Institutionalization of changed behavior and the role of the change recipients’ beliefs
The systematic development of an assessment instrument
Master thesis, Msc Business Administration, Change Management
University of Groningen, Faculty of Economics and Business
October 4, 2012
2
ABSTRACT
To effectively plan and improve the institutionalization of changed behavior in organizations a change agent needs to be able to assess salient precursors to change recipient’s behavioral reactions. Using a systematic item development framework as a guide (e.g. item development, questionnaire administration, exploratory item reduction, and scale evaluation), a thorough analysis of an existing instrument measuring change recipient’s beliefs, and literature on institutionalization the author has developed an instrument which is able to predict the progress of the institutionalization of changed behavior. Based on the literature five different precursors for institutionalization were expected but the factor analysis revealed that these five factors are perceived as only two factors by the change recipients showing evidence for a changed individual cognitive process during institutionalization. Moreover, the results show that the salience of beliefs for performing certain behavior change during a change process as well, offering valuable information for successful institutionalization of changed behavior. Although the instrument should be regarded as a preliminary step in the development of a psychometrically sound instrument, the results are encouraging and new insights have been gained about the process and precursors of
institutionalization.
3
INTRODUCTION
The institutionalization of change is an important part of the change process and is also often mentioned as one of the reasons why changes fail (Buchanan, Fitzgerald, Ketley, Gollop, Jones & Lamont, 2005; Kotter, 1995; Goodman & Dean Jr., 1981). In order to give change agents a helping hand, Armenakis, Bernerth, Pitts, & Walker (2007) developed an instrument which intends to gauge the process of an organizational change effort via a self-report questionnaire measuring salient individual beliefs towards the change and is called the Organizational Change recipients’ Beliefs Scale (OCRBS). Via the instrument a change agent can retrieve important information regarding the beliefs of the change recipients on which actions can be based to improve the readiness, adoption, and institutionalization of a change project. The five beliefs which Armenakis et al. (2007) state as most salient in determining the reactions of change recipients are discrepancy, appropriateness, self-efficacy, principal support, and valence.
The salience of these beliefs for the success of organizational change is often proven (van Dam, 2005; Kotter, 1995; Bandura, 1986; Rafferty & Griffin, 2006; Rousseau & Tijoriwala, 1999; Vroom, 1964). However, there are two general reasons to doubt whether the assessment instrument is useful in the institutionalization phase.
4
be less interesting regarding the institutionalization of change compared to the readiness and adoption phases (Lewin, 1947).
As institutionalization is seen as an important aspect of the change process and as it is useful for the change agent to be able to obtain key information regarding the institutionalization it will be valuable to develop an instrument focused on measuring the most salient beliefs for institutionalization, with the correct items to measure those beliefs, and phrased correctly. Therefore, the aim of this research paper is to develop a measurement instrument which is better able to predict institutionalization compared to the OCRBS. This analysis is a preliminary step in the development of a psychometrically sound instrument on which scholars can make further improvements and can apply a confirmatory analysis on. Besides, to increase improvement possibilities and to gain extra insights about institutionalization, the beliefs which are expected to have less influence on institutionalization are tested as well.
To develop the new instrument it is important to know which aspects and beliefs are of importance for institutionalization. To answer this questions several areas are researched and analyzed. First the meaning of institutionalization and the role of the individual’s cognitive process during institutionalization are discussed. Subsequently the framework of Goodman & Dean (1981) is used to explain the process and important aspects of institutionalization. Besides, an extra dimension is given to that framework by expanding it with the model about different individual cognitive modes which is seen as important for institutionalization (Louis & Sutton, 1991). Lastly, a critical analysis of the OCRBS shows opportunities for improvement of the ORRBS instrument.
The new instrument will be developed in line with the recommended scale development process of Hinkin (1998). The instrument will be tested for psychometrically soundness and will be compared with the OCRBS to assess whether it indeed predicts institutionalization better using five criterion related validity measures. The analysis, the comparison of the analysis, and the results will be discussed and further research opportunities are identified.
THEORATICAL FRAMEWORK Institutionalization
5
The first definition is used by Buchanan et al. (2005) in a literature review on sustaining organizational change where institutionalization is defined as “the process through which new working processes,
performance goals, and improvement trajectories are maintained for a period appropriate to a given context” (p189). Another definition is “when new ways of working and improved outcomes become the norm. Not only have the processes and outcomes changed, but the thinking and attitudes behind them are fundamentally altered and the systems surrounding them are transformed in support…..” (NHS
modernization agency, 2002, p12). Moreover, Goodman & Dean (1981) focus on an act and say that “an
institutionalized act is defined as behavior that is performed by two or more individuals, persists over time, and exists as social fact, where, behavior as a social fact means that it exists external to any individual as part of the social reality, that is, it is not dependent on any particular individual” (p229).
The fourth definition is one of Beer (1976) who defines institutionalization as “the stabilization of change
at a new equilibrium state through supporting changes in reference group norms, culture or organizational policy and structure” (p939) which also is used in the paper of Armenakis et al. (1999)
A comparison of these definitions shows that there is not a clear view on institutionalization. In the first place, Buchanan et al. (2005) and the NHS modernization agency (2002) focus on the influence of institutionalization on persistence of behavior (e.g. the processes resulting in maintenance), while Goodman & Dean (1981) and Beer (1976) see institutionalization as an (end) state with a certain degree persistence which occurs by meeting certain conditions. The difference between the two groups is the focus of institutionalization; process focused and behavior focused. Another difference between the scholars is that Buchanan et al. (2005) describe it as sustaining change without reference to specific how aspects, while the other three talk about aspects like altered thinking / attitudes, behavior as a social fact, and changed reference group norms, culture, policy, and structure. This implies that the process of institutionalization has different influencers (e.g. separate processes) like a more social institutionalization (e.g. behavior becomes the norm) and more formal institutionalization (e.g. structural alignment). Finally, the institutionalization process can be seen on a collective and an individual level. Except for the definition of Buchanan (2005), the definitions of NHS modernization agency (2002), Goodman & Dean (1981), and Beer (1976) see a role of the group within the institutionalization process as phrases like ‘behavior becomes the norm, performed by two of more individuals, and exists external to any individual’ are mentioned in the definitions assuming influence of group processes on the institutionalization of the individual behavior.
