.785 0.217 4 More profitable 0.596 0.429 5 More innovative 0.796 -0.028 6 Larger in size 0.105 0.902 Initial Eigenvalues 2.992 1.055 Variance explained (%) 49.86% 17.58% Cronbach's α 0.741 0.753 Cronbach's α (total) 0.794 Cronbach's α - 1 0.794
Strategy has three initial eigenvalues exceeding 1, and all three components have a sufficient-ly high alpha. The components of strategy can be labelled as being focused on differentiation (DIFF), cost efficiency (COST), and service-oriented or quality (QUAL), which are depicted as such in table 10. These three components focus on such different strategies that it is not prudent to throw all strategy-related questions in a single variable. As such, they will be en-tered as separate control variables in the regression analyses. The questions are divided as follows: COST contains questions 1 through 3, DIFF uses questions 4 through 7, while QUAL is made up of questions 8 through 11. For all three of these variables alpha is at a reli-able level, making alterations unnecessary and ineffective.
Table 10: Factor analysis Strategy
Strategy Component
Factor DIFF COST QUAL
1 Lower cost than competition 0.057 0.786 -0.152
2 Increasing cost efficiency 0.155 0.728 0.263
3 Improving cost for coordination 0.321 0.711 0.184
4 Improving equipment utilisation 0.512 0.297 0.121
5 Introducing new services quickly 0.578 0.280 0.179
6 Providing distinct services 0.762 0.010 0.206
7 Offering broad range of services 0.804 0.093 0.080
8 Improving time to consumer 0.110 0.476 0.550
9 Providing high quality services 0.192 -0.073 0.757
10 Customizing services to customers' needs 0.080 0.158 0.751 11 Providing after-sale service and support 0.376 0.084 0.511
Initial Eigenvalues 3.713 1.372 1.078
Variance explained (%) 33.76% 12.47% 9.80%
Cronbach's α 0.680 0.713 0.646
Cronbach's α - 1 0.669 0.707 0.592
Table 11 shows that Environmental Uncertainty (ENV), the final control variable, is also split up in two components. This is reflected in the questions of the variable, with questions 1 through 3 specifically being about the intensity of the market (INT), while questions 5 and 6 questioned respondents about the predictability of the market. Question 4, about the number of new products and services brought to market by the organisation, seems like an odd inclu-sion, yet a reliability analysis shows it is indeed relevant for ENV, as Cronbach’s alpha de-creases from 0.604 to 0.566 when omitting the question.
Table 11: Factor analysis ENV
Environmental Uncertainty Component
Factor INT PRED
1 Intensity of bidding for contracts 0.835 0.041
2 Intensity of competition for manpower 0.586 0.133
3 Intensity of price competition 0.815 0.053
4 New products bought to market 0.296 0.542
5 Industry predictability -0.033 0.825 6 Client predictability 0.056 0.744 Initial Eigenvalues 2.026 1.320 Variance explained (%) 33.77% 22.00% Cronbach's α 0.643 0.510 Cronbach's α (total) 0.604 Cronbach's α - 1 0.585
5. Results
The fifth chapter details the results found after the testing of the hypotheses through linear regression. First, however, the variable descriptives and the correlation between variables are given for a greater understanding of the variables used.
5.1 Descriptive statistics
As with the survey respondents, the variables used in hypothesis testing give us insight in professionals, though we are now looking at the PSF itself more than the professional as indi-vidual. As shown in table 12, nearly all variables used for this research are above the middle value of 3, with only Results Control Tightness below this midpoint and Behaviour Control Tightness nearly equal to it. The highest mean value for one of the independent variables is Cultural Control Tightness, an interesting point to keep in mind.
Reputation has a particularly high mean, averaging a score higher than 4 out of a max-imum of 5. This indicates that the respondents consider their organisation’s reputation to be good, or expect their clients to consider the organisation’s reputation to be good. One other thing of note about the descriptive statistics of the variables is the size of firms represented in the PSF survey project. The mean value for organisation size is rather high at a value of 2.960 – only 31% of respondents work in an organisation totalling fewer than 500 employees. This can be explained through the survey design, as one of the criteria for eligible respondents was that their organisation had to total at least 50 employees. Finally, the Staff variable isn’t shown in the descriptive statistics table since it is a dummy variable scoring either 0 or 1.
Though this would give the percentage of respondents in a staff function (42%), this infor-mation is also shown in table 3.
