The effect of Middle Managers’ Age Stereotypes on the employees’ IS Adoption
MASTER THESIS
Corlien Verkerk S2120593
c.c.c.w.verkerk@student.rug.nl January 2018
Supervisor: Drs. M.L. Hage Assessor: Dr. B. Mueller
MSc. Business Administration Change Management Faculty of Economics and Business
University of Groningen
Word count: 14539
ABSTRACT
The increasing use of information systems in business in combination with the aging of the workforce might lead to difficulties for organizations, which requires more research. This study aimed at
contributing to this literature gap by investigating the role of middle managers' age stereotypes on the individual IS adoption. However, the data was not sufficient to conduct a multi level analysis.
Therefore, the data was solely analyzed on a single level. We examined the perception of a negative age-discrimination climate and the intention to use an IS system as dependent variables. We found some surprising results, where only nationality explained significant variance of the perception of negative age discrimination climate, and only experience with the implemented IS system explained significant variance on intention to use the system. Remarkably, the variables from the extensively studied UTAUT IS adoption model (Venkatesh, Morris, Davis & Davis, 2003) were not significantly related to the intention to use the system. This was possibly due to the timing of the survey, a low response rate in combination with a selection bias. However, although we did not find many
significant results, we did find some interesting trends in the data that with additional research could potentially lead to new valuable insights regarding to age and the perception of a negative age- discrimination climate in relationship to the IS adoption model. Possibly our results indicate that the IS implementation is a context where age becomes salient, which might influence younger people to use the IS implementation as an opportunity to get ahead in the organization. Additionally, it might be that organizational tenure might buffer the effect of the perception of a negative age-
discrimination climate for older employees. Lastly, it seems that that the differences between the Dutch and Indian participants become larger with age, which would be in line with previous research.
Our observed trends require subsequent research with a larger sample size to examine the hypotheses.
Keywords: Age stereotypes, perception of negative age-discrimination climate, IS adoption, UTAUT model, middle managers
1. INTRODUCTION
A current business trend is the increasing use of Information Systems (IS) use (Galliers, Jarvenpaa, Chan & Lyytinen, 2012). Information Systems both take a more important role in business and are growing in complexity (Ziefle & Bay, 2005). Another trend in business is the aging of the workforce in organizations (Organization for Economic Co-operation and Development, OECD, 2011). Due to, among other things, health improvements, longer life expectancies, greater financial incentives, and pension policy changes the amount of older employees (55 years and older) is growing and the amount of younger employees is declining (OECD, 2011). This trend is called ‘the greying of the workforce’ and is expected to persist the coming years (Tams, Grover & Thatcher, 2014). Tams, Grover and Thatcher (2014) state that these two developments in business require more research to examine ‘the managerial challenges that arise from the interplay between a greying workforce and rapidly evolving use of Information Systems’.
The past decades there have been a fast growth in the digitization of work and an increase on reliance on computers at the workplace (Valletta, 2015). IS research investigates this trend and has identified many variables that affect organizational adoption of IS (Sharma & Rai, 2015). IS research can be defined as the IS developments in the organizational application of digital computer and communications technologies (Swanson, 1994). Research on the adoption and individual use of IS is one of the most mature streams within IS research (Venkatesh, Thong & Xin Xu, 2016). For an organization it is key that employees adopt an IS system as fast and effectively as possible to benefit the most.
During an IS implementation most often management is in the lead. Employees are the one’s that have to get used to working with a new system (Venkatesh, Morris, Davis & Davis, 2003). Several studies have examined the role of management on IS adoption by employees. E.g. there has been research concerning managers’ personal characteristics (Sharma & Rai, 2003), and managers roles (Benbasat & Barki, 2007) in the IS context. Top management support was found to be a critical predictor of IS adoption, middle management support was found to be a promising predictor but has not been studied extensively (Jeyaraj , Rottman & Lacity, 2006). However, overall there is still a lack of understanding of managerial behavior in IS implementation (Sharma & Rai, 2015).
One of the aspects that could be further explored is the role of managerial age stereotypes within organizations. Age stereotypes are present and persistent in organizations, and often negative (Posthuma & Campion, 2009). Negative age stereotypes can lead to ageism. ‘Ageism is prejudice or discrimination against or in favor of an age group’ (Palmore, Branch & Harris, 2005). Literature shows
that age stereotypes influence managerial actions. Managers who hold age stereotypes may believe that older employees are less productive, less motivated and they have a lower ability than younger employees (Posthuma & Campion, 2009). These stereotypes can lead to discrimination of older employees. A negative IS related age stereotype is the belief that it is more challenging for older employees to adjust to new IS in comparison to their younger colleagues. Research has shown that negative age stereotypes can have severe negative consequences for, especially, older employees (Boone James, McKechnie, Swanberg & Besen, 2013; Posthuma & Campion, 2009; Truxillo, Finkelstein, Pytlovany, & Jenkins, 2015). Although the empirical support to justify these believes is largely lacking, age stereotypes have proven to be persistent in organizations (Ng & Feldman, 2012;
Posthuma & Campion, 2009). Therefore, within the IS field it is highly relevant to study worker behavior in response to managerial actions and how the worker behavior is influenced by age stereotypes.
IS implementation is currently an important topic within all kinds of organizations. Within large organizations it is important to make the distinction between top and middle management in the context of IS implementation. Top management usually has the power to decide to implement a new IS. Middle management has more daily contact with the employees, and are the one’s that provide direct resources towards their employees during the process (Jasperson, Carter & Zmud, 2005). Top management support has been found to be one of the critical factors in IS adoption (Jeyaraj et al., 2006). There is less known about the precise effect middle managers have in the context of IS implementation. However, there a indications that, because of there close connection to the end user, middle managers and their decision making influence IS adoption (Venkatesh & Bala, 2008).
Overall, within the managerial IS research there is no research about the effects of age stereotypes of top and middle managers on IS adoption.
1.1 Research Aim and Relevance
This study is a reaction to the limited research about the effect of managerial age stereotypes on employee IS adoption behavior within the implementation context. In order to fill this literature gap, this research aims at answering the following research question:
‘What is the effect of middle managers’ age stereotypes on employees IS adoption?’
The aim of the research is to understand more about the role of middle management on individual
stereotypes. Hypotheses were developed by combining insights from the IS and age stereotype literatures.