6
cognitive process. The intention to continue performing behavior as reflection of institutionalization is in line with behavioral scientists who identified the important role beliefs play in the individual’s cognitive process and affect their intention and performance of behavior (Ajzen, 1991; Fishbein & Ajzen, 1975). In these theories the beliefs of individuals are seen as important precursors for behavioral intentions and performance of behavior. This implies that when the individual’s beliefs about the new behavior change the intention and performance of the new behavior will change as well which is reflected in Bandura’s learning theory (1982). Based on this, the difference between the degree and process of institutionalization is clear as well as the degree of institutionalization is the result of individual cognitive process which in itself is affected by all processes affecting the believes of the individual (e.g. development of norms, existence of rewards for performing the behavior etc.).
OCRBS (Armenakis et al., 2007)
The OCRBS is a measurement instrument measuring five individual’s beliefs in order to predict the level of readiness, adoption, and institutionalization and is therefore also based on the individual’s cognitive process in line with the Ajzen’s (1991) theory of planned behavior. The five beliefs are developed in a theory about institutionalization of change where it is argued that the change message should include messages which affect the individual’s beliefs regarding discrepancy, appropriateness, efficacy, principal support, and personal valence in order to increase institutionalization where higher beliefs reflect higher institutionalization (see table 1 for the definitions). Armenakis et al. (2007) measure the several beliefs and assume that when the beliefs are judged as ‘too low’ actions can be planned and executed to improve those beliefs and accordingly improve readiness, adoption, and institutionalization.
TABLE 1
The salient beliefs for institutionalization (Armenakis et al., 1999) Belief Definitions
Discrepancy Change recipient’s belief that the need for change exists.
Appropriateness Change recipient’s belief that the new behavior is appropriate for the situation
Efficacy Change recipient’s belief about the perceived capability of the change recipient and the organization to implement the change
Principal support
Change recipient’s belief that the formal and informal leaders are committed to successful implementation and institutionalization of the change
7
The significance of these five beliefs cannot be denied as much research shows the importance of those beliefs in the change process (van Dam, 2005; Kotter, 1995; Bandura, 1986; Rafferty & Griffin, 2006; Rousseau & Tijoriwala, 1999; Vroom, 1964). However, although Armenakis et al. (1999) argue that that these five beliefs result in commitment which is according to them the natural indicator of institutionalization, not much literature and research specifically focused on institutionalization is included. Therefore it is important to review the literature on institutionalization and related concepts in order to discuss the functionality of their instrument. The OCRBS will be the basis for the development of a new instrument solely focused on measuring the institutionalization of new behavior.
Theory on Beliefs and Processes of Institutionalization
Underneath the important aspects of institutionalization will be discussed. To structure this discussion the framework of Goodman & Dean (1981) will be used as it reflects the important aspects of institutionalization and as it is one of the only few comprehensive theories on institutionalization. Goodman & Dean (1981) developed a theory with five different aspects which explain the degree as well as the process of institutionalization. They argue that institutionalization exists of knowledge of the behavior, performance, preference, normative consensus, and value consensus. The first three aspects Goodman & Dean mention (e.g. knowledge, performance, preference) are seen as conditions for institutionalization to happen while the latter two (e.g. normative consensus and value consensus) do reflect the degree of institutionalization. Moreover, Goodman & Dean (1981) see the first three aspects on an individual level while the other two are expected to happen on a collective level.
The knowledge aspect involves the knowledge a change recipient has of the behaviors associated with the change intervention. According to Goodman & Dean (1981) it is important that the individual knows enough about the behavior to be able to perform it and to understand what will happen when performed. For institutionalization, the extent to which a common cognitive representation of each behavior exists among participants is important as institutionalization is a social construction of the reality (Goodman & Dean, 1981)
The second aspect is the performance of the behavior which consists of the adoption and continuance decision. Important aspects of this adoption decision are the perceived ability to perform the behavior, the expected outcomes, and the attractiveness of the outcomes (Goodman & Dean, 1981). The adoption decision is followed up by repeating decisions to continue performing the specific behavior (Goodman & Dean, 1981; Armenakis et al., 2000; Buller & McEvoy, 1989; Conlon, 1980). The decision to continue performing the specific behavior is central to institutionalization as is reflects the persistence of behavior. Conlon (1980) calls the moments where the decision to continue behavior are made the moments of
8
expectation of the behavior was incorrect), unexpected outcomes (e.g. different results than expected), and / or the availability of a (new) alternative (e.g. new behavior is found with potentially better results or intrinsically more beneficial) (Conlon, 1980). The continuance of the performance of the new behavior can happen for two reasons. In the first place when there is no perceived cause to reevaluate (e.g. none of the three triggers apply), resulting in an automatic continuance of the new behavior. Secondly, when there is a reason to reevaluate but the new behavior is still preferred over other potential behaviors. The fewer moments of reevaluation exist and the greater the preference towards the new behavior compared to other behavior, the higher the intention and a actual performance of the behavior will be which therefore reflects a higher degree of institutionalization.
The third aspect is the development of a preference towards the behavior which is expected to advance after the adoption / continuance phase. This refers to whether the participants like or dislike performing the behavior (Goodman & Dean 1981) and therefore is about the private acceptance of the behavior. Especially the intrinsic valence will influence the private acceptance as the private acceptance is about whether the change recipient personally accepts the behavior and its direct consequences (Lofquist & Dawis, 1978). Moreover, the intrinsic valence is extra important for institutionalization as the alignment between work values and the individual’s own values, especially on the long term, is expected to be important for workers. When the intrinsic valence is relatively low, moments of reevaluation are expected to increase and other behavior could earlier be preferred compared to the low valued behavior.
These first three aspects (e.g. knowledge, performance, and preference) are on an individual basis. They are about whether people know, do, and prefer the behavior. The last two aspects are about whether there is consensus on the norms and values of the behavior and develop on a collective level (Goodman & Dean, 1981). First, the individual preference towards the behavior is expected to develop into a normative consensus which reflects a collective consensus on the appropriateness of the behavior. When the individual believes that the norm consensus is high he / she will feel the pressure to conform to the behavior (Coleman, 1990 in Knez & Simester, 2001) and will, via the attribution process, also develop more positive beliefs about the behavior (Goodman & Dean, 1981). Moreover, when the individual knows that the new behavior is ‘the way how the work should be done’ less reason to reevaluate the behavior exists (see taken-for-grantedness, P 10).
9
work environment. A consensus on values means that the members agree on certain work values as relatively important (Dose, 1997). Values are slow to change so a high value consensus on values which important for the behavior will result in more stable beliefs regarding the behavior and therefore more institutionalization. Moreover, when the individual believes that all colleagues think that the values of the new behavior are indeed important, less reason to reevaluate the behavior exist (see taken-for-grantedness, P 10).
