University of Groningen Career roles de Jong, Nicole

(1)

Career roles

de Jong, Nicole

DOI:

10.33612/diss.101440931

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

de Jong, N. (2019). Career roles: their Measurements, Determinants and Consequences. Rijksuniversiteit Groningen. https://doi.org/10.33612/diss.101440931

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

536582-L-sub01-bw-Jong 536582-L-sub01-bw-Jong 536582-L-sub01-bw-Jong 536582-L-sub01-bw-Jong Processed on: 16-10-2019 Processed on: 16-10-2019 Processed on: 16-10-2019

Processed on: 16-10-2019 PDF page: 19PDF page: 19PDF page: 19PDF page: 19 19

Measuring Identification

with Career Roles

CRIQ: An Innovative Measure using Comparison

Awareness to avoid Self-presentation Tactics*

2

The chapter presents a new measure for career role identification, the Career Role Identification Questionnaire (CRIQ). In constructing the CRIQ, we used the Comparison Awareness Inducing Technique (CAIT), a new and innovative method to reduce the effects of self-presentation tactics. The results show that the CRIQ measures identification with the six career roles conceptualized by Hoekstra (2011). Furthermore, the inventory has reliable scales and a clear factorial structure. The CAIT receives some support as a new way to deal with the problem of social desirability in self-report measures as this technique is thought to induce comparison awareness and thus suppress various response tendencies. Further limitations and implications are discussed.

* Chapter based on De Jong, N., Van Leeuwen, R. G. J., Hoekstra, H. A., & Van der Zee, K. I. (2014). Measuring identification with career roles. CRIQ: An innovative measure using comparison awareness to avoid self-presentation tactics. Journal

(3)

In a dynamic and complex work environment, the nature of jobs is changing. While jobs used to be described in terms of a set of fixed tasks that had to be performed by one person, today tasks are subject to rapid changes and development (Arthur, Khapova, & Wilderom, 2005). As a result, the dominant view on career development changed in such a way that jobs can better be described in terms of a set of roles (Hoekstra, 2011; Parker, 2007). With the changing nature of jobs, the view on careers also changes towards a perspective that is less focused on the sequence of jobs, but more on combinations and sequences of different roles in one’s work. The development of practices and instruments that address the changes in the occupational landscape has just begun. Recently, a model was proposed that specifies the nature and development of career roles (Hoekstra, 2011). Career roles refer to the stable and repetitive patterns in one’s way of functioning; attributed on the basis of perceived reality of functioning, not on assigned job titles. Workers may perceive their own career roles differently than their environment. Furthermore, persons may identify with career roles not conquered yet, as much as with the roles they believe to fulfill. In these perceptions the dynamics of careers are formed.

The purpose of this chapter is twofold. First, we present a new measure of role identification, the Career Role Identification Questionnaire (CRIQ). Second, we apply the Comparison Awareness Inducing Technique (CAIT) to reduce the effects of self-presentation tactics. We will first explain the construct of career roles and role acquisition processes, and then discuss career role identification. After describing the development of the CRIQ as a measure of career roles identification, and the new item-presentation technique CAIT that was used, we report a study of the psychometric properties of the CRIQ.

Roles in Life and Career

People play a variety of roles in their lives; some existent from childhood, others appear as people mature. In this regard, one of the pioneers in this area of research, Super (1980) distinguished nine major life roles (e.g., child, student, worker, partner, parent, citizen, homemaker, leisurite and pensioner). Each of these life roles may play an important part in people’s lives across different domains and settings (e.g., home, school, work and community). A career in this sense is a constellation of interacting, varying and changing life-roles. Conceptualizing a career as such, career development can be described in terms of a sequence of positions and occupations (Super, 1957). Overall, a more diverse role repertoire may result in greater satisfaction than simultaneously playing very similar roles (Super, 1980, 1990).

The notion of holding multiple roles is important, not only for life-roles in general, but also within today’s complex work environment. Nowadays there is increased room for job crafting within many jobs, affecting the meaning of one’s work and allowing workers to see beyond the boundaries of their job description (Wrzesniewski & Dutton,

(4)

2001). As a result, role boundaries and perceptions change; allowing employees to broaden their role orientation and become more flexible (Parker, 2000). How people view their job in terms of roles can have a great influence on their job satisfaction and performance (Parker, 2007; Porter, Lawler, & Hackman, 1975). First, flexibility in terms of role breadth can influence one’s job performance, career potential and career success (Briscoe & Hall, 2006; Morrison, 1994; Parker et al., 1997). Second, experience in a variety of functional experiences (e.g., diversity of roles) has been associated positively with promotion, salary level, and overall positive affect (Campion, Cheraskin & Stevens, 1994).