This research focuses solely on the effect of middle management’ age stereotypes on IS adoption.
Middle managers are the one’s in direct contact to the end user of the system. The determination of the effect of middle managements’ stereotypes on IS adoption contributes to scientific research by filling the literature gap. Previous research showed that managerial age stereotypes influence their managerial actions; it influences how managers treat their older employees in comparison with their younger employees (Taylor, 2001). Negative age stereotypes can lead to multiple negative
consequences on the individual and organizational level (Collella & King, 2015), which will be explained in more depth in the literature review. Within the IS implementation research there is a literature gap when it comes to the effects of managerial behavior (Sharma & Rai, 2015), especially on middle management. No research has focused on the specific aspect of middle managements’ age stereotypes in the context of IS implementation. Therefore, this research will give insights in to an underexplored area of IS research and is theoretical relevant.
In terms of managerial implications, it is key for managers to know how IS adoption is affected by their actions, so they can apply right methods to approach their employees in an IS implementation process. It might be that employees adjust their behavior in response to managerial actions, which can eventually lead to differences in the IS adoption. Age stereotypes might be particularly salient in certain industries and context, and the information and technology may be a situation in which older employees are in a disadvantaged position in comparison to the younger employees (Kanfer &
Ackerman, 2004). Therefore, it is important to investigate this topic more thoroughly. More
knowledge will lead to more insights about the effect of managerial age stereotypes in the context of an IS implementation. Finally, these insights can lead to a constructive approach and development of guidelines that will help to guide implementation of new systems as effectively as possible for especially older employees.
Furthermore, the study examines the effect on the IS adoption of the employee. The scope of the research is to examine the effects it has on the end user of the system. The main goal of IS
implementation is that the system is used by the employee and thereby improves job performance and adds benefits by improving efficiency and effectiveness in the organization (Business Software, 2017). Therefore, the actual IS adoption of the employee is the most relevant outcome measure.
Overall, this study aims at contributing to the scientific literature by investigating the unexplored effect of middle managements’ age stereotypes employee behavior in the context of IS
implementation.
2. LITERATURE REVIEW
This literature review provides coverage of several relevant areas of literature. First, age stereotypes at the workplace are discussed in more depth. Second, age stereotypes in the context of IS
implementation are reviewed. Third, managerial actions in context of IS implementation are examined. This is followed by employee responses to managerial actions within the context of IS implementation to gain a deeper understanding of this relationship. Lastly, the conceptual model and hypotheses are derived from the available literature.
2.1 Age Stereotypes at work
Beliefs and expectations about employees based on their age are called age stereotypes (Hamilton &
Sherman, 1994). Often stereotypes are ‘negative, inaccurate, or distorted opinions about people based on their membership in a particular group’ (Fiske & Neuberg, 1990). Stereotypes are based on unfounded assumptions that lead to the incorrect conclusion that all people within a group are the same. This makes stereotypes different from prejudice and discrimination. Prejudices are more conscious (Fiske, 1998); discrimination is more behavioral (Nelson, 2002). However, despite the fact that age stereotypes are often subconscious, they can trigger and lead to discriminatory behavior (Bal, Reiss, Rudolf & Baltes, 2011). The discrimination of older employees may become manifest in decisions such as hiring, training opportunities, performance appraisals, and termination/layoffs.
Additionally, overt and subtle forms of discrimination might influence interpersonal contact and different expectations or perceptions of older employees (Collella & King, 2015). Thus, by
unconsciously classifying a person into a certain group, attributes related to that group are retrieved and connected to the individual. Consequently, the stereotypes concerning that group are used to predict the behavior of an individual, which might result in discriminatory behavior.
Age stereotypes can be negative but also positive. In their meta-analysis Posthuma & Campion (2009) summarized the most common age stereotypes in the workplace. They found the most common negative stereotypes; older employees are believed to have a lower productivity, to be more resistant to change, to be less able to learn, to have a shorter tenure, and to be more costly in comparison to younger employees (e.g. Brooke & Taylor, 2005; Chiu, Chan, Snape, Redman, 2001;
Gordon & Arvey, 2004; Hedge, Borman, & Lammlein, 2006; Ostroff & Atwater, 2003). Additionally, they summarized the most common positive age stereotypes: older employees are seen as more stable, reliable, honest, trustworthy, loyal and committed to the job (e.g. Bal et al., 2011; Hedge et al., 2006).
Age stereotyping is not necessarily done based on facts. In fact, little empirical evidence has found negative age stereotypes to be true (e.g. Ali & Davies, 2003; Broadbridge, 2001; Hedge et al., 2006;
Kubeck, Delp, Haslet & McDaniel, 1996). For example, a persistent and prevalent age stereotype is that older employees are less innovative (Ng & Feldman, 2013). However, in a meta-analysis Ng &
Feldman (2013) found that that older employees do not engage in less innovation-related behavior than the younger employees. Overall, whether the age stereotypes are valid or not, the reality is that they are very persistent in organizations these days and have effect on managerial behavior
(Posthuma & Campion, 2009).
This research focuses solely on negative age stereotypes. As aforementioned negative age
stereotypes can lead to discriminatory behavior of older employees, which on an individual level may result in less training opportunities (Wrenn & Maurer, 2004), more negative performance appraisals (Truxillo et al., 2015), quicker layoffs (McGoldrick & Arrowsmith, 2001) and a higher chance of being bullied (Einarsen & Skogstad, 1996) for the older worker in comparison to the younger worker.
Additionally, Boone James et al. (2013) found that on an individual level older employees are limited from substantives job responsibilities or job-related career development opportunities because of their age. Furthermore, Ng and Feldman (2013) found that managerial stereotypes reduce
innovated-related behavior of less productive employees. Because of the serious implications age stereotypes may have they are of practical and scientific relevance.
In addition, age-related assumptions are shown to be influential at the workplace and require a better understanding. In age-diverse teams age subgrouping occurs (Wegge et al., 2012). Age subgrouping is when people classify themselves within a group of the same age category. When age subgrouping occurs, and becomes a factor of social categorization, that is noticeable within the group, this can lead to the perception of a negative age-discrimination climate (Kunze, Boehm &
Bruch, 2013). Additionally, Kunze et al. (2013) found those top managers’ negative age stereotypes moderates the relationship between age diversity and the perception of a negative age-
discrimination climate. When top managers had high levels of negative age stereotypes the relationship between age diversity and a negative perception of the age-discrimination climate became positive.