Beliefs influencing the stability of beliefs. As explained, the process of institutionalization is influenced by the beliefs change recipients have about the new behavior which, via an individual cognitive process influences the intention to continue that behavior. The beliefs play an important role in this process as changing beliefs can influence the intention and therefore the performance of the behavior. Consequently, not only the level of the beliefs are important for institutionalization but also the stability of those beliefs over time (Ajzen & Fishbein, 2000). Therefore it is important to know which factors and beliefs influence the stability of the salient beliefs and the cognitive process. An analysis of these factors and beliefs is given underneath by using the theoretical model about the different cognitive gears of individuals (Louis & Sutton, 1991).
Louis & Sutton (1991) proposed a cycle of individual cognitive processing with different cognitive modes; the automatic mode and the conscious mode. The standard mode of behavior of every individual is the automated mental routine in which behavior is habitualized and the individual will draw on previous developed mental and behavioral schemata’s (Louis & Sutton, 1991). To change that habitualized behavior or to routinize behavior, individuals need to sense conditions for switching to the conscious cognitive mode and conditions to switch back to the automatic cognitive mode which reflect the more institutionalized mode (Louis & Sutton, 1991).
10
So, when the individual performs the new intended behavior the cognitive mode of the individual needs to switch back to the automatic mode resulting in the stability of the beliefs and cognitive process. Therefore it is important to know which factors influence this cognitive switch. In the information technology academic area the institutionalization of behavior and the creation of habits got much attention (Jasperson et al., 2005; Ortiz de Guinea & Markus, 2009). A habit (e.g. routinized behavior) develops when the behavior is frequently performed, when the outcome has a high satisfaction, and when it happens in a relatively stable context (Limayem, Hirt & Cheung, 2007). The frequency has to do with the familiarity of the behavior as it is expected that the more often a behavior is performed the more is known about the behavior. The high satisfaction is comparable with positive behavioral beliefs resulting in a stable positive attitude (Limayem et al., 2007). Moreover, the relationship between strong attitudes and stability of beliefs is often seen as positive (Ajzen & Fishbein, 2000; Ajzen, 2002). This means that the beliefs important for the decision to perform the behavior are still important later in the process and need to be high for institutionalization. The stable context refers to the past where it is expected that when behaviors are performed in stable context less cognitive monitoring is needed (Limayem et al., 2007). This reasoning will also hold for the future as when individuals expect that the outcomes of the specific behavior will be the same in the future, less cognitive monitoring or sense making is expected to be needed resulting in less sense making.
When the change recipient believes to know everything about the consequences of the behavior, the behavior is expected to become more automatic (Limayem et al., 2007). The same argument is made by Kim, Hornung & Rousseau (2011) as they say that when the novelty of the situation decreases also the active sense making decreases. Consequently, Ajzen & Fishbein (2002) say that when motivation to process consequences of behavior is low that the attitude towards the behavior will be earlier automatically activated. Therefore, is can be expected that when the change recipients believe that he / she knows all the outcomes / consequences and believes that those outcomes will not change in the future (taking into account the environment in which the behavior is and will be performed) that the change recipient will switch to the automatic cognitive mode. So, these beliefs are therefore seen as the switching conditions to the automatic mode.
11
Critical Evaluation of the OCRBS and its Underlying Cognitive Process
As the new institutionalization measurement instrument will be based on the beliefs used in the OCRBS and on the individual cognitive process it is important to know what the assumptions of the OCRBS and of the theory of planned behavior are. This evaluation will be combined with the theory on institutionalization in order to start developing the new instrument.
Armenakis et al. (2007) argue that five specific beliefs are important during all three phases of organizational change, thereby assuming that those five beliefs have the same importance in the beginning as at the end of a change process. However, an important aspect of the theory of planned behavior is that many beliefs are created about an object but that only the salient beliefs result in a certain attitude while the salience of beliefs can change over time as well (Fishbein & Ajzen, 1975). Consequently, salient beliefs could gain and lose in explaining power during the change process. Based on these insights, it is doubtful that the five beliefs stated by Armenakis et al. (2007) are of the same importance for the recipients in the beginning, during, and at the end of the change process.
The first belief which is expected to lose explaining power during the change process is discrepancy. Although the discrepancy belief is often stated as important aspect of the change process, it is mostly related to the beginning of the change process where people are expected to be more ready for the change when the need to change is clear (Kotter, 1995; Coch & French, 1948). In the field theory of Lewin (1957) the discrepancy of the former behavior does not play a role anymore in the institutionalization phase where focus is laid on the minimization of the discrepancy of the new behavior which will become a restraining factor for change of the changed behavior. Next to that, it can be assumed that the new behaviors and its consequences are less known by the change recipients in the beginning of the change process compared to the end of the change process resulting in more attention towards the insufficiency of the former behavior then towards the new behavior for de adoption decision. Hence, when the new behavior and its consequences become clearer later in the process, the appropriateness of the new behavior can expected to be more important than the discrepancy belief. Based on this, the explaining power of discrepancy is expected to decrease during the change process. Next to that, the discrepancy belief does also not fit in the theory of planned behavior as it is not about the intended behavior which is seen as important condition for explaining specific behavior (Ajzen & Fishbein, 1977; Fishbein & Ajzen, 1975). Therefore the discrepancy belief is not used in the new instrument.
12
change agents and opinion leaders which are expected to influence whether the change will be embraced (Armenakis et al., 2007). However, normative consensus goes further as it is about how the change is embraced by the whole organization and therefore more influential in the institutionalization phase as norms which should be shared by everyone need to develop (Goodman & Dean, 1981). Yet, this does not mean that principal support does not have any influence anymore but it will influence the degree of believed normative consensus as when principals are less supportive (e.g. not walk the talk) less consensus is expected. Therefore this belief is also not used in the new instrument.