Table 12: Descriptive statistics
N Min Max Mean Std. Dev. Items
REP 491 1.50 5.00 4.123 0.686 4 BCT 491 1.00 4.86 3.006 0.664 7 RCT 491 1.00 4.29 2.807 0.599 7 PCT 491 1.00 5.00 3.184 0.584 8 CCT 491 1.63 5.00 3.519 0.664 8 SIZE* 489 1.00 4.00 2.963 1.018 1 PERF 491 1.00 5.00 3.417 0.726 6 COST 491 1.00 5.00 3.511 0.803 3 DIFF 491 1.00 5.00 3.547 0.740 4 QUAL 491 1.50 5.00 3.839 0.660 4 ENV 491 1.33 4.83 3.188 0.615 6
* 1 = Less than 100; 2 = More than 100 but less than 500; 3 = More than 500 but less than 5000; 4 = More than 5000
Moving on to table 13’s correlation matrix showing the correlations between the variables that are of interest to this research, it appears the control variables were well-chosen. All control variables correlate with REP to a greater or lesser degree, with COST and ENV correlating to the lesser degree, being significant only at p < 0.05 rather than p < 0.01. Surprisingly, two of the independent variables do not correlate with reputation: BCT (r = 0.080; p > 0.05) and RCT (r = 0.074; p > 0.05). As a result, the hypotheses involving these formal controls will probably not be supported by the statistics, though this will still be tested in the next para-graph. All of the independent variables do correlate with SIZE, PERF, and COST, however. DIFF correlates with all but one of the independent variables (p < 0.01), but does not correlate with BCT. There are no correlations that could indicate multicollinearity (r > 0.8), meaning there is no apparent risk of multiple variables explaining the same thing when running the regression analyses.
With this correlation matrix, we are now aware of the correlations that exist between the variables used. Correlation does not necessarily signify causation, which is what will be tested for in the regression analyses of the next paragraph.
Table 13: Correlation matrix
REP BCT RCT PCT CCT SIZE PERF COST DIFF QUAL ENV Staff
REP BCT 0.080 RCT 0.074 0.340** PCT 0.150** 0.153** 0.226** CCT 0.356** -0.032 0.072 0.274** SIZE 0.205** 0.265** 0.094* 0.226** 0.095* PERF 0.449** 0.144** 0.126** 0.160** 0.196** 0.223** COST 0.094* 0.134** 0.160** 0.112* 0.104* 0.053 0.208** DIFF 0.321** -0.017 0.144** 0.143** 0.221** 0.005 0.310** 0.422** QUAL 0.367** -0.015 0.097* 0.102* 0.260** -0.016 0.285** 0.371** 0.481** ENV 0.093* -0.059 0.110* 0.010 0.034 -0.004 0.147** 0.219** 0.289** 0.217** Staff -0.028 0.010 0.005 -0.064 -0.038 0.072 0.002 0.157** 0.125** 0.053 0.078 * p-value significant at < 0.05 ** p-value significant at < 0.01
5.2 Regression analysis
Before running the regression analysis, there are yet more tests that will determine whether multicollinearity exists in this data set. Field (2013) mentions that multicollinearity can be identified by scanning a correlation matrix for values higher than 0.8 or 0.9, but this method may not pick up on less obvious forms of correlation. There are also other diagnostics availa-ble, such as the variance inflation factor (VIF) and the tolerance statistic. The general rule of thumb for the VIF is that its largest value should not be greater than 10 (Bowerman & O’Connell, 1990; Field, 2013). Dividing 1 by the VIF leads to the tolerance, which may be a cause for concern at a value below 0.2 (Menard, 1995). These rules of thumb are merely that, however: guidelines that may cause researchers to question statistically solid results (O’Brien, 2007). In all regression analyses performed in this research VIF does not exceed 10 or even 5, which means there is no cause for concern on this front either.
Hypotheses 1a and 1b predicted that PSFs with low formal control tightness would have a higher reputation when compared to PSFs with high formal control tightness. This would mean that if BCT and RCT were to be made tighter, the organisation’s reputation would lessen. This predicted negative relation would see a graph with a downward slope. On the other hand, hypotheses 2a and 2b predicted that PSFs with high informal control tightness would have a higher reputation when compared to PSFs with low informal control tightness. On a graph this would show PCT and CCT as rising lines, increasing the firm’s reputation as the controls become tighter. With this in mind, the regression analyses were run, the results of which can be seen in table 14.
The regression analyses are set up as follows: the first model looks at the effect of just the control variables on reputation, the following four models look at the effect of the control variables and each of the independent variables in order, while the sixth and final model looks at the total of control variables and independent variables. In all models, all control variables except for environmental uncertainty have a significant effect on reputation. Of these, only the COST strategy variable has a negative beta. This lends a measure of trustworthiness to the models, as they behave as would be considered logical in practice. An organisation with a cost leadership strategy will not be particularly invested in producing high quality services, thus lowering reputation in that regard. On the other hand, a focus on service and quality would lead to positive word-of-mouth, while the positive beta of DIFF can be explained by Deephouse and Suchman’s argument that differentiation is an important aspect of reputation (2008).