Table 1 The Career Roles Model: Six Career Roles

Dominant Personal Motives

Organizational Performance Domains Exploitation

Production, Results Innovation, ChangeExploration

Distinction motives

Autonomy / Agency

Self – assertion Maker Expert

Integration motives

Connectedness, Belonging,

Cooperation, Sharing Presenter Guide

Structure motives

Collective meaning, Cohesion,

Institutional structure Director Inspirer

Recently (Hoekstra, 2011) developed a model of 6 universal career roles (see Table 1). The model assumes that over time enacted work roles may grow into stable and enduring career roles (Hall, 1976; Hoekstra, 2011). Hoekstra defines a career role independent of jobs and functioning level, as ‘a coherent and enduring set of characteristics of the perceived effects of the way a person is doing his or her work’ (Hoekstra, 2011, p. 164). Hoekstra based his model on the systematic combination of three classes of individual motives (distinction, integration and structure, derived from Hogan, 2007) and two essential organizational themes (exploitation and exploration, derived from March, 1999). First, distinction refers to dominant personal motives of autonomy, agency, and self-assertion; integration refers to motives of connectedness, belonging, cooperation, and sharing; structure refers to motives of collective meaning, cohesion, purpose, institutional structure. Second, exploitation refers to performance directed at production and results, exploration is aimed at innovation and change. Combining both individual and organizational motives, the Career Roles Model distinguishes between six career roles, each of which typically serves a certain group of motives, characterized by the combination of an individual and an organizational theme. Hoekstra proposes the Maker, Expert, Guide, Presenter, Director and Inspirer role as building blocks for career

(5)

development (Hoekstra, 2011). The Maker role involves direct results and making things happen. The Expert role involves questions and problems rather than direct results. The Presenter role involves influencing others. The Guide role involves helping and guiding others. The Director role involves the use of means and resources of the current collective to the best possible use in the long run. The Inspirer role involves exploration of possibilities for change and innovation.

Role Acquisition Processes

Role acquisition is, to a large extent driven by the processes of role taking and role pressure. Both processes drive actual fulfillment of career roles, but are also involved in identifying with those roles. For most people, the first career roles that develop will result from the expectations and demands of others, by a process called role pressure. Later on, people may increasingly work their way towards roles they prefer for themselves, a process we denote as role taking (see Wille, Beyers, & De Fruyt, 2012).

Individual, organizational and external motives may all play an important part in a person’s role acquisition. When role pressure is very high, presumably there is not much room for changes guided by identification with other than actual roles. When role pressure is moderate, identification with certain roles will determine a person’s role-taking ambitions and behaviors. Good performance in turn will enhance the possibilities to work one’s way towards preferred roles. In sum, as people grow in their work and become more confident and autonomous, although still bound to external pressure on the job, the process of selecting tasks and roles ideally would become a more conscious decision, reinforcing and differentiating career role identification, and hence also role taking (Crant, 2000; Parker, 2007).

An instrument to measure the six career roles just described has recently been published (Hoekstra, 2011). In Hoekstra’s conception of career roles, self-perceived and other-perceived fulfillment of roles, as well as the identification with career roles are all important elements of career dynamics. Too much discrepancy between career role enactment and career role identification may have negative effects on job performance, job satisfaction, and psychological well-being (Fugate & Kinicki, 2008). Subsequently, this can have major consequences for career role enactment in terms of motivation or performance (Hall, 2002). Therefore, the identification with each of these career roles was the measurement target for the CRIQ instrument described next.

Career Role Identification

Differences between role taking and role pressure acquire additional meaning when looking at the way people relate to and identify with their own roles. Enacted roles will not always be what a person identifies with and aspired roles may in reality be out of reach for all sorts of reasons. By career role enactment we mean the observed activities of a person in a job and their effects, fitting a specific role. However, people may feel

(6)

attracted to, or strive for, roles that are not frequently (or not at all) enacted. In doing so, people can relate to career roles in multiple private ways as well.

Specifically, we define career role identification as ‘the mental act of identifying the career role as part of the self’. This can be done by elaborate self-construction (Holland, 1997), unreflected commitment (Brisbin & Savickas, 1994) or by external pressure. Career role identification is not publicly observable by definition, but a private event in the person’s mind. This ‘private’ career role identification can be a conscious identification as well as an unconscious preference. For instance, a person may aspire to work in the Director role and thus have a preference for strategic leadership, also when this position is still out of reach effectively. Or, one might unconsciously long for a role as Presenter, attracting attention and having influence on others, without knowledge of such a preference.

Measuring Career Role Identification

The purpose of this study was to develop a measure for career role identification, the Career Role Identification Questionnaire (CRIQ). Because career role identification is such a private matter, a reasonable way to access one’s identification with career roles is through the use of self-report measures. One of the challenges in constructing such measures is to prevent self-serving motives dominating the responses (Robins & John, 1997). As previous research has shown, a high number of people tend to identify to a considerable extent with all career roles, possibly fearing to present a narrow profile that might reduce career chances (Hoekstra & Groen, 2008).

Self-presentation Tactics

People tend to enhance one or more self-components or defend themselves against negative views using serving motives such as enhancement and self-protection (Alicke & Sedikides, 2009). Furthermore, self-serving motives enable a form of damage control and operate when a part of the self is threatened to drop below a tolerated point. Under influence of both self-enhancement and self-protection processes, people can (consciously as well as unconsciously) adopt self-presentation tactics while answering to self-report measures.

An important self-presentation tactic is socially desirable responding, which is responding in such a way to portray oneself as appealing. This response set includes faking answers and impression management (Paulhus, 1991). The tendency towards socially desirable responding is even more likely to occur when the evaluation is perceived as important to one’s self-concept and global self-esteem (Alicke & Sedikides, 2011; Paulhus et al 2003). As career roles are situated in a work context, and respondents are likely to feel evaluated in their work, the tendency to respond in a socially desirable way must be considered very likely.