The last belief with is expected to lose explaining power in the institutionalization phase is the belief of extrinsic valence. As argued before, the intrinsic valence is seen as important aspect of institutionalization as that is expected to result in acceptance of the behavior’s intrinsic values. On the other hand, extrinsic valence improves the intention to perform a behavior as people get motivated when offered extrinsic rewards and has a positive influence on the decision to perform behavior according to the rational model (Vroom, 1964, Tolbert & Zucker, 2006). However, motivation to perform or continue certain behavior due to extrinsic rewards does not say much about the real acceptance and valence of the behavior itself. The changing importance of beliefs does not only increase / decrease the salience of the five beliefs of Armenakis et al. (2007) but could also result in different beliefs which have a stronger relation with the institutionalization of changed behavior. Based on the theory about institutionalization it can be expected that the beliefs about norm consensus, value consensus, stability of behavioral consequences in the future, and about the total knowledge of the behavior and behavioral consequences increase in salience regarding the continuance of the new behavior. Next to these beliefs is it expected that the beliefs about appropriateness, efficacy, and intrinsic valence which are used in the OCRBS are still salient for institutionalization. These three beliefs do fit within the framework of Goodman & Dean (1981) as people need to perceive the behavior as the right behavior, need to have the perceived ability to perform the behavior, and need to personally accept the behavior. Moreover, especially the beliefs of appropriateness and efficacy fit very well in the theory of planned behavior. The perceived appropriateness of the behavior is expected to be an important factor for a favorable / unfavorable evaluation of the behavior and the perceived ability to perform the behavior (e.g. efficacy) will have a strong relationship with the perceived ease or difficulty of performing the behavior (e.g. perceived behavioral control) which are both expected to be related to the intention to perform (Ajzen, 1991)
13
measuring the beliefs when used in the institutionalization phase. In the beginning of the change process (e.g. readiness phase), when the behavior and its consequences are less known, it could be appropriate to use more general behavior questions to measure the beliefs. The reason for this is that most of the time no specific behavior is known and in line with the principle of compatibility the questions measuring the general behavior should be more general as well (Ajzen & Fishbein, 1977). As the changed behavior and its consequences become clearer for the change recipient the focus on item level should also be more focused towards the specific behaviors instead of the general change. Secondly, some items in the Armenakis et al’s (2007) questionnaire are based on a comparison of the former situation with the intended situation while the former situation is, just as the arguments regarding the discrepancy belief, expected to be less influential for institutionalization. Also the content of beliefs is different; in the readiness phase the beliefs are expected to be more based on informational beliefs while beliefs during the institutionalization will be more descriptive beliefs (Fishbein & Ajzen, 1975). So, in the beginning, the beliefs are more about what people expect while during the institutionalization the beliefs are more based on their experience. This can have an impact on the factors on item level as all items should be based on evaluative beliefs. Moreover, some attributes (e.g. consequences of the new behavior) used in the items of the OCRBS are also assumed not to change during the change process. In the questionnaire evaluations of specific consequences are asked like performance of the organization, improve operations, level of payment, and feelings of accomplishment. As the questions are more based on the adoption of new behavior and are based on readiness theory, the influence in the continuance decision can be questioned. This is also argued by Ajzen & Fishbein (2000) who say that behavior and therefore intention can be best predicted when dimensions that are salient at the time of the assessment are those that are naturally salient at the time of the behavior. So, it is important to focus on items which are expected to have a relation with institutionalization when developing a questionnaire regarding institutionalization.
14
the ultimate goal of the instrument (Liu & Perrewé, 2005). So although emotions do indeed play a role in the change process, the cognition, especially during the institutionalization phase, is still seen as the most important for the cognitive process and the measurement of emotion does not add much value to the instrument. Therefore, emotion will not be included in the instrument.
VALIDATION STUDY
The American Psychological Association (1995) requires that a quantitative survey instrument has construct validity which follows when meeting the standards of content validity, internal consistency, and criterion related validity. In this section a description is given of the process to meet those standards making use of the scale development process explained by Hinkin (1998). The study will be an exploratory study, therefore is this analysis a preliminary step in the development of a psychometrically sound instrument.
Three different studies have been done; two studies to check for content adequacy and one study for the exploratory factor analysis and to assess the criterion related validity. Moreover, the predictive power of the new instrument, assessed via the criterion related validity, will be compared with the OCBRS in order to assess whether the new instrument indeed predicts institutionalization better.
Step 1: Item Development
The first important step for the development of a new scale is to articulate the construct domains clearly (MacKenzie, Podsakoff & Podsakoff, 2011). The construct domains (e.g. beliefs) are set out underneath and are based on the discussed literature. The first belief is the belief of appropriateness which reflects the belief that the specific changed behavior is currently and in the future the correct behavior for the organization. The belief of appropriateness is also used is the OCRBS. However, the belief of the stability of the behavioral consequences is included in the current appropriateness belief by adding the near future in the definition. By increasing the time horizon of the appropriateness, not only the appropriateness but also the expectations of the future consequences are incorporated which influences the attitude towards the behavior and the stability of the beliefs due to less sense making.
15
The third belief is the belief of intrinsic valence which reflects whether the new behavior is believed to be intrinsically beneficial for the change recipient. This belief is comparable with the valence belief of the OCRBS but the extrinsic valence has been deleted. The fourth belief is the individual´s belief regarding the existence of norm consensus about the appropriateness of the behavior for the organization. The last belief is the individual´s belief about the existence of a value consensus within the organization about the values relevant for the changed behavior. For the value consensus domain it will be important to identify the important work values of the changed behavior to be included in the items.
In order to generate the items the theory of the theoretical framework is used. Moreover, to develop the items, many articles which measured (parts) of the beliefs are reviewed and used (Chen, Gully & Eden, 2001; Bandura, 1986, Verplanken, 2006; Charn, Piliavin, Callero, 1988; Limayen et al., 2007; Aarts, Verplanken & Knippenberg, 1998; Lofquist & Dawis, 1978; van Dam, 2005; Kelly, Janet, Lawrence, & bressfield, 1990). Lastly, the guidelines for developing items have been taken into account (Hinkin, 1998).
Content validity of the items. In order to legitimize the psychometric properties of the measuring instrument the content adequacy of the different domains should be assessed (Schoenfeldt, 1984). The content adequacy test measures the degree to which the measure´s items are a proper sample of the theoretical content domain of a construct (Schriesheim, powers, Scandura, Gardiner & Lankau, 1993). The content validity test suggested by Hinkin (1998) is used. The respondents were provided the construct definitions and were asked to match the items with their corresponding definition or mark it as unclassified. As advised by MacKenzie, Podsakoff & Fetter (1991) a threshold of 80% percent of correct matches (e.g. definition with correct domain) was applied. The items were randomly listed and seven different questionnaires (e.g. items were in a different order) were distributed to minimize a potential bias. Students were chosen as sample as they have sufficient intellectual ability to perform the item rating task while being relatively free of potential biases (Schriesheim et al., 1993). 20 judges is seen as the minimal acceptable number (Hinkin, 1998).