At firm level, research examined the role of age-related policies directed at older employees. It showed that these policies are beneficial to counteract the negative consequences of age
outcomes. They hypothesized that collective sense making HR-practices can lead to a more positive age diversity climate. The positive age diversity climate was hypothesized to positively influence the firm outcomes. Their study confirmed their hypothesis. Moreover, the study showed that the age- diverse HR-practices had a positive effect on individual collective turnover intentions, these declined.
This study extends the line of research by focusing on lower levels within the organization, namely the effect of middle managers’ age stereotypes on the individual.
To sum up, age stereotypes are widespread and negatively affect organizational and individual level outcomes, even though they are largely unsupported by empirical data. Managerial action (e.g. HR policy) might help to counteract negative consequences of age stereotypes on a top management level. On a middle management level research found considerable evidence that age stereotypes influence multiple employment related decisions such as hiring, performance appraisals, and job responsibilities and opportunities (Boone James et al., 2013; Posthuma & Campion, 2009; Truxillo et al., 2015). However, even though the practical and scientific relevance of age stereotypes has been shown, there is still a gap in the literature when it comes to the specific effect on middle managers’
age stereotypes on individual outcomes of the employee.
2.2 Age stereotypes in the context of IS implementation
IS implementation could be a distinct situation in which older employees may be discriminated because of their age (Kanfer & Ackerman, 2004). Research has shown that high-tech work is strongly associated with younger employees (Warr & Pennington, 1994). A ‘young and masculine’ worker is perceived as the most suitable for high-tech work (Ensmenger, 2003). Even young adult employees are considered old in technical occupations. For example, Comeau & Kemp (2007) found that within the IT field sports and military metaphors were used a lot. Within these metaphors ‘old’ was
considered around 30-year of age and these ‘older’ individuals must move from the team to a more coaching role because then they do not fit well in the ‘younger’ team anymore. This example illustrates that there might be negative association between technical work and employees’ age.
Eventually, in the high-tech industry research showed that a negative association leads to discrimination of older employees (Comeau & Kemp, 2007).
An IS implementation is not related solely to jobs in the high-tech industries but due to technological advantages are of great importance within all types of organizations nowadays. Because of the aforementioned association between technological related work and a young age, this can have possibly negative effects for older employees in the context of an IS implementation. For example negative age stereotypes such as; older employees are less flexible, more resistant to change, and a
lower ability to learn in comparison to their younger co-employees might be drawn upon by co- employees and managers. Therefore, the IS implementation context is very suitable to investigate the relationship between managerial ageism and effect on employees.
2.3 Managerial actions during IS implementation
In IS research there is still a lack in understanding managerial behavior (Sharma & Rai, 2015). Top management usually imposes the IS implementation within an organization, while middle management supervises the implementation process. Jeyaraj et al. (2006) reviewed in a meta- analysis the main predictors, linkages, and biases in IS innovation adoption research. They found that top management support was, among perceived usefulness, computer experience, behavioral intention, and user support to be the main linkage in individual IS adoption. Thus, top management support might not have a uniform effect on every one, depending on how inclusive they are. Middle management support was found to be a promising predictor. However, they indicated that more research is required to be able to draw any conclusions.
The role of middle managers within an organization can be defined as: ‘middle level managers endorse, refine, and shepherd entrepreneurial opportunities and identify, acquire, and deploy resources needed to pursue those opportunities’ (Kuratko, Ireland, Covin & Hornsby, 2005). Middle managers are very important within the organization because they both understand the operational issues in the organization and additionally have access to top management (Wooldridge, Schmid &
Floyd, 2008). Literature suggests that because of their position within the organization middle managers can promote and develop innovative ideas and at the same time implement strategy (Floyd & Wooldridge, 1992). The contribution of middle managers has been linked to positive outcomes such as, profit growth (Mair, 2005), strategy realization (Floyd & Woolridge, 1992), and steering the organization through difficult times (Beck & Plowman, 2009).
Although the important role of middle managers within an organization has been acknowledged in the literature, there is still a gap in literature on the role of middle managers in the context of IS implementation (Jeyaraj et al., 2006). In IS implementation middle managers usually do not have the direct power to decide over the type of IS. However, they are the one’s that are in direct contact with the end user. Middle managers are in charge of providing support to the employees during the implementation process (Venkatesh & Bala, 2008). To address the gap in literature Paavola, Hallikainen & Elbanna (2017) conducted an explorative study to the role of middle managers in the
managers adjust their behaviors according to the stages of the implementation process and can be considered ‘organizational glue’. The linking role middle managers play in IS implementation could potentially be critical when it comes to the effect of age stereotypes.
Although research has proven that middle management has an influence on the IS adoption of the end user, there is still a lack of understanding about the exact consequences. Additionally, no research has examined the effect of middle managers age stereotypes in the context of IS adoption.
This while it is known that for a successful IS implementation employees depend on support and training, which is usually controlled by management (Venkatesh & Bala, 2008). When management draws upon these negative age stereotypes they might unfairly offer older employees less IS training (OECD, 2011; Wrenn & Maurer 2004), assign them to more routine rather than complex tasks
(Maurer, 2001), and reduce their employee engagement (Boone James et al., 2013), which eventually might not be the best for the individual and organizational outcomes. By taking into account middle managers’ age stereotypes this study investigates possible important factors that influence end users’ IS adoption.
2.4 Employee responses to managerial actions
IS adoption by employee’s has been studied profoundly the past decades. A widely tested and accepted IS adoption model is the Unified Theory of Acceptance Use of Technology (UTAUT) by Venkatesh et al. (2003). The UTAUT model integrated eight IS adoption models into one (Venkatesh et al., 2003). Within the model there are three determinants that have an effect on the end users’
intention to adopt. Thereby, intention to adopt a new IS System has proven to be a valid indication for actual IS adoption. The determinant performance expectancy is defined as ‘the degree to which an individual believes that using the system will help him or her to attain gains in job performance’, effort expectancy is ‘ the degree of ease associated with the use of the system’ and social influence is
‘the degree to which an individual perceives that important others believe he or she should use the new system’ (Venkatesh et al., 2003). Since the introduction of the model it has been tested in many countries and many settings (e.g. Donaldson, 2011; Escobar-Rodriguez & Carjaval-Trujillo, 2013;
Gholami, Ogun, Koh &Lim, 2010; Im, Hong & Kan, 2011; Khechine, Lakhal & Ndjambou, 2016).