(7)

The Influence of Self-presentation Tactics

There are a number of ways in which self-presentation tactics can influence the rating process and lead to biases in self-reports. First, self-presentation motives may reflect roles that people aim, or hope for (Alicke & Sedikides, 2009; Paulhus, 1991). Second, identification with a career role may (un-)consciously be influenced by the importance that a role has been given in the employment context in such a way that a general base level is required by the company. This is, for example, the case for health care institutions where the Guide-role is prevalent. Third, identification with an already acquired role may unconsciously enhance identification-tendencies as one is already familiar with that role. Related, general attractiveness of a career role may influence personal attractivity ratings above personally felt identification. For instance, ratings of the Director role will presumably seldom be very low among management, because such scores might reflect a lack of ambition which is socially undesirable for them. In the rating process, a consequence can be that the self-presenting person will focus more on the image of the role than on one’s own feelings about the role. As such, self-presentation can be used if employees have specific goals in portraying themselves or wish to impress others. Self-report measures include a self-evaluative component; therefore numerous individual differences possibly play a role in self-presentation biases such as narcissism, self-concept certainty or clarity and self-handicapping (Morf, Horvath, & Torchetti, 2011; Paulhus & John, 1998; Sedikides & Gregg, 2008; Sedikides & Strube, 1997).

Minimizing the Effects of Self-presentation Tactics

Overall, Likert type scales are the dominant measurement tool in questionnaires (Ogden & Lo, 2011). To minimize self-presentation tactics and avoid response style problems, use of other types of scales, such as the forced-choice, ipsative measures have been suggested. This technique results in Forced Choice Ipsative Data (FCID) (Edwards, 1953). Ipsative data however have the clear disadvantage of implying scale interdependence. Scales are no longer measured independently (Meade, 2004), which poses serious problems when analyzing the scales, especially with multivariate techniques (e.g., Baron, 1996; Cornwell & Dunlap, 1994).

To overcome problems of both ipsative and traditional Likert scales and still reap some of the intended advantages of the choice paradigm, we developed a new item format designated the Comparison Awareness Inducing Technique (CAIT). Applying the CAIT people are free to respond on the items as they would on a Likert scale, responding to every single word-item on a [1 – 7] range without forced comparison between the items, thus avoiding the problems of forced choice measures. At the same time, some awareness of comparison between different items is induced in respondents by presenting the items in a threefold set. For the CAIT, on each page three word-items representing different roles are presented in the form of a triangle imposed on a circle. The respondent is asked to rate each of the three items by indicating a preferred

(8)

position for each of the three items. By the threefold presentation of the word-items a comparative mindset is thought to be induced in respondents so as to mitigate self-presentation tendencies. A more elaborate description will be presented on the following pages.

The Comparison Awareness Inducing Technique (CAIT)

The Comparison Awareness Induction Technique (CAIT) is based on a number of assumptions about the item response process. First, responses to Likert items often fall prey to acquiescence (Dittrich, Francis, Hatzinger & Katzenbeisser, 2007). Consciously or unconsciously respondents tend to agree with seemingly desirable items and disagree with undesirable items, engaging in socially desirable responding (for an overview see Zickar & Gibby, 2006). Thus, on any rating scale representing single dimension scale-points in the direction of the socially desirable scale-end will tend to be overrated. Second, when actively comparing equally desirable or equally undesirable options, respondents will be induced to moderate strong preferences and extreme ratings. And third, even if active comparison is not required, the presence of comparable options will enhance awareness of relative differences and thus moderate the ratings of options unless they are very strongly preferred or rejected. In comparison, previous research has shown high mean scores when asking for identification as well as high correlations between scales with career roles with a Likert method (Hoekstra & Groen, 2008). With the CAIT, we intend to induce comparison awareness, resulting in a downward rating adjustment. We expect that this comparison will result in lower overall ratings and increased differences between the item-words within item-sets.

As Figure 1 shows, the CAIT provides three word-items at once, introducing comparison awareness between options. Hence, chances for one desirability dimension to dominate the mind-set of the respondent are supposed to diminish. Even if such differences are not consciously processed and weighted in the response process, the format will contribute to avoidance of simple desirability connotations. The number three is thought to be crucial here. For example, presenting two options simultaneously, the respondent will most likely enter the mind-set of an either-or script. The comparison leads to the association with choice and competition between items. However, presenting three word-items instead is thought to balance the options within the given context (career roles). Simultaneously identification is being operationalized as being part of the self, by asking participants: ‘To what extent do roles relate to you as a person’ (Expert, Guide, and Inspirer role). As a result, we believe this fits the assumption that through comparison awareness people show less desirable responding patterns and will more inclined to balance extreme scores.

(9)

Figure 1 Example Item Comparison Awareness Inducing Technique. People are asked to rate to what

extent each of the presented words relate to them as a person. The item “Know” refers to the Expert role, “Connect” to the Guide role, and “Stimulate” to the Inspirer Role, respectively.

Graphic item presentation

As shown in figure 1 the word-items in an item-set are graphically positioned on a circle in such a way that they form an equilateral triangle. The orientation within the circle of the item-set triangle is chosen randomly, to make sure the location of the word-items does not influence the reported ratings. As starting position, each item is shown with a rating value of 4, which is the scale-midpoint. Respondents can vary each rating value independently between [1 - 7].

Item-sets as threesomes

An item-set consists of three word-items that are simultaneously shown to participants. Participants are asked to rate all three word-items on their own 7-point Likert scale. There are no constraints how to answer the three individual items. All three word-items can be rated on their own 7-point Likert scale by moving the points to the inner or outer part of the circle.