16
Content validity assessment two. A second content validity assessment was done as too many questions had been altered or deleted. The second content validity assessment was send to 30 respondents who did not receive the first content validity test. Only ten respondents returned the survey resulting in a too small amount of respondents. However, as the first content validity assessment was done already, these ten returned surveys were used to make further improvements when mixed answers on an item were given. Of the ten respondents, the average age was 25 years, 60 % was male, and 40 % is currently studying (all at the university). Of the people who are working the average working experience is 2 years and they all finished the university.
In this second content validity assessment nine items were, next to the 16 already validated items, validated by meeting the threshold of 80% correctness. For further improvement only the items which did get not validated by both tests were judged. Items 23, 28, 30, and 31 did not meet the threshold. However, respondents mentioned that they had problems understanding these abstract questions, especially since they were in English. Therefore it is expected that when the questions are translated into Dutch and when the organizational goal and important work values of the new behavior will be inserted into the questions that the questions are easier to understand and that the questions will have content validity. Therefore are these questions kept. Ultimately, six items did not meet the threshold and were altered to fit the content domain better (see appendix 1 for the final questions).
The items were after the validation translated into Dutch. The translation was checked by a graduated communication student (Bachelor) for a potential language bias (Bouckenooghe, Davons & van den Broeck, 2009). After the acceptance of the Dutch questions they were adjusted to fit the local context. This adjustment was done based on three conversations with managers who have been present before and during the change. Subsequently, questionnaires were sent to three advisors of the sample group and one team leader to check whether they understood the survey and to find out how long it would take to fill it in. The feedback they gave resulted in slight changes in the questions (without changing the measurement goal).
Step 2: Questionnaire Administration
Organizational context. The Vodafone Customer Management department is Vodafone’s customer contact center. Customers can call the employees of Vodafone for all their questions regarding their subscription or potential subscriptions.
17
of the new behavior showing evidence that the institutionalization phase is not over yet. So, the sample is very useful for this research as the change program is still in its institutionalization phase and clearly after the adoption phase. Totally 762 advisors were working in the Customer Management department when the survey was set out of which 362 did already work in the department before the behavioral change was introduced.
Old behavior: The old behavior was characterized by clear operating rules and clear procedures were
present about how to deal with the customer. The solutions for problems were already set for the advisors resulting in clarity for them and he results were measured via clear indicators like length of calls.
New behavior: The new behavior is characterized by many opportunities for the advisor to find solutions for the customer. Therefore, it is expected from the advisors that they are very assertive towards the customer. The operating procedures are now guidelines about how a customer could be helped which results in much freedom and responsibility for the advisor. The results are measured via customer evaluations which are seen as more difficult to interpret and less clarity exists about how to influence those indicators.
Sample. The sample taken for the analysis consisted of all people who already worked for the Vodafone Customer Management before the NPS heartbeat was introduced. The total group was 362 of which all were asked to fill in the questionnaire and the sample was taken in June 2012 (18 months after the introduction).
Procedure. The 33 questions of the new instrument were all asked using a 5 point Likert-scale (strongly agree - strongly disagree) and were asked in a random order. Solicitation to participate was send to the 362 employees and to motivate the employees to participate, a price was promised to one of the respondents who filled in all questions. In total, 213 employees (58,9 %) responded of which 148 filled in all questions (37,3 %). The high amount of people who did start but did not finish the questionnaire was chiefly the consequence of a failed internet connection which caused people to stop filling in the questionnaire. As this happened randomly the high amount of not finished questionnaires was not expected to influence the data. For an exploratory factor analysis a sample size of 150 observations should be sufficient, therefore the sample for the factor analysis (N = 148 (e.g. 40,9 %)) is big enough (Guadagnoli & Velicer, 1988 in Hinkin, 1998).
18
followed by HBO (29,7 %), VMBO (7,4), and WO (7,4 %), VWO (2,7%), and HAVO (2%). Lastly, all employees work at the same layer in the organization and worked in 32 different teams which is a good reflection of the department.
Step 3: Initial Item Reduction
The first step in analyzing the data was to assess the variance of each of the 33 items. Items which do not show enough variance are not useful for an instrument and should be removed from the instrument (Clark & Watson, 1995). However, no clear cutoff score exists for the level of variance while in some articles a Standard Deviation of 1.0 as cutoff score has been used (Armenakis et al., 2007; Liden & Masylin, 1998). Of all items only two had a standard deviation higher than 1.0. However, as no clear cutoff score exists no items are deleted.
After the variance assessment, the intercorrelations matrix between items within domains was analyzed (appendix 3). The items should have intercorrelations above ,4 (Hinkin, 1998; Armenakis et al., 2007; Kim & Mueller, 1978). Five of the efficacy items and one norm consensus item did not survive this test (resp. 7, 8, 10, 14, 15, and 27). Due to the potential deletion of items 7, 8, and 10 the future perspective disappeared in the efficacy belief while the future perspective is theoretically important, therefore item 10 is kept. Items 7 and 8 cover the knowledge aspect and partly the future aspect within the efficacy belief but these items are deleted since those aspects are covered by other items measuring the belief’s domain. Items 14 and 15 have very low intercorrelations. A reason for the low intercorrelation could be that before the change is introduced, when the change recipients do not have much knowledge yet about the execution of that behavior, that the efficacy belief regarding the behavior is more or less in line with the perceived influence of the company on the execution. When the changed behavior is introduced the perceived influence of the change recipient on the execution is probably bigger as respondents learned more about the behavior, probably decreasing the role of the company which creates the difference between the concepts. As the intercorrelations are too low and the belief regarding own efficacy and influence of the company disperse from each other these items are deleted. The deletion of item 27 does not have an impact on the intent of the norm consensus belief as the purpose of that domain is covered by other questions as well and is therefore deleted.
19
Successively the factor analysis can be done, which allows the reduction of the set of observed items to a smaller set of items. The exploratory principle axis factor analysis is suitable to analyze whether the scales are independent enough from each other (Hinkin, 1998). Much discussion exists about which rotation to use for the analysis (Reise, Comrey & Waller, 2000). Spector (1992) says that the principal component analysis with an orthogonal rotation is well accepted for scale construction and therefore this rotation is executed. Based on the eigenvalue and scree-test a three factor solution was retrieved. However, none of the items did clearly load on the third factor making that factor not useful (Hinkin, 1998). Therefore, a two factor solution rotation was executed. The two factors account for 58,2 % of the variance, have an eigenvalue of 14, 1 and 2,1, and meet the scree plot test. As this two factor solution was not expected also an exploratory factor analysis with an Oblique rotation (e.g. direct oblimin) was executed. However, also with the oblique rotation the two factors with the same items were extracted and as the item distribution over the two factors was comparable with the orthogonal rotation. As the orthogonal rotation is easier to interpret this rotation is used for the factor analysis (Gorsuch, 1983, p197). However, as many cross-loadings were present it can be expected that both factors are to a certain degree related to each other.