Because the proven importance of the model on IS intention to adopt, this research takes the three determinants into account by measuring the intention to adopt.
Research has thus provided valuable insights in the how and why employees make a decision about adoption of IS in the workplace. However, the effect management has on the IS adoption process is less examined but seems a promising factor (Venkatesh & Bala, 2008). Management decisions can
influence perceived ease of use of the system and has been identified as a relevant direction for future research. Jasperson et al. (2005) propose that middle managers have an important role within IS adoption. Namely because they can directly (e.g. direct modification of the IS, change tasks and process) and indirectly (e.g. providing training and resources) influence the IS adoption process.
Thereby management support might influence users’ performance expectancy, effort expectancy and social influence.
Additionally, there has been little attention to how managements’ age stereotypes influence the responsive behavior of the worker, and especially no attention to the effect in the context of IS adoption. Employees experience managerial actions on a daily basis, which has a large influence on their work practices. Managers set rules, deadlines, provide support and thereby guide employees continuously trough their jobs. Boone James et al. (2013) found that unintentional discrimination negatively influences older employees more than younger employees: whereas intentional
discrimination negatively influences younger employees more than older employees. However, there has been no research on how age stereotypes or discrimination influences employees’ actual
behavior and IS adoption in relationship to managerial actions in IS implications. During IS
implementation employees have to adjust their usual behavior and change it to be able to use the new IS. To know more about the actual consequences of age stereotypes of middle management in relationship to worker behavior is valuable to the IS research.
2.5 Conceptual model
Overall, the reviewed literature highlights that there is a gap when it comes to the effects of age stereotypes held by middle management in the IS literature. This research aims at contributing to the IS field by addressing this gap. The quantitative study will aim at answering the research question:
‘What is the effect of middle managers’ age stereotypes on employees IS adoption?’
The hypotheses are derived from the aforementioned literature. It was found that top managers negative age stereotypes’ moderate the relationship between an age diverse climate and the
perception of negative age-discrimination climate (Kunze et al., 2013). We argue that the same effect occurs for middle managers’ age stereotypes since they closely interact with the employees in the context of an IS implementation (Kuratko et al., 2005), and middle managers’ age stereotypes already have shown to negatively impact the employee (Boone James et al., 2013; Posthuma &
other words, age can become a salient factor in an age-diverse climate (e.g. is an important
noticeable categorization factor within a group), which can lead to a negative perception of the age- discrimination climate. As discussed, an IS implementation might be a context in which age can become a salient factor as well (Kanfer & Ackerman, 2004). We argue that because we expect age to be a salient factor in the context of IS implementation, middle managers’ age stereotypes would positively influence the perception of a negative age-discrimination climate. This leads to the first hypothesis:
Hypothesis 1: The presence of middle managers’ negative age stereotypes will have a positive effect on the employees’ perception of a negative age-discrimination climate in the context of an IS implementation.
Furthermore, the perception of a negative age-discrimination climate has proven to have a negative effect on organizational outcomes but has not yet been investigated on a personal IS adoption level.
However, on a personal level there have been found some negative effects of a negative perception of the age-discrimination climate. For example a negative age-discrimination climate on an individual level leads to a higher turnover rate (Kunze et al., 2011; 2013), and worse personal health
(Liebermann, Wegge Jungman & Schmidt, 2013). We argue that because an IS implementation is a context where age might become salient a perceived negative age-discrimination climate will have a negative effect on the behavioral intention to use the system. This leads to the second hypothesis:
Hypothesis 2: The perception of a negative age-discrimination climate will have a negative effect of the behavioral intention of the employee to use the implemented IS system.
All the aforementioned literature is summarized in a developed conceptual model. The model depicts the hypotheses and controls of the study (Figure 1).
Figure 1. Conceptual model
H1+ H2-
Middle managers’
negative IS age Stereotypes
Employees’ perception of negative age- discrimination climate
Employees’ behavioral intention to use IS
system
Control variables employees:
Performance expectancy Effort expectancy
Social influence IS experience
Gender Age Nationality Educational level
Job tenure System experience Control variables middle
managers:
Performance expectancy team
3. METHODS
The aim of this research is to explore the role of middle managers’ age stereotypes on employee IS adoption.
3.1. Research context: implementation of SCRUM
For this study the data was collected at a large Dutch organization (20000 - 25000 employees). The organization decided to change their way of working to an agile way of working. The agile way of working was developed within the IS industry and has received increasing attention the past years (Abrahamsson, Salo, Ronkainen & Warsta, 2002). Agile working entails working in small teams, in short time frames, towards concrete results. The old way of working was long-term project-based way, in which teams sometimes worked towards an end goal one year from now. The agile way of working was chosen as a preferred way of working since it is more flexible, does not have many constraints from management layers and enables the achievement of fast results. The agile approach of working is nowadays implemented in many different types of organizations (Cooper, 2017). The SCRUM method facilitates the agile way of working. The SCRUM method states that the organization consists of small self-steering teams (minimum three, maximum nine members), which is part of a larger group of teams (department). These teams work in sprints in which they take small steps instead of focusing on a deadline a year ahead. Within a team there are different roles. The two leading roles are the product owner (PO) and the scrum master (SM). The PO focuses on the end goal, sets priorities, communicates with own and other teams within the department, and handles the stakeholder management within the organization. The SM facilitates the team processes,
coaches the new way of working, and facilitates the practical aspects. The other team members have executing roles to achieve the goal of the team, and department.