Table 14: Regression analyses control tightness’ effect on reputation
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
CV CV + BCT CV + RCT CV + PCT CV + CCT CV + CT
Variable Pred β t-value Sig. β t-value Sig. β t-value Sig. β t-value Sig. β t-value Sig. β t-value Sig. (constant) 8.208 0.000* 7.145 0.000* 7.695 0.000* 6.963 0.000* 5.642 0.000* 4.656 0.000* SIZE 0.143 3.654 0.000* 0.137 3.409 0.001* 0.143 3.650 0.000* 0.136 3.396 0.001* 0.126 3.320 0.001* 0.119 2.996 0.003* PERF 0.324 7.764 0.000* 0.321 7.650 0.000* 0.325 7.753 0.000* 0.322 7.685 0.000* 0.303 7.437 0.000* 0.299 7.281 0.000* COST -0.145 -3.349 0.001* -0.149 -3.404 0.001* -0.144 -3.321 0.001* -0.146 -3.379 0.001* -0.137 -3.281 0.001* -0.143 -3.362 0.001* DIFF 0.162 3.465 0.001* 0.164 3.496 0.001* 0.162 3.464 0.001* 0.158 3.378 0.001* 0.135 2.955 0.003* 0.141 3.063 0.002* QUAL 0.260 5.779 0.000* 0.262 5.800 0.000* 0.260 5.774 0.000* 0.259 5.751 0.000* 0.218 4.923 0.000* 0.220 4.959 0.000* ENV -0.027 -0.687 0.493 -0.026 -0.636 0.525 -0.027 -0.677 0.499 -0.026 -0.647 0.518 -0.017 -0.427 0.670 -0.011 -0.292 0.770 BCT - 0.025 0.630 0.529 0.051 1.222 0.222 RCT - -0.005 -0.135 0.893 -0.025 -0.606 0.545 PCT + 0.035 0.877 0.381 -0.015 -0.383 0.702 CCT + 0.216 5.555 0.000* 0.223 5.572 0.000* r² 0.307 0.307 0.307 0.308 0.349 0.351 Adjusted r² 0.298 0.297 0.297 0.298 0.339 0.337 F 35.538 30.480 30.402 30.557 36.758 25.828 Sig. F 0.000* 0.000* 0.000* 0.000* 0.000* 0.000* * p-value significant at < 0.01
Of the independent variables, only CCT is significant in either the individual models (β = 0.216; p < 0.01) or the complete model (β = 0.223; p < 0.01). The positive effect of CCT’s beta is in line with the predicted direction, supporting the hypothesis that tighter cultural con-trols increase a professional service firm’s reputation. As such, H2b is supported by the data.
As the data shows that BCT, RCT, and PCT do not have a significant effect on reputa-tion, this means that hypotheses H1a, H1b, and H2a are not supported. In the case of BCT, the predicted negative direction of effect does not correspond to the actual result either. The non-significance of the BCT and RCT is not unexpected, but rather confirms their lack of correla-tion with REP seen earlier in the correlacorrela-tion matrix. For the first four models, the r-squared (r²) and adjusted r-squared (adjusted.r²) are close to 0.307 and 0.297 respectively, indicating that close to 30% of reputation can be explained by the variables used in the regression analy-sis. Since these values barely change for the first four models, this is another indicator that BCT, RCT, and PCT have no significant effect on reputation. Meanwhile, the models contain-ing CCT explain reputation roughly 4% more than the other models.
This trend continues in table 15, which shows the results of the regression analyses performed to find support for hypotheses 3a, 3b, 3c, and 3d. While the first four hypotheses predicted directional effects of control tightness on reputation, the prediction for these other hypotheses is that the effect is more subdued. In the case of the formal BCT and RCT, which were expected to have a negative effect on reputation, this would result in a less negative rela-tionship. For the informal PCT and CCT, the expected direction for H3c and H3d is negative, as they are expected to be less positive.
The models for regression analysis are structured similarly to the previous regressions. First, four models are used for each individual variable of control tightness, with the last mod-el including all variables of control tightness. None of the modmod-els show any kind of signifi-cant effect for any of the Staff-related variables, however. Although the direction of the beta is in line with the predicted direction for each individual control tightness variable, there is not enough statistical evidence to support any of the H3 hypotheses. Only RxSt has an effect, though it is very minimal (β = 0.362; p < 0.1). This research does not consider a p-value high-er than 0.05 to be sufficient statistical support, howevhigh-er, meaning that only one of the con-ceived hypotheses is supported by the data: H2b. In the next chapter I discuss why this may be the case.