This two factor model was unfortunately not expected based on the theory. However, three reasons exist why this two factor model can be used. Firstly, no empirical analysis of the role of beliefs in institutionalization has been done before, therefore no clarity does exist about the validity of the five factors. Moreover, the theory of Goodman & Dean (1981) makes a separation between the individual and collective processes which is reflected in the two different factors which could be a sign that the two factor model suits the reality. Lastly, an exploratory factor analysis was done for the OCBRS as well which resulted in a comparable construct (intrinsic valence, appropriateness, and efficacy are analyzed as being one factor). As no empirical research has been done of the several beliefs during institutionalization it can be the case that the beliefs play a different role in the institutionalization phase. For these three reasons, the two factor model is used for further refinement by decreasing the number of items.
20
Based on these conditions, items 3, 5, 10, and 25 are deleted. Unfortunately has item 10 a factor loading of ,205. Although the item was kept after the correlation matrix for theoretical reasons, the factor loading is too much below the cut off score to be kept. The deletion of the other three items does not impact the domains of the beliefs and are therefore deleted. Factor 1 now consists of 14 items. Two original items measuring valence with the lowest loading are deleted since six out of the fourteen items originally measured valence which therefore gives too much weight to that belief (e.g. 18 and 21 are deleted). For the same reason, two norm consensus beliefs are deleted as well (e.g. 30 and 31). A new rotation with the new items shows that items 13 and 23 are under the cut of score of ,6. However, to balance the items of the original beliefs in the current factors the items are kept. So, after the factor analysis the new instrument consists of 20 items divided by two factors (table 2). The two factors account for 63,29 % of the variance and the factors had an eigenvalue of 10,71 and 1,95 for factor 1 and 2 respectively. The first factor is a combination of items originally included in the appropriateness, valence and efficacy beliefs. This belief is called the individual belief as all those items measure what the individual believes about the behavior, execution, and the consequences of that specific behavior. The second factor consists of the items originally included in the norm consensus and value consensus beliefs and is labeled perceived
group consensus belief (PGCB) as all items included measure whether there is perceived consensus about
the appropriateness of the behavior and importance of the new behavior’s work values.
TABLE 2
21
TABLE 2
Results of Factor Analysisa
VALUE 29 ,386 ,652
VALUE 32 ,253 ,753
VALUE 33 ,291 ,768
Note: N = 148. APP = Originally Appropriateness; EFF = Originally Efficacy; VAL = Originally Valence; NORM = Originally Norm Consensus; VALUECON = Originally Value consensus A. Item numbers based on original questionnaire
Reliability is an important condition when developing an instrument with validity and can be checked using Cronbach’s alpha (Hinkin, 1998). The coefficient alpha for the Individual belief factor was ,95 and for the PGCB factor ,899. The suggested standard is ,7 and therefore both factors are internal consistent (Nunnally, 1978).
Step 4: Scale Evaluation
Hinkin (1998) recommends that the validity of scales should be evaluated beyond the factor analysis. Therefore, the criterion related validity will be tested. Criterion related validity reflects the success of the measure for prediction or estimation (Cooper & Schindler, 2006). If the statistical relationship is significant, evidence of criterion related validity is provided. Moreover, the variance explained by the new instrument of the criterion related validity measures will compared with the OCRBS’ explained variance of the related measures to assess whether the new measure is better able to gauge the process of institutionalization. Five different scales of related criteria and the items of the OCRBS have been administered along with the items of the new instrument. The related criteria are emotion (negative and positive), habit, organizational cynicism, and continuance intention.
22
(e.g. positive and negative) and measures to which extent the respondents experienced a specific emotion. As this measure can measure emotions retrospectively, is used often in the social science, and is related to work related emotions this measure is used to measure emotion (Fisher, 2000; Dasborough et al., 2008). The second related scale is the scale of habit. A habit is defined as “learned sequences of acts that have become automatic responses to specific cues, and are functional in obtaining certain goals or end-states” (Verplanken & Aarts, 1999, p. 104).When a certain behavior becomes a habit the individual will keep on performing the behavior without active sense making which reflects a stable intention to perform the behavior, only totally different situations will result in potential alternations in the cognition of the individual (Jasperson et al., 2005). So, a habit is comparable with the automatic cognitive state of taken-for-grantedness and is therefore expected to relate to institutionalization. Limayem, Hirt & Cheung (2003) have developed a scale of six items for measuring a habit in the context of the introduction of information systems. The usage of information systems is just as in this paper focused on the changed behavior of the change recipient and is therefore used as criterion related validity measure, although altered to the circumstances.
The third related scale is the scale of organizational cynicism. According to Vance, Brooks & Tesluk (1996), cynicism would breed suspicion towards the motives of the change agents, apathy towards those efforts, and perhaps even sabotage. Therefore the relation of organizational cynicism and institutionalization is expected to exist in two different ways. In the first place can it be expected that when the change is successfully institutionalized that the employees are not feeling (or at least have less) cynicism towards the organization as the last big change program was a success. Secondly, low cynicism after the change program which had a big impact on the change recipients will indicate that the change recipients still believe in the good motives of change agents and will behave positively towards the former introduced change indicating institutionalization. Moreover, also Armenakis et al. (2007) used organizational cynicism as criterion related validity measure which was in their analysis significantly related to the beliefs. The scale with six items of Atwater, Waldman, Atwater & Cartier (2000) is used to measure organizational cynicism.
The last criterion related validity measure is about the intention to continue performing the behavior. As institutionalization is about the (automatic) decision to continue performing the behavior this measure is the natural indicator of institutionalization. In the often cited article of Bhattacherjee (2001) a scale consisting of three items measuring the continuance intention for the use of information systems is developed which is also used in this research.
23
cynicism is an attitudinal reaction and a habit is a behavioral reaction. Next to that, the measure of usage continuance is a measure for the intention to perform a certain behavior.
Armenakis et al. (2007) also used normative and affective commitment as criterion related validity measures. They argue that commitment would be the natural indicator of institutionalization (Armenakis et al., 2007; Armenakis et al., 1999). However, when looking at the items used by Armenakis et al. (2007) to measure affective and normative commitment (taken from Herscovitch & Meyer, 2002) then the items more or less reflect the same aspects as the five beliefs in the OCRBS and the new instrument. As this does not have any added value the scale of commitment was not included in this paper.