The organization that is investigated made the decision to change to an agile organization with the SCRUM method in 2015. To be able to facilitate the SCRUM method a new software tool system was chosen by the leading IS team. First, higher management formed departments, with one end goal, and subsequently teams. For employees it was possible to state a preference for a department or team or to interview for a certain position in the team. 50 teams were formed in 2016, and another 350 teams in 2017. From the moment the team is officially formed it is mandatory to work with the new software system. The software system is key in the new way of working; all employees need to work daily in the system to note their activities, progress and communicate with their team
members. At the time of the study 7500 people work with the system. However, this study only considers the users that are in an official team and department because then it is mandatory to work with the system. In this study the PO and the SM roles are considered as middle manager roles
because they have the middle managers’ responsibilities such as; working together with the team members, being the link between the team and higher management, and the point of contact in relationship to the implemented software. Additionally, the product owner and scrum master had to apply for these functions, which indicates the more responsibilities they have in comparison with the rest of the team members. All other team members are named end users. Within a department the different PO’s and SM’s have weekly meetings to keep track of all teams within the department. In this study we analyze the average age stereotypes of the SM and PO per department, and the effect it has on the end users within that department. Since the agile way of working is introduced in many different types of organizations nowadays, this has very practical relevance to study.
3.2 Data collection strategy
To test the hypotheses an online questionnaire (Appendix A) was distributed through the systems’
blog to collect the data. The online questionnaire could be administered in English and Dutch. The preference language was set in English, and on the first page of the questionnaire it was stated that participants had the possibility to change this into Dutch. Most questionnaires were already available in Dutch, except the managers negative IS age stereotypes scale and the perception of negative age- discrimination climate scale. These questionnaires were translated with in Dutch by a back-
translation procedure (Brislin, 1970).
3.3 Sample
The online questionnaire was available for five full working days (Monday to Friday) and had a range of 434 teams, with 434 PO’s and 434 SM’s. As a reward for the company’s participation they received an extensive management report. In total 212 participants filled in the questionnaire of which, 165 end users, 30 scrum masters, and 17 product owners. However, because this is a multilevel analysis we could only use data from the departments that at least one middle manager and one end user filled in the questionnaire. Participants within departments that not fulfilled that condition were left out of the study. Eventually, 158 participants fulfilled that condition. In total 147 participants were included in the analysis (11 were left out for not providing all requested information). The
participants were part of a total of 19 departments, of which 25 SM’s and 14 PO’s, which makes a total of 39 middle managers. 108 end users were included. The distribution of middle managers and end users per department can be found in Appendix B. 82 participants answered the questionnaire in English, 65 in Dutch. The descriptive statistics per middle managers and end users can be found in Table 1.
Table 1. Descriptive statistics
Middle Managers End users Gender, n (%)
Men Women
23 (59%) 16 (41%)
69 (64%) 39 (36%)
Age (years), M (SD) 42 (10.89) 32,69 (9.44)
Nationality, n (%) Dutch
Indian Other
33 (85%) 6 (15%) 0
38 (35%) 68 (65%) 2 (2%) Educational level, n (%)
Primary to high school Bachelor (HBO/WO) Master/Phd (WO)
3 (8%) 17 (43%) 19 (49%)
6 (5%) 73 (68%) 29 (27%) Organizational tenure, n (%)
<1 year 1-3 years 3-5 years 5-7 years > 7 years
5 (13%) 11 (28%) 4 (10%) 3 (8%) 16 (41%)
29 (27%) 50 (46%) 8 (7%) 3 (3%) 18 (17%) Experience with system (months), M (SD) 23 (14.36) 14 (14.54)
3.4 Measures
All of the below explained items could be answered on a 7-point Likert scale, unless stated otherwise, ranging from strongly disagree (1) to strongly agree (7).
Middle managers negative IS age Stereotypes. Middle managers’ negative age stereotypes were measured by a three-item scale. These items followed from the age stereotype scale, which was developed by Chiu et al. (2001). Per department the individual scores of the scrum masters and products owners were averaged. This measure is treated as a pure aggregated, because there is no assumption of shared consensus between scrum masters and product owners. A higher score indicates stronger negative age stereotypes.
Perception of negative age-discrimination climate. The negative age-discrimination climate was measured using the 5-item scale that was applied and validated by Kunze et al. (2011). The measure was slightly adjusted so that it would apply to the studied organization in an IS implementation context. The items were answered on a 5-point Likert scale, ranging from strongly disagree (1) to strongly agree (5). A higher score indicates a higher perception of a negative age-discrimination climate.
Behavioral intention to use IS system. Three items measured the intention to use the IS system.
These items were taken from the developed and validated measurement of Venkatesh et al. (2003).
A higher score indicates a higher perception of intention to use the system.
Control variables. In this study there is controlled for factors that potentially bias the results. The general control variables were gender, age, and educational level. Gender was measured using dummy coded items, where 1 = male and 2 = female. Age was measured on a continuous scale where participant filled in their age in years. Tenure was measured on a 5-point scale ranging from working over 1 year (1) at the organization to working more than 7 years at the organization (5). We controlled for both age and tenure, since they might have an effect on the age-discrimination climate (Boehm et al., 2014). Educational level was measured on a 5-point scale from highest accomplished level 1 = primary school education to 5 = University Master/Phd (Boone James et al., 2013). For all users we controlled for mean number of months experience with the system, because not all participants started to work with the system at the same time. Additional control variables come from Venkatesh et al. (2003) UTAUT model. As aforementioned, the UTAUT model has repeatedly proven its significance in the IS adoption literature. The items to measure the different scales were minimally adjusted from the original questionnaire (Venkatesh et al., 2003) to fit the current study.
They were all rated at a seven point Likert scale from completely disagree (1) to completely agree (7).
At middle managers’ level, we controlled for the teams’ performance expectancy. Middle managers are also users of the system and therefore their performance expectancy of the system might influence their behavior, and eventually the adoption behavior of the end user (Venkatesh & Bala, 2008). For the end user we controlled for the four factors, performance expectancy, effort
expectancy, social influence and experience. These factors have proven to have an effect on behavioral intention to use the system.
3.5 Data preparation and analysis
Middle managers’ variables. In this study we look at the effect of the average middle managers’ age stereotypes per department. Therefore, the average values for age stereotypes and performance expectancy team per department were calculated.
Missing data. With a Little MCAR test (Little, 1988) it was tested if there were missing values that need to be adjusted for. The survey required answering all question in order to proceed to the next question so no missing values were expected. This was confirmed by the test that showed that there were no values that needed to be adjusted for.