Item combinations

The number of word-items, and hence item-sets, were chosen such that full competition between all different scales was realized. Thus, the six role scales are pitched against each other an equal number of times.

To what extend do the following words relate to you as a person.

1 = does not relate to me 7 = strongly relates to me

(10)

Development of the Career Role Identification Questionnaire (CRIQ)

The Career Role Identification Questionnaire (CRIQ) was constructed, using the CAIT, to measure identification with each of the six career roles in the Career Roles Model (Hoekstra, 2011).

Generating word-items

Word-items were developed using existing literature and specification of the career role construct (Hoekstra, 2006: Hoekstra, 2011). Starting point was the VLR-30 (a Dutch 30 item-survey for career roles) which was originally developed by Hoekstra (2006). Furthermore, concept definition and presumed effects of the six different career roles served as the basis for word-item generation. All generated word-items were intended to reflect the career role in actions or outcomes. With the career role definitions as guidelines, the first and the third author generated as many verbs (for role-typical actions) and nouns (for role-typical outcomes) as possible to generate a large item-pool.

Word reduction

The initial item pool with both verbs and nouns was reviewed by experts to maximize content validity (DeVellis, 1991). Six experts in career development and psychology were provided with the pool of word-items along with the construct definition for the career roles. The experts were asked to rate each word on a scale from 1 (not representative) to 5 (completely representative). To remain in the pool a word-item had to be rated a three or higher by each of the experts. Subsequently, the five nouns and the five verbs for each of the six career roles that received the highest average scores from the experts, and fitted best, were used for further scale development (see Table 2). This resulted in a total of 60 word-items.

Table 2 Mean Scores Expert-ratings for Each Career Role

Maker Expert Presenter Guide Director Inspirer Verbs 4.73 4.67 4.20 4.83 4.70 4.70

Nouns 4.30 4.63 4 4.54 4.77 4.23

Total 4.52 4.65 4.10 4.69 4.73 4.50

Note. Ratings range from 1-5.

Operationalizing item-sets

The five verbs and nouns representing each scale were all used twice in constructing item-sets, in order to have a full competition for the six career role scales. This resulted in a total of 120 word-items composing 40 item-sets of three word-items each. For each item-set, three word-items from different career role scales are used. Word-items referring to the same career role scale were never used in one item-set. For every word-item in the word-item-set we asked participants to rate on a 7-point scale: ‘To what extent do

(11)

the following words relate to you as a person’ ranging from 1 (I do not relate to this word) to 7 (I strongly relate to this word). Examples of (translated) word-items are ‘Make’(Maker role), ‘Know’(Expert role), ‘Show’ (Presenter role), ‘Connect’ (Guide role), ‘Control’ (Director role) and ‘Stimulate’ (Inspirer role). All Likert rating combinations are possible in every item-set (for example, 2-2-2, 5-3-1 or 2-6-7, see Figure 2).

Overview of the Studies

In the first study we investigated the characteristics of the CAIT, exploring the CAIT-assumptions comparison awareness and sufficiently independent answering. The second study focused on reliability and construct validity of the CRIQ assessing internal consistency and factorial structure of the CRIQ with exploratory factor analyses.

Figure 2 Example Item Comparison Awareness Inducing Technique People are asked to rate to what

extent each of the presented words relate to them as a person. The item “Know” refers to the Expert role, “Connect” to the Guide role, and “Stimulate” to the Inspirer Role, respectively.

To what extend do the following words relate to you as a person.

1 = does not relate to me 7 = strongly relates to me

(12)

Study 1: A Simulation Study of Scale Independence

The goal of study 1 was to test the assumptions behind the CAIT. The first assumption is that the scores for the six role-scales are non-ipsative, rendering sufficiently independent scales. Ipsative measurements result in negative correlations between scales. This follows from the notion that the sum-score of each respondent is a constant, and therefore has zero variance. Because the variance of the sum equals the sum of the variances and covariances, some, if not all of the covariances should be negative to cancel the positive sum of the variances. The correlations will also be negative, at least their mean. If all scale-variances are about equal the mean correlation is about equal to -1/(k-1), where k is the number of scales. For example, with two scales a correlation of -1 is found between them, while the mean inter-correlation between three scales will be around -.50. It is important to note that these correlations are an artifact of the ipsative measurement method. The question is whether this method will by itself generate negative correlations between scales as does the ipsative method. As stated, we suppose this will not be the case. The second assumption is that the semi-comparative response format leads to moderation of all-out desirable responding, in turn causing low to medium correlations between items within sets (i.e. from different role scales).

The assumption behind the ‘awareness inducing’ method is that, although the respondent is free to rate each item on a 1 - 7 Likert scale, the three word-items in an item-set will presumably not be rated independent of each other. We suppose that each word-item is rated in comparison to the other two word-items in the item-set, resulting in some kind of ‘weighted’ scores. This line of reasoning has two implications for the characteristics of surveys using CAIT: First, the data should show no negative correlations within item-sets as would be the case for ipsative data because with CAIT people are free to rate word-items independent of each other. Furthermore, through comparison awareness word-items within item-sets should show low to medium positive correlations. Finally, we expect that word-items from the same scales show moderate to high correlations between same role word-items over different item-sets confirming consistency of the scale. This leads to the following hypotheses:

Hypothesis 1. There are no negative correlations between different scales.