Table 15: Regression analyses staff function's effect on control tightness’ effect on reputation
Model 1 Model 2 Model 3 Model 4 Model 5
CV + BCT CV + RCT CV + PCT CV + CCT CV + CT
Variable Pred β t-value Sig. β t-value Sig. β t-value Sig. β t-value Sig. β t-value Sig. (constant) 6.568 0.000* 7.331 0.000* 5.410 0.000* 4.762 0.000* 3.667 0.000* SIZE 0.141 3.479 0.001* 0.147 3.721 0.000* 0.136 3.381 0.001* 0.128 3.345 0.001* 0.115 2.871 0.004* PERF 0.319 7.570 0.000* 0.321 7.658 0.000* 0.316 7.540 0.000* 0.302 7.394 0.000* 0.293 7.131 0.000* COST -0.144 -3.256 0.001* -0.147 -3.306 0.001* -0.132 -3.027 0.003* -0.135 -3.191 0.002* -0.149 -3.396 0.001* DIFF 0.169 3.586 0.000* 0.168 3.581 0.000* 0.161 3.426 0.001* 0.140 3.060 0.002* 0.146 3.167 0.002* QUAL 0.259 5.711 0.000* 0.258 5.732 0.000* 0.256 5.684 0.000* 0.215 4.840 0.000* 0.214 4.798 0.000* ENV -0.023 -0.571 0.568 -0.023 -0.568 0.570 -0.015 -0.374 0.709 -0.016 -0.416 0.678 0.003 0.082 0.935 BCT - 0.010 0.186 0.853 0.055 0.981 0.327 RCT - -0.040 -0.815 0.416 -0.086 -1.625 0.105 PCT + 0.076 1.511 0.132 0.030 0.566 0.572 CCT + 0.243 4.913 0.000* 0.249 4.788 0.000* Staff -0.120 -0.683 0.495 -0.248 -1.324 0.186 0.272 1.253 0.211 0.160 0.795 0.427 0.102 0.326 0.744 BxSt + 0.077 0.425 0.671 -0.002 -0.011 0.991 RxSt + 0.209 1.092 0.275 0.362 1.807 0.071 PxSt - -0.323 -1.483 0.139 -0.337 -1.488 0.137 CxSt - -0.198 -0.981 0.327 -0.155 -0.735 0.463 r² 0.310 0.311 0.313 0.351 0.360 Adjusted r² 0.297 0.298 0.300 0.339 0.340 F 23.876 23.985 24.233 28.772 17.737 Sig. F 0.000* 0.000* 0.000* 0.000* 0.000* * p-value significant at < 0.01
6. Discussion
In this final chapter, the results found previously are discussed in detail before drawing sever-al conclusions from them. These conclusions are then set against the limitations this research faced, before finishing with a few directions future research could take in this field.
6.1 Conclusions
This paper contributes to the growing body of literature about professional service firms and reputation in several ways. In the first place, it looks at the effect of control tightness on repu-tation for PSFs specifically, as it seemed probable that service professionals would differ from other workers in this regard due to the nature of their work and their professional disposition. Secondly, it provides more research that primarily views reputation as a consequence of what goes on inside an organisation along the lines of Lai, Chiu, Yang, and Pail (2010) and Selnes (1993). Previous research on reputation has commonly regarded it only as an antecedent to other aspects of business, such as the effect it has on performance (e.g. Aqueveque & Ravasi, 2006; Boyd, Bergh, & Ketchen Jr., 2010; Roberts & Dowling, 2002), credibility (Hutton & Stocken, 2007), or management decisions (Dollinger, Golden, & Saxton, 1997). Since reputa-tion is an important asset for professional service firms (Hitt et al., 2006), I expected it to be a goal PSFs strive towards. As such, I set up hypotheses that would test the direct effect of the level of control tightness on reputation. Additionally, I set up similar hypotheses that would test whether the effect differed between staff professionals and primary service professionals.
Out of the eight hypotheses set up to test both of these things, only one found signifi-cant support in the data: there is evidence supporting the claim that cultural control tightness has a positive direct effect on reputation. This is in line with Grunig and Hung-Baesecke’s (2015) findings, which state that reputational reputation has the greatest effect on a firm’s reputation. Since cultural controls revolve around the organisation’s core values and beliefs, applying these controls tightly to ensure members of the organisation act as excellent repre-sentatives would result in positive coverage rather than negative hearsay. The direct effect of cultural control tightness is thus not only supported by the findings, but also by some earlier literature. The lack of support for behavioural, results, and personnel control tightness’ effect on reputation is not terribly surprising, however, considering the regression analyses run test-ed for a direct relational effect. It appears that this is too direct of a relationship, which is fur-ther corroborated by the lack correlation between BCT and REP and between RCT and REP.