Results. Coefficient alpha for the positive emotion was ,949 and for the negative emotions ,856. For habit, organizational cynicism, and continuance intentions the Coefficient alphas are respectively ,939; ,665, and ,792. So, only the Alpha coefficient of organizational cynicism does not meet the threshold of ,7. However, since this is an exploratory research the scale was included in the research.
Simple regression analyses with the averages of every factor were used to determine whether the instruments could predict a significant amount of variance in the five criterion variables (Hinkin, 1998).
TABLE 3
Explained variance, intercorrelation, and significance to establish criterion related validity
New two-factor instrument* OCBRS*
R2 β significance R2 Habita ,703 ,743 (Ind.)c ,132 (PGCB)d P < ,001 P < ,05 ,664
Organizational cynicism a ,289 -,516 (Ind.) -,031 (PGCB)
P < ,001 P > ,05
,382
Continuance intention a ,630 ,855 (Ind.) -,095 (PGCB)
P < ,001 P > ,05
,644 Negative emotion b ,221 -,438 (Ind.)
-,045 (PGCB)
P < ,001 P > ,05
,252 Positive emotion b ,543 ,601 (Ind.)
,183 (PGCB) P < ,001 P < ,05 ,559 Note: a N = 148; b N = 134; c = Individual belief; d
= Perceived group consensus belief; * all significance levels of the whole regression analysis were P < ,001
The new instrument with two factors does significantly explain the variance of all five related criteria so criterion related validity has been established. Also the original OCRBS does explain the variance of all five criteria significantly (table 3).
24
An interesting point which can be seen in this analysis is that the difference between the β of the individual belief and the PGCB. The β of the individual belief is considerably higher than the PGCB. Moreover, the PGCB belief does in this model not have a significant relationship with organizational cynicism, intention, and negative emotion. In order to understand the separate influence of the two beliefs on institutionalization better and to go more into depth regarding the different roles of the individual belief and the PGCB an extra test has been conducted. This test does generate extra information about the beliefs to understand the functioning of the questionnaires better which can help when improving the new questionnaire further. Moreover, these tests can also generate extra information to create new research opportunities regarding the role of beliefs in the institutionalization process.
The analysis is a duplication of the criterion-related validity assessment but now with the two separate factors of the new instrument. Based on the theory, it was expected that both factors have a significant relationship with the criterion related measures and explain a big amount of variance (table 4).
TABLE 4
Explained variance, intercorrelation, and significance to establish criterion related validity
Individual Belief PGCB R2 β significance R2 β significance Habita ,694 ,833 P < ,001 ,405 ,637 P < ,001 Organizational cynicisma ,288 -.537 P < ,001 ,145 -,381 P < ,001 Continuance intentiona ,625 ,791 P < ,001 ,236 ,486 P < ,001 Negative emotionb ,220 -,469 P < ,001 ,116 -,340 P < ,001 Positive emotion b ,525 ,724 P < ,001 ,346 ,588 P < ,001 Note: a N = 148; b N = 134
25
The last analysis is done to generate some more specific insights concerning the beliefs which were expected to lose explaining power during institutionalization. These three beliefs (e.g. discrepancy, extrinsic valence (only one item), and principal support) were assessed for their predictive power of the criterion related validity measures.
TABLE 5
Explained variance, intercorrelation, and significance to establish criterion related validity Discrepancy Extrinsic valence Principal support
R2 β significance R2 β significance R2 β significance Habita ,095 ,473 P < ,001 ,044 ,210 P < ,05 ,396 ,629 P < ,001 Organizational cynicisma ,053 -,230 P < ,01 ,019 -,139 P < ,1 ,253 -,503 P < ,001 Continuance intentiona ,234 ,484 P < ,001 ,032 ,178 P < ,05 .424 .651 P < ,001 Negative emotionb ,028 -,168 P < ,1 ,069 -,262 P < ,005 ,163 -,404 P < ,001 Positive emotionb ,169 ,411 P < ,001 ,085 ,292 P < ,005 ,445 ,411 P < ,001 Note: a N = 148; b N = 134
The results show that most beliefs have significant relationship with the criterion related measures (e.g. except for discrepancy with negative emotion and extrinsic valence with organizational cynicism) table 5). However, the beliefs of discrepancy and extrinsic valence do not explain much variance of the criterion related measures while principal support does explain quite some variance.
DISCUSSION
26
stability of behavioral consequences in the future and knowledge of the behavior and behavioral consequences.
Comparison With The OCRBS (Armenakis et al., 2007)
The new two factor instrument (e.g. individual belief and the PGCB) does explain for four out of five criterion related validity measures less variance compared to the OCRBS five-factor model. The differences regarding the explained variance between the two models are, except for organizational cynicism, not very big. That the OCRBS explains almost 10% more variance for organizational cynicism compared to the new instrument can be the result of the belief of principal support which is included in the OCRBS. Since organizational cynicism partly comprises of the belief that the organization lacks integrity (Dean, Brandes & Dhanvadkar, 1998) it can indeed be expected that when the change recipients do not perceive that the leaders / managers / respected peers show support for the change that the change recipients start to question the integrity of the organization resulting in increased cynicism towards the organization.
As the OCRBS does explain more variance for most of the criterion related measures, the expectation that the new instrument would be better able to predict institutionalization was not correct. On the other side, as every factor does add extra explained variance to the model it could have been expected that a model with five factors (e.g. OCRBS), of which all factors were assumed to be important for institutionalization, explains more variance compared to a model with only two factors. Next to that, when only the individual belief is used to explain the variance of the related criteria it still explains more variance for habit while the prediction of the other related measures is only slightly less than the five factor OCRBS model. This does not only show the importance of the individual belief but also confirms that not all beliefs of the OCRBS are as important as expected for the institutionalization of changed behavior.
Role Of Beliefs During Institutionalization
27
theory which states that a change recipient starts an attributional process to explain why others value and
perform the behavior which subsequently influences his / her own beliefs and consequently the performance of the behavior (Goodman & Dean, 1981).