Hypothesis 2. Within item-sets, presented word-items have lower or equal correlations compared to same scale word- items between-sets.

Hypothesis 3. Between item-sets, same scale word-items show moderate to high correlations (.30 < r > .60).

(13)

Method Study 1

To test the CAIT- assumptions a program was written by the second author (in DelphiXE2) to perform a number of simulations. The program generated uncorrelated data resembling the data-structure from the CRIQ, N = 235, 40 item-sets consisting of 3 word-items each, constituting six scales. Although uncorrelated, a number of restrictions and parameters apply to the generated simulation-data. First, the upper and lower bounds for the sum-score allowed could be set to any value from 3 to 21, resulting in intervals with a range of 0 (both bounds set to the same value) to 18 (bounds set to the interval [3-21]). Second, the variability of preferences for the six scales, and (independently) the variability of the word-items within each scale, could be varied. The ‘comparison awareness’ is simulated by generating three ratings for the word-items independently on the basis of preference for the scales, and then adjusting them to the given range and standard deviation of the sum-score. The simulation-data is used to (a) test the CAIT-assumptions in a simulated data-set with uncorrelated scales and (b) compare the simulated results for different parameters, with the CRIQ-results from the empirical sample. The found differences can be indicative for the ‘true state of mind’ when answering the CAIT.

Results Study 1

Table 3 shows the results of multiple simulated data-sets varying in item-set interval ranges. For comparison, the final row of table 3 shows the values found in the empirical sample (Study 2). As Table 3 shows, for fairly ipsative conditions (range 3 - 7) negative overall correlations between the six scales were found. However, with a relatively small range of allowable sum-scores of the item-sets (range > 9) negative between scales correlations disappear. These results are comparable for all values of the between-scales and within-scales variances, within reasonable boundaries. Furthermore, the mean between-set correlation for word-items of the same scale remains .33 overall, which indicates that the six scales are reliable (note that the value is the mean correlation between two word-items, not a reliability coefficient alpha). The mean between-set correlation for word-items of a different scale is slightly negative in the near-ipsative condition and rises to low positive values. The mean within-set word-item correlation is close to -.50 in the near-ipsative condition, as expected, and becomes positive when the interval range increases. The mean correlation between scales is close to -.20 in the near-ipsative condition, as expected, and becomes positive for larger interval ranges. Altogether, it seems fair to conclude that this procedure generates moderate correlations for the six scales when there is no range restriction for the item-set [3 - 21] and considerable range within item-sets [1 - 7].

Hypothesis 1. There are no negative correlations between different scales.

(14)

deviation of item-scores and item-sets, a mean r = .21 over all scales is found confirming hypothesis 1.

Hypothesis 2. Within item-sets, presented word-items have lower or equal correlations compared to same scale word- items between-sets.

In the simulation that resemble the empirical data most, in mean and in standard deviation of item-scores and item-sets, a within-set of r = .22 is found, which is similar to the empirical value of r = .26. These values are lower than the between item-set same scale word-item r =.34 in the simulation study and r = .48 in the empirical study. Therefore, Hypothesis 2 is confirmed.

Hypothesis 3. Between item-sets, same scale word-items show moderate to high correlations (.30 < r > .60).

In the simulation that resembled the empirical data most, a between item-set same scale word-item r = .34 is found which is slightly lower than the empirical value of r = .48. Both results fall in the expected range of correlations, therefore Hypothesis 3 is accepted.

To a large extent the results support our assumptions. In the next study we test the assumptions in an empirical sample and evaluate and compare the empirical sample to the simulation-study.

Table 3 Generated Simulated Results Varying Degrees of Ipsativity.

Range interval Mean item-score SD item-score Mean sum item-set SD item-set mean r within-sets (on item level) Mean r between-sets same scale (on item level) Mean r between-sets different scale (on item level) Mean r scales over all scales (scale level) 3 4.37 1.16 13.11 0.64 -0.43 0.33 -0.06 -0.19 5 4.37 1.27 13.10 1.51 -0.21 0.29 -0.05 -0.14 7 4.44 1.39 13.33 2.04 -0.08 0.31 -0.04 -0.07 9 4.49 1.48 13.47 2.58 0.06 0.33 -0.01 0.06 11 4.52 1.57 13.57 3.17 0.22 0.34 0.04 0.21 13 4.54 1.59 13.65 3.32 0.25 0.36 0.05 0.21 15 4.66 1.73 14.03 4.05 0.42 0.34 0.12 0.40 17 4.51 1.74 13.57 4.08 0.42 0.37 0.13 0.39 19 4.36 1.78 13.10 4.27 0.46 0.37 0.14 0.44 19 4.90 1.55 14.75 3.20 0.26 0.48 0.28 0.54

Note. N = 235. The last row shows results found in the empirical sample.

(15)

Study 2: Psychometric Properties of the CRIQ in a Field Sample

In Study 2 we aimed to map the factorial structure of the CRIQ and establish robustness of the structure across two different samples. Based upon initial results we made minor changes for the second sample and adapted the CRIQ 1.0 to CRIQ 1.1 by replacing a total of seven word-items for the Presenter- and Inspirer- scale. We used both data-sets in the analysis.