Multicollinearity. The Variable Inflation Factor (VIF) is checked for each independent variable. The
the tolerance should be above .4. Table 2 shows the possibility of multicollinearity. All VIF’s are below 10, and all tolerance is above .4, which indicates that there are no signs of multicollinearity.
Table 2. Multicollinearity diagnosis
Normality. The distribution of the variables is inspected by a Shapiro-Wilk test (Royston, 1982). For all the variables the p-value is calculated. All variables had a p-value below 0.05. Therefore the null hypothesis is rejected and normality is not assumed. However, per variable the Q-Q-plots were visually inspected since solely relying on the Shapiro-Wilk test can be misleading (Tabachnick & Fidell, 2013). The Q-Q plots (Appendix C), do not all show a perfect straight line. However, since the
deviation does not seem great we continue the analysis and keep in mind the potential effects of non-normality when problems are encountered.
Variation. In order to check for variation between the average age stereotypes per department a one-way between subjects ANOVA was conducted. A one-way between subjects ANOVA assumes that there are no differences between groups. The one-way ANOVA showed a p-value < .05, which indicates that there are differences between the departments in the average age stereotypes and allow to continue with the multilevel analysis.
Data analysis. The statistical package IBM Statistical Package Social Sciences (SPSS) 25 was used in combination with the extension SPSS AMOS 25 for the analysis. First, we tested the reliability of the measures by calculating reliability and validity measures. Second, the hypotheses were tested applying structural equation modeling (SEM) techniques. This technique was chosen because it allows testing the relationship between observed and latent variables (Hoyle, 1995). The recommendation of Anderson and Gerbing (1988) to test the measurement structure and subsequently the structural relationship was followed. First, an exploratory factor analysis was performed to uncover the underlying structure of the large variable set (Norris & Lecavalier, 2009).
Second, a confirmatory factor analysis was performed to test whether the constructs were consistent
Variable Tolerance VIF
Age Stereotypes
MM’s Performance expectancy team
Perception of negative age-discrimination climate Performance expectancy
Effort expectancy Social Influence
0.910 0.937 0.755 0.421 0.447 0.514
1.099 1.067 1.325 2.376 2.238 1.945
with the hypothesized measurement model (Kline, 2010). Consequently, the model fit and common method biased was performed. Even though the sample size exceeded the threshold of 100 (Hair, Anderson, Tatham & Black, 1998), SPSS AMOS did not allow the performance of the aforementioned tests. Our sample size did not reach the threshold of 150 (Bollen, 1989), which might be the reason for not being able to continue the SEM analysis. However, multilevel models can also be tested with regression analysis in SPSS. In the result section the sequence and decisions are explained in more detail.
4. RESULTS 4.1 Descriptive statistics
The correlation table (Table 3) shows that the variable age stereotypes does not seem to be correlated with the perception of a negative age-discrimination climate or intention to adopt the system. Neither does a negative age-discrimination climate seem to be related to intention to adopt the system. Regarding the control variables, performance expectancy, effort expectancy, social influence, IS experience, experience with the system, age, nationality, and tenure to be significantly related to our outcome variables. Neither gender, nor educational level, nor middle managers’
performance expectancy team are significantly correlated to any of the outcome variables.
Table 3. Bivariate correlation of variables
Variable 1 2 3 4 5 6 7 8 9 10 11 12 13
1. Perception of negative age-discrimination climate 2. Intention to use system 3. Age stereotypes
4. Performance expectancy team
5. Performance expectancy 6. Effort expectancy 7. Social Influence 8. IS Experience
9. Experience with system (months)
10. Gender 11. Age 12. Nationality 13. Educational level 14. Tenure
.02 .07 -.01
.49*
.32*
.33*
.05 -.15
.03 -.47**
.54**
.12 -.45**
-.05 -.02
.22*
.28**
.39**
.33**
.24*
-.13 .02 -.03 .18 .01
-.24*
.13*
.10 .01 -.10 .15
-.13 -.26**
.13 .02 -.25**
.02 .03 -.01 -.15 -.16
.10 -.15 .13 -.10 -.08
.67**
.61**
.14 -.03
-.03 -.56**
-.56**
.13 -.46**
.64**
.14 .07
.12 .41**
.48**
.03 -.34**
.19*
.04
-.01 -.27**
.33**
.19*
-.31**
.24*
-.27**
.16 .04 .04 .08
-.11 .20*
-.11 .16 -.00
-.22*
.11 -.10 -.00
-.57**
.01 .67**
-.04
-.50** -.32**
Note. *p < .05. **p < .01. ***p < .001
4.2 Exploratory Factor Analysis
To reduce a large number of observed variables to a small number of factors exploratory factor analysis is the preferred statistical method (Hadi, Abdulla & Sentosa, 2016). First, we performed a Keizer-Meyer-Olkin (KMO) measure of sampling adequacy and a Bartlett’s test of sphericity to assess the factorability of the data. A KMO sampling of adequacy of .05 is considered the minimum (Kaiser, 1974); the data shows a KMO of .76. The Bartlett’s test of sphericity was significant (p < .05) (Bartlett, 1954). Therefore, the data is suitable for factor analysis (Tabachnick & Fidell, 2013).
The items were analyzed by the Maximum Likelihood as extraction method and by Promax as the rotation method. Promax is an oblique rotation which should give a more accurate, and perhaps more reproducible, solution than orthogonal rotation (Osborne & Costello, 2009). After the first run four items were removed because of cross-loading with other factors (‘If I use the system it will increase my chances of getting a raise’ from performance expectancy, which cross-loaded on
perception of negative age-discrimination climate; from social influence the items ‘my team is helpful in the use of the system’ and ‘In general, our company supports the use of the system’, which loaded on effort expectancy were deleted, and from middle managers’ performance expectancy team ‘If my team uses the system my team increases its chance of getting a raise’ ,which cross-loaded on
performance expectancy was deleted). Communalities were all above .3 (Table 4). For good
convergent validity is above the threshold .3 (Hair et al., 1998). All the items are above .4 so there is good convergent validity. There are no cross-loadings, which shows discriminant validity.
Additionally, the correlation matrix (Table 4) does not show factors that exceed the threshold of .7, which confirms good discriminant validity.