Next to the likely positive relationship between the two beliefs it is also possible that the PGCB has a stabilizing effect on the individual belief due to the process of social comparison (Goodman & Dean, 1981). Change recipients validate and confirm their beliefs by comparing them with the behavior and beliefs of other change recipients. When their own beliefs are confirmed the individual will earlier assume the beliefs to be true without further verification / less questioning resulting in the development of a certain degree of taken-for-grantedness (Festinger, 1954). This possible positive influence of the PGCB on the level and stability of the individual belief means that that PGCB and the process of perceived consensus development are still important for institutionalization. However, as the individual belief in this research is the most important belief for explaining variance, the perceived norm and value consensus do in this study not reflect institutionalization as argued by Goodman & Dean (1981). Institutionalization can possibly also occur with a relatively low level of perceived group consensus, which is therefore according to this research probably just one of the factors influencing institutionalization. Yet, it is also possible that the ambiguous and small direct role of the PCGB is a result of difficulties measuring institutionalization on a unconscious level. The criterion related measures measure the known (e.g. explicit) beliefs, emotions, and intentions of the change recipients while the concepts of taken-for-grantedness and of automatic decisions are very much based on unconscious and implicit beliefs. If the PGCB has indeed influence on the stability of the individual beliefs and on the concept of taken-for-grantedness while the criterion related measures are not (totally) able to measure these concepts the role of the PGCB will be underestimated.
28
In this exploratory research the five beliefs could not be separated from each other. The beliefs of appropriateness, efficacy, and intrinsic valence are perceived as one belief and also the norm consensus and value consensus beliefs were perceived as one belief by the change recipients. However, at least four of all original five beliefs are included in the two beliefs which both have a significant relationship with the criterion related measures showing initial evidence that the original five beliefs do all have some importance in the institutionalization process. That the respondents perceived the five beliefs as two constructs was not expected but can be explained by the framework of Goodman & Dean (1981). They argue that the constructs are analytically independent but they also argue that the constructs could be developmentally unidimensional as they represent a process. That could mean that the several beliefs would be analytically independent in the beginning of the change process while evolving later in the change process into a more coherent construct.
That the beliefs of appropriateness, efficacy, and valence become unidimensional can be the result of the change recipient’s tendency to decrease the cognitive load later in the change process causing
taken-for-grantedness (Schein, 1992). During the introduction of new behavior change recipients evaluate every
belief separately to judge the behavior while during institutionalization the evaluation is simplified by combining several original separate beliefs, possibly reflecting a certain degree of automaticity as well. Next to that, it can be expected that when appropriateness, efficacy, and valence are perceived as one construct that any (potential) fluctuation of one of these beliefs will be minimized due to the influence of other two beliefs resulting in less fluctuation of the former belief. Therefore the unidimensionality could reflect taken-for-grantedness as well as institutionalization as when the three beliefs are perceived as one construct only situations which significantly alter the cognitions of the change recipients could change their cognitive script and therefore their intention (Jasperson et al., 2005). However, it is important that the unidimensionality is combined with strong positive beliefs to be able to speak of high institutionalization as positive beliefs without stability or high stability but negative beliefs will probably not result in institutionalization.
That the individual belief and the PGCB are still different constructs in this research could therefore mean that the institutionalization is in this case not that far (yet) that the individual belief and PGCB consensus have become one construct. When the individual belief and the PGCB tend to become one construct as well it could be a reflection of the existence of strong norms as the personal beliefs are in line with the perceived consensus on the appropriateness and valuation of the new behavior. However, no empirical information exists whether the two constructs tend to become one construct when institutionalization is high.
29
This research and the new instrument do have added value compared to the existing instruments and knowledge of institutionalization. The new instrument is in the first place very short while still explaining much variance. As the new instrument with only the individual belief consists of twelve items it will be possible to combine the new instrument with other interesting instruments. Moreover, although PGCB does in combination with the individual belief not explain much extra variance, it can still be included in the instrument in order to judge how far the process of institutionalization moved forward and to test whether the PCGB can be improved to increase and stabilize the individual belief.
Another added value is that the new instrument is the only instrument which is solely focused on institutionalization while for example the OCRBS is focused on the whole change process making it less effective as institutionalization is different than the readiness and adoption phase. Moreover, this instrument can be used to evaluate and assess other instruments which also want to measure (aspects of) institutionalization as it has showed to explain a relatively big variances of five related measures of institutionalization. A last added value of the new instrument is that it can probably be used to predict institutionalization of behavior for new employees as well. The new employees who did not experience the old behavior will be able to fill in the questions as the new instrument is focused on only the new behavior and on situation of this moment or in the future. As Goodman & Dean (1981) mentioned, the socialization of new members is important as bad socialized new members could cause de-institutionalization. By using this instrument it can be predicted whether the behavior is institutionalized for the new employees as well. Moreover, the PCGB plays possibly a more important role for new employees as new people could feel more pressure to conform to their perceived norm and value consensus.
Limitations
30
indeed predictive validity, it does not necessarily imply a causal relationship (DeVellis, 2003). The goal of the exploratory analysis was to start developing an instrument and to generate insights concerning institutionalization which can open new research possibilities. Therefore, no hard conclusions can be drawn regarding all insights generated during the analysis. These insights needs to be dealt with as possible explanations and therefore need some thorough attention to assess the legitimacy.
New research possibilities
As the several insights gained during the research are interesting and could explain the process and role of beliefs regarding institutionalization much new research possibilities can be extracted from this paper. In the first place is it interesting to research the role of the PGCB in the process of institutionalization. It could impact institutionalization directly, but can also do this via the positive relationship with the individual belief and via the development of the stability of the beliefs (e.g. taken-for-grantedness). An important step in discovering the role of the PGCB is to assess whether the criterion related validity measures do indeed measure all aspects of institutionalization including the unconscious aspects since the role of the PGCB could be bigger when the measures do not measure all unconscious aspects.
The role of the beliefs about stability of behavioral consequences in the future, and about the total knowledge of the behavior and behavioral consequences remains unclear after this research. Only four questions of the new instrument include parts referring to these beliefs. Two items measuring appropriateness include the future perspective and of items measuring efficacy is one focused on knowledge and one on the stability of the behavior in the future. This does not say that the beliefs are not salient but that their role in this questionnaire is minimal. The fact that the other questions including parts of these beliefs were deleted due to low intercorrelations and low factor loadings suggests that these concepts are not (much) related to appropriateness and efficacy. Therefore is recommended to research the role of these beliefs separately to assess their role in the process of institutionalization.
Furthermore, in the theory section it was stated that different attributes will be salient in the individual cognitive process during institutionalization, decreasing the importance of the attributes used in the OCRBS. However, not much is known about the important attributes in the institutionalization phase. It is interesting to research the important attributes as change agents can focus on those attributes when improving institutionalization and the questionnaire can be improved by inserting those attributes in the items.