Method Study 2

Participants and Procedure

For CRIQ 1.0 respondents were a Dutch random sample of 310 employees from different companies in the Netherlands. The sample represented a wide range of professions (e.g., technicians, nurses, doctors, policy makers) that participated in a longitudinal portal survey on a voluntary basis. The survey was part of a larger survey, not reported here. Based upon initial results, for the second study the CRIQ1.0 was adapted into CRIQ1.1 by replacing a total of seven word-items for the Presenter- and Inspirer- scale. For CRIQ 1.1 respondents were a Dutch random sample of 235 employees that participated in an online survey on a voluntary basis. Respondents were visitors on the website of a Dutch HRM company and received no payment. Personal data like age, gender and job type were not known for this group; generally this site is visited by persons interested to improve on their work situation in some way.

Analyses

To analyze the CRIQ scales Cronbach’s alpha, mean scores, standard deviations, and correlations were examined. Subsequently, an exploratory factor analysis (varimax rotation) was conducted to map the factorial structure. In addition, confirmatory factor analysis were conducted in order to further test the model. The comparative fit index (CFI) and the root-mean-square error of approximation (RMSEA) are used in determining model fit. The CFI ranges from 0 – 1, measuring the relative fit in relation to a null model of complete independence, CFI measures larger than 0.95 are indicative of a good fit (Bentler, 1990). The RMSEA is a close fit measure, known to be relative insensitive to sample size. An RMSEA value of 0.08 or smaller is indicative of acceptable fit (e.g., Schermelleh-Engel, Moosbrugger, & Müller, 2003).

After investigating both samples separately, orthogonal procrustean results are presented rotating the CRIQ1.1 factor solution to the CRIQ1.0 (target) factor solution. Although the CRIQ1.1 has the seven improved word-items, CRIQ1.0 was chosen as the target solution because more background information is available for this sample.

(16)

Results Study 2

Preliminary Exploration

Figure 3 shows a summary of item-characteristics of the 40 item-sets of the CRIQ from the field sample (N = 235). The results show that the range of the item-sets differ from [3 - 21] indicating that there is no observable restriction for the total score of the item-set as the total sum varies between all possibilities.

Figure 3 Frequencies of the sums of item-sets of the CRIQ.

Reliability and Descriptive Statistics

Tables 4a and 4b show the reliabilities and descriptive statistics for the different CRIQ-scales. The six career role scales had satisfactory reliabilities (α ranging from .94 to .96). The different career role scales were moderately to highly correlated (correlations ranging from .21 to .79). High correlations were found between Maker and Guide role (CRIQ1.0 r = .38; CRIQ1.1 r = .66), Presenter and Inspirer-role (CRIQ1.0 r = .79; CRIQ1.1 r = .68), Director and Inspirer role (CRIQ1.0 r = .70; CRIQ1.1 r = .59), and Guide and Inspirer role (CRIQ1.0 r = .63; CRIQ1.1 r = .65).

Complete independence is theoretically not to be expected as roles are, to some extent, related to one another and therefore may be connected in respondent’s perception (Hoekstra 2011). Furthermore, often people are likely to identify with more than one role. In light of the simulation study we can estimate differences between correlations found empirically and correlations generated due to the CAIT. The increase of variability within-sets appears to generate accumulative overall scale correlation. As shown in table 3 in the simulation study an overall scale correlation of r = .44 is found compared to an overall scale correlation of r = .54 in the empirical sample, when the interval range is 19. As table 3 shows, for the simulation that resembles the empirical data most in mean and in standard deviation of item-scores and item-sets, we find a mean r = .21. Therefore, based on the simulation study we may estimate that part of the correlations will be an effect of the CAIT, resulting in an over estimation of the theoretical r = .21 dependence.

(17)

Table 4a Scale Characteristics of the CRIQ 1.0

Roles α M SD 1 2 3 4 5 6 1.Maker .97 5.07 1.15 - .43** _.38** _.38** _.38** _.39** 2. Expert .96 4.94 1.12 - .29** _.36** _.30** _.41** 3. Presenter .93 4.17 1.02 - .57** _.62** _.79** 4. Guide .95 4.96 1.10 - .43** _.63** 5. Director .95 4.43 1.12 - .70** 6. Inspirer .94 4.68 1.06

-Note. Results of the CRIQ1.0 (N = 310) are presented. ** p < 0.01.

Table 4b Scale Characteristics of the CRIQ 1.1

Roles α M SD 1 2 3 4 5 6 1. Maker .96 5.10 1.07 - .54** _.53** _.66** _.54** _.51** 2. Expert .96 5.15 1.11 - .37** _.58** _.44** _.47** 3. Presenter .92 4.21 .95 - .53** _.55** _.68** 4. Guide .95 5.19 1.01 - .49** _.65** 5. Director .96 4.88 1.10 - .59** 6. Inspirer .94 4.97 1.04

-Note. Results the CRIQ1.1 (N = 235) are presented. ** p < 0.01.

Factorial Structure

To test the proposed six-factor solution we performed an exploratory factor analysis For CRIQ1.0 and CRIQ1.1 on the 120 word-items. After varimax rotation a clear structure of the six career role scales emerged. For CRIQ1.0 all items loaded highest on their own scale, except for 1 Presenter-item, that loaded highest on the Director-scale and 10 Inspirer-items that loaded higher on the Director- Maker- and Guide-scale. After slight changes were made for the CRIQ1.1 all items loaded highest on their own scale, except for 5 Presenter-items, they loaded higher on the Inspirer-, Maker-, and Director-scale (see Table 5).