Table 4. Pattern matrix
1 2 3 4 5 6 7
Perception of negative age-discrimination climate
Q1 Q2 Q3 Q4 Q5
.897 .962 .965 .935 .915 Intention to use system
Q1 Q2 Q3
.837 .972 .893 Age stereotypes
Q1 Q2
.892 .945
Q3 .882 Performance expectancy
team Q1 Q2 Q3
.775 .833 .938 Performance expectancy
Q1 Q2 Q3
.844 .922 .888 Effort expectancy
Q1 Q2 Q3 Q4
.780 .857 .929 .885 Social influence
Q1 Q2
.908 .941
Table 5. Component correlation matrix
Factor 1 2 3 4 5 6 7
1. Perception of negative age-
discrimination climate 1.00
2. Intention to adopt system .341 1.00
3. Age stereotypes -.035 -.038 1.00
4. Performance expectancy team .275 .402 -.075 1.00
5. Performance expectancy .316 .609 -.021 .421 1.00
6. Effort expectancy .071 .097 -.280 -.019 .082 1.00
7. Social influence .007 .199 .039 .242 .277 -.043 1.00
4.3 Confirmatory Factor Analysis
Validity and reliability checks. To assess the validity and reliability of the variables multiple tests were performed. First, the construct reliability (CR) was assessed. Construct reliability above .70 is a sign of good reliability (Hair et al., 1998). In our sample the variables scored above .70 CR (Table 6). Second, the convergent validity was checked with the average variable explained (AVE), which is a summary measure of convergence of a set of items representing a construct. For convergent validity a score above .50 AVE is required. All the variables pass this threshold (Table 6).
Table 6. Validity and reliability table
When the square root of the AVE is greater than the inter-construct correlations it is a sign of good discriminant validity (Hair et al., 1998). Table 7 shows that for all the variables this is the case.
Variable CR Cronbach’s alpha AVE
Perception of negative age-discrimination climate
Intention to use IS system Age stereotypes
Performance expectancy team Performance expectancy Effort expectancy Social Influence
.965
.897 .907 .871 .894 .903 .910
.97
.89 .87 .72 .87 .90 .91
.848
.745 .771 .719 .744 .699 .835
Table 7. Factor correlation matrix
Note. Square root of the AVE in bold on the diagonal
Measurement model fit. The model was assessed on the model fit. Hair et al. (1998) advice to state
the following measurement fit indicators; χ2
/df, CFI and RMSEA. Based on the results we assume that the model has a good fit (Table 8).
Table 8. Measurement model fit
Measure Threshold Score
χ
2
/df < 3 good; < 5 sometimes permissible 1.398
CFI > .95 great; > .90 traditional .961
RMSEA < .05 good; .05 - .10 moderate .06
Note. Hu & Bentler (1999) thresholds
Common method bias. Podsakoff, MacKenzie, Lee & Podsakoff, (2003) states that to test for any variance that is attributable to the measurement model instead of the constructs, a common method bias test has to be performed. This procedure required adding a dummy latent factor variable to the model. However, AMOS did not allow us to perform this step with our sample. A possible explanation could be that we did not reach a threshold of 150 participants (Bollen, 1989).
In consequence, the initial approach, structural equation modeling with AMOS, was revised and
Factor 1 2 3 4 5 6 7.
1. Perception of negative age-discrimination climate
.921
2. Intention to adopt system .013 .863
3. Age Stereotypes .040 -.084 .878 4. Performance expectancy
team
.034 -.095 -.291 .848
5. Performance expectancy .384 .162 .081 .035 .862
6. Effort expectancy .335 .255 .071 -.008 .684 .836
7. Social influence .302 .215 -.013 .004 .462 .510 .914
suitable for multilevel analysis because it allows parameters to vary at more than one level and they are generalizations of linear regressions (Raudenbush & Bryk, 2012; Tabachnick & Fidell, 2013).
4.5 SPSS Regression
To test the hypothesis multiple regression analysis was performed in two steps. First, the multilevel hypothesis was analyzed. Consequently, the other variables that might have an effect on the perception of negative age-discrimination climate and intention to use the system were analyzed by multiple regressions.
Multilevel model regression. To run a multilevel model we use the mixed model command in SPSS.
First, we examined the effect of age stereotypes on perception of negative age-discrimination climate when controlled for performance expectancy team. When the random intercept, which allows the intercepts to vary in the model, is added to the model, SPSS shows that convergence has not been achieved. This statement is in line with the finding of no difference in the χ
2
between the first model without including the random intercept and the second model with including the random intercept. This shows that the differences between the groups, even though the one-way ANOVA did show significant results, are not large enough to calculate a multilevel effect (Maas & Hox, 2005).
Second, we examined whether there was a direct effect of age stereotypes on intention to use the system when controlled for performance expectancy team. Also in this analysis, there is no significant difference in the χ
2
between the model without and the model with including the random intercept. So in conclusion, the differences between the average of the middle managers’ age stereotypes between departments is too small to calculate a multilevel analysis. This might also explain why SPSS AMOS was not able to continue with the analyses. Overall, we will not be able to analyze the effect of middle managers’ age stereotypes on the perception of negative age-
discrimination climate or on the intention to adopt the system. Hence, we will continue with the study by analyzing the data on a single level.
Multiple regressions. From now on we focus solely on the single level data, which results in the fact that we do not only have to include end users of whom also a middle manager of their department filled in the questionnaire. This allowed us to include all the end users that filled in the questionnaire (N = 148). Because we only study the single level we will examine what the influence of our control variables are on employees perception of negative age-discrimination climate and on employees’
behavioral intention to use the IS system. To analyze the data we performed two multiple
regressions; one multiple regression on the perception of negative age-discrimination climate and
another multiple regression on the behavioral intention to use the IS system. The summary statistics of the variables can be found in Appendix D.
Assumptions. The assumptions for multiple regression are checked before continuing. Two outliers are removed and there are no sign of multicollinearity because all variables had a correlation below .7, a VIF below 10 and a tolerance above .4 as shown in Table 9 (Neter et al., 1990). The assumption of normality of the outcome variable was not met because the distribution was negatively skewed.
The negatively skewed distribution was adjusted for by a log transformation. After the
transformation the assumption of normality is met. Despite the transformation the scatterplot shows some signs of non-linearity and homeoscedasticity. However, since the data was already
transformed, we will continue with the regressions for now.