Subsequently, we fitted the CRIQ1.0 in the proposed six factor model. This model did not fit adequately, (RMSEA = 0.08, CFI = 0.618). Inspecting modification indices, we detected, as expected in light of the item content, correlated residuals between the identical word-items of the survey. Allowing these residuals to correlate resulted in an improvement in model fit (RMSEA = 0.059, CFI = 0.794).

To test robustness of the CRIQ-scales we performed an orthogonal procrustean rotation. Table 4 shows results of the CRIQ1.1 factor solution rotated to the CRIQ1.0 (target-) factor solution. Table 5 shows results of the Phi-coefficients for the scales, coefficients vary between .91 and .96 corresponding to a fair similarity, indicating that most factors can be considered equal (Lorenzo-Seva & Ten Berge, 2006).

(18)

Table 5 Factor Loadings of the CRIQ1.0 & CRIQ 1.1 Rotated Component Matrix

Career Role Director

1.0 1.1 1.0 1.1Maker 1.0 1.1Expert 1.0 1.1Inspirer 1.0 1.1Guide Presenter1.0 1.1

1.doen .78 .74 4.maken .67 .56 7.aanpakken .80 .67 10.uitvoeren .81 .66 13.realiseren .80 .64 16.doen .86 .77 19.maken .81 .64 22.aanpakken .82 .72 25.uitvoeren .83 .71 28.realiseren .82 .69 61.actie .70 .74 64.daadkracht .47 .53 .57 67.resultaat .58 .58 70.oplevering .64 .52 73.handeling .60 .62 76.actie .76 .72 79.daadkracht . 44 .67 .64 82.resultaat .67 .57 85.oplevering .68 .60 88.handeling .72 .70 3.analyseren .70 .66 14.onderzoeken .75 .78 17.onderzoeken .77 .79 20.bestuderen .79 .69 32.weten .67 .58 35.analyseren .78 .71 38.denken .74 .70 49.weten .80 .66 52.bestuderen .79 .78 55.denken .78 .76 63.kennis .77 .68 74.precisie .56 .63 77.onderzoek .81 .80 80.inzicht .60 .50 .44 92.inzicht .63 .52 .41 95.specialisme .61 .60 98.onderzoek .81 .82 109.kennis .41 .72 .65 112.specialisme .62 .55 115.precisie .53 .65 2.vormgeven .55 .43 .37

2

(19)

1.0 1.1 1.0 1.1Maker 1.0 1.1Expert 1.0 1.1Inspirer 1.0 1.1Guide Presenter1.0 1.1

5.presenteren .46 .29 8.indruk maken .68 .63 11.beinvloeden/ verleiden .42 .23 .44 31.etaleren .74 .66 34.etaleren .41 .74 .68 37.vormgeven .50 .49 .40 40.presenteren .46 .41 43.beinvloeden/ verleiden .26 .48 46.indruk maken .72 .71 62.stijl .56 .51 65.uitstraling .52 .32 68.houding/ show .46 .44 .77 71.vorm .52 .44 91.uitstraling .43 .62 .37 94.vorm .55 .38 97.stijl ,63 .42 100.impact .40 .41 ,03 .17 103.houding/ show ,41 .72 106.impact .48 .19 .19 6.helpen .67 .58 15.bemiddelen .60 .54 23.verbinden .58 .55 26.inleven .77 .60 33.begeleiden .69 .62 41.bemiddelen .67 .66 44.inleven .46 .77 .57 50.verbinden .68 .51 53.helpen .41 .69 .57 58.begeleiden .70 .56 66.feedback .43 .39 75.begrip .66 .50 83.steun .77 .64 86.coaching .70 .59 93.hulp .69 .62 101.feedback .48 .36 104.steun .77 .68 110.coaching .44 .72 .62 113.begrip .75 .53 118.hulp .73 .62 9.stimuleren/ ruimte scheppen .24 .52 ,59 18.inspireren .44 .40 .61 ,43

(20)

1.0 1.1 1.0 1.1Maker 1.0 1.1Expert 1.0 1.1Inspirer 1.0 1.1Guide Presenter1.0 1.1 24.uitdagen/* .47 .43 .38 29.vernieuwen .65 .71 36.prikkelen .44 .45 .49 42.uitdagen/ ter discussie stellen .52 .47 .46 47.inspireren .46 .62 51.stimuleren/ ruimte scheppen .50 .33 .49 .49 56.vernieuwen .69 .67 59.prikkelen .52 .59 .40 69.overtuiging/ verandering .52 .46 .22 .58 78.visie/ toekomstvisie .51 .41 .49 .57 84.waarden/ idealen .20 .45 .43 89.creativiteit .62 .67 96.creativiteit .66 .69 102.verbeelding .43 .61 107.waarden/ idealen .22 .46 .44 111.visie/ toekomstvisie .56 .47 .49 .56 116.verbeelding .47 .63 119.overtuiging/ verandering .45 .42 .25 .61 12.reguleren .44 .45 .44 21.beslissen .73 .77 27.besturen .70 .74 30.besturen .74 .73 39.beslissen .73 .78 45.leiden .76 .79 48.regisseren .78 .70 54.reguleren .53 .55 57.leiden .77 .76 60.regisseren .76 .67 72.strategie .64 .54 81.overzicht .48 .49 87.sturing .72 .71 90.regie .78 .73 99.koers .67 .58 105.koers .64 .65 108.strategie .65 .60 114.regie .77 .78 117.overzicht .48 .49 120.sturing .76 .73

Note. This table shows factor loadings for both CRIQ 1.0 and CRIQ 1.1. CRIQ1.0 (N = 310). Explained variance is 59% . CRIQ 1.1( N = 235). Explained variance is 57%. Loadings < .40 were omitted. 24.uitdagen* in CRIQ1.1 is a residue of CRIQ1.0.

(21)

Table 6 Phi-coefficients Procrustean Rotation

Career Role

Maker Expert Presenter Guide Director Inspirer Maker .97 Expert .93 Presenter .96 Guide .93 Director .91 Inspirer . .91

Note. Procrustean rotation matrix available upon request.

General Discussion

Despite increased attention for the changing nature of jobs and careers, development of practices and instruments supporting these developments is still in its infancy. The aim of this chapter was twofold. First, we introduced and evaluated the “Comparison Awareness Inducing Technique (CAIT); an innovative method to reduce the effects of self-presentation tactics. Second, we presented and evaluated the Career Role Identification Questionnaire (CRIQ), a new measure for role identification. We tested the assumptions of the CAIT and psychometric properties of the CRIQ in two studies.

The results from Study 1 indicate that the CAIT receives encouraging support as a new way of dealing with self-report measures and socially desirable responding. The method does not generate negative correlations when the range of sum-scores is sufficiently large. As the empirical data show, this is what happens in a real application. Furthermore, the method does generate correlations between scales, if the within-set variability is large. Meaning that the overall role-correlations found in an empirical study are probably overestimated. Furthermore, the results from Study 2 show encouraging support for our conceptualization of the Career Role Identification Questionnaire (CRIQ) as a measure of identification with the six career roles. After adapting the CRIQ1.0 to the CRIQ1.1 a clear factorial structure emerged. Additionally, item loadings were found to be equivalent across the two samples, indicating the robustness of the CRIQ. In summary, the results indicate that the CAIT is an innovative method for dealing with self-presentation tactics and used across samples the CRIQ is a reliable and robust measure of career role identification.

Limitations and Future Research

Although the proposed factorial structure emerges from both samples indicative of existence of six career roles, there is room for improvement. In line with previous research, the Presenter-, Director- and Guide-scale show high inter-correlations (Hoekstra, 2011). Due to these high correlations, the Inspirer-scale emerges less clear from the factorial structure. In spite of these results, overall the career roles were found to be reliable and

(22)

constitute distinct dimensions, indicating conceptual accuracy of the items. However, future research may offer insight in these results. For example, although all career roles can vary independently, environmental influence can vary resulting in broad and narrow jobs. Subsequently, workers may adopt and identify with more or less roles causing (high) inter-correlations between roles.

Furthermore, although the RMSEA (0.059) is satisfactory, the CFI fit is below the 0.90, which is generally acknowledged as an acceptable fit. It has been reasoned however, that fit indices may reflect other uncontrollable variables in addition to reflecting model fit. Different aspects such as model-type and model complexity may influence the fit indices and model-fit results, respectively (Fan & Sivo, 2007). With regard to our model, a simple structure does not apply, because the items consist of two parts. There is a general part, that applies to all latent variables (a good worker role, that comprises all six career roles) and an item-specific part, that applies especially to the one role. For example, there are different ways in which an individual can contribute to the organization, which is reflected in the items. It is through set tasks that task performance is assessed, which is job specific (e.g., career roles). Another contribution is made through organizational citizenship behavior (e.g., the good worker role). Which is ‘Individual behavior that is discretionary, not directly or explicitly recognized by the formal reward system, and in the aggregate promotes the efficient and effective functioning of the organization (Organ, Podsakoff, & Mackenzie, 2006). This type of contribution is similar across jobs and roles, and supports the social & psychological environment in which task performance takes place (Organ & Ryan, 1995). The complexity of the model should be taken into account when interpreting the model-fit.

Implications

Implications of the CAIT

One of the challenges in constructing identification measures is to prevent the risk that self-serving motives come to dominate the responses (Robins & John, 1997). The CAIT provides a new way for measuring identification through comparison awareness. When introducing awareness between options, it is thought chances will diminish for one desirability dimension to dominate the mind-set of the respondent. Although we used the CAIT to measure identification with career roles the CAIT can be transferred to other areas of identification measures. Furthermore, we have shown that with this technique people are free to respond on the items as they would on a normal Likert scale, tackling the problem of forced choice ipsative data. Taking the present results and possible future applications into consideration we believe that the CAIT has the potential to make a significant contribution to research in various domains beyond career role identification.

(23)

Implications of the CRIQ

Within the same job, employees can differ in how they perceive their job. From this perspective, it seems relevant how people perceive their roles and how these beliefs influence role behavior (Neale & Griffin, 2006). Although job descriptions might be the same, how an employee perceives his/her own job can lead to different role orientations and subsequently, different role behavior (Neale & Griffin, 2006; Parker et al., 1997). How people respond to different role expectations will be influenced by expectations of a role, but also how salient a role is for that person given the context (Neale & Griffin, 2006). Specifically, the CRIQ may give insight into how important certain roles are for people and consequences on career role enactment in terms of motivation and/or performance (Hall, 2002).

Acknowledgements

The authors would like to thank all experts for their help concerning the development of the CRIQ questionnaire.

(24)

(25)