Table 9. Multicollinearity diagnosis
Multiple regressions on perception of negative age-discrimination climate. To examine the effect of the variables on the perception of negative age-discrimination climate all the variables were entered in the regression. Results show that nationality is the only variable that significantly explains variance of the perception of a negative age-discrimination climate (Table 10). All the variables together explain 30% of the variance (R2 = .299).
Variable Tolerance VIF
Perception of negative age-discrimination climate Performance expectancy
Effort expectancy Social Influence
0.847 0.425 0.485 0.695
1.182 2.352 2.061 1.438
Table 10. Regression on perception of negative age-discrimination climate
Note. SE = Standard error. *p = <.05,**p = <.01, ***p = <.001.
Kunze et al. (2011; 2013) discussed that age might influence the perception of negative age- discrimination climate. Therefore, we check for a moderating effect of age on the significant correlated variables by one-by-one adding the interaction to the model (Appendix E, table 14 through table 18). The variables that are significantly correlated with age-discrimination climate in the sample are performance expectancy (.383), effort expectancy (.218), social influence (.225), age (-.405) and nationality (.347) and tenure (-.429). All variables have a non-significant interaction effect with age on perception of negative age-discrimination climate (p > .05). However, to visualize the results we computed a group (group 1) for the top 25% youngest people in the sample, and a group (group 2) for the top 25% eldest people in the sample. The results are plotted in the graphs
(Appendix E, graph A trough graph F), which displays the direction of relationship, although it is not significant.
Multiple regressions on intention to use the IS system. To examine the effect of the variables on the intention to use the system all variables were entered in the regression. Results show that
experience with the system is the only variable that significantly explains variance of the intention to use the system (Table 11). All the variables together explain 17% of the variance (R2 = .166).
Variable B SE β
Performance expectancy Effort expectancy Social Influence Experience with IS Gender
Age Nationality Educational level Tenure
Experience with system R2
N
.148 -.140 .090 .050 .076 -.013 .276 .114 -.141 -.006
.107 .099 .064 .059 .153 .011 .106 .126 .075 .005
.299 146
.167 -.149 .126 .069 .038 -.129 .212*
.071 -.201 -.091
Table 11. Regression on intention to use the system
Note. SE = Standard error. *p = <.05,**p = <.01, ***p = <.001.
Table 3 shows that performance expectancy, effort expectancy, and social influence, all significant correlated to both perception of negative age-discrimination climate and intention to use the system. This might indicate a mediation effect of these variables on intention to use the system.
However, because perception of negative age-discrimination climate is not significantly correlated to nor significantly explains any variance of intention to use the system, no mediation can be
determined (Preacher & Hayes, 2004).
To assess the possible moderating effect of age and perception of negative age-discrimination climate on intention to use the system the interaction effect was added to the regression, which showed non-significant results (p > .05) (Appendix E, table 19). However, to test the direction of the relationship we plotted the interaction effect for the youngest (group 1) and the eldest group (group 2), see Appendix E (graph F).
Variable B SE β
Perception of negative age-discrimination climate Performance expectancy
Effort expectancy Social Influence Experience with IS Gender
Age Nationality Educational level Tenure
Experience with system R2
N
.008 -.005 .039 .028 .003 -.019 -.002 -.008 .049 .020 .002
.020 .025 .023 .015 .014 .035 .003 .025 .029 .017 .001
.166 146
.037 -.026 .200 .189 .020 -.045 -.098 -.029 .145 .135 .178*
5. DISCUSSION
This final chapter discusses the results described in chapter 4. First, the theoretical implications and the future research directions are described. Secondly, we will go into more depth concerning managerial implications. Thirdly, limitations of the study are discussed. Finally, the discussion ends with an overall conclusion.
5.1 Theoretical implications and future research directions
This research aimed at looking at the effect of middle managers’ age stereotypes on individual IS adoption by answering the main research question:
‘What is the effect of middle managers’ age stereotypes on employees IS adoption?’
As stated in the result section the acquired data was not sufficient to be able to analyze the multilevel stated research question because the number of end users that could be linked to a middle manager was too small. In the limitations section further explanation for the non-sufficient data will be discussed. However, on an individual level it was possible to examine the variables that might affect the perception of negative age-discrimination climate and the intention to use the system. Below we first discuss the two multiple regressions that were performed and the direct effects found. Subsequently, the non-significant findings that might indicate interactions effects that lead to suggestion for further research are discussed.
The first multiple regression examined the relationship between the measured variables and the perception of negative age-discrimination climate. The variance of the perception of negative age- discrimination climate was solely significantly explained by nationality. The significance of nationality in our sample indicates cultural differences between the Netherlands and India. This is in line with the age stereotype theory of Marcus and Fritzsche (2016). They classify India as a collectivistic tight culture, where it is not accepted to deviate from the group norm, and the Netherlands as an individualistic loose culture, where it is accepted to deviate from the group norm. They argue that because of the strong social norms in a tight collectivistic culture, those cultures are the most age- discriminatory. Individualistic loose cultures are argued to be the least age-discriminatory because of weaker social norms. In our sample on average Indian participants rate the perception of negative age-discrimination climate more negatively than Dutch participants, which in is line with their theory.
The original hypothesis was that middle managers’ age stereotypes would increase the perception of negative age-discrimination climate. We were not able to examine the relationship between middle
managers’ age stereotypes and the perception of a negative age-discrimination climate. However, we can compare our results with previous studies done on the perception of negative age-
discrimination to gain a better understanding of the perception of negative age-discrimination climate in our sample. It is remarkable that when we compare our mean of the perception of a negative age-discrimination climate (M = 3.20) with the mean of the study of Kunze et al. (2011) (M = 2.13) Kunze et al. (2013) (M = 1.83), it shows that our mean is a lot higher. An explanation of this could have to do with age diversity in the context of an IS implementation. Namely, the perception of a negative age-discrimination climate has been proven to dependent on the level of age diversity in a company. Kunze et al. (2013) showed that in an organization with greater age diversity and high age salience (e.g. age is an important noticeable categorization factor within a group) the perception of a negative age-discrimination climate is higher than in organization with smaller age diversity. These findings were in line with the social identity theory (Tajfel & Turner, 1986) and the social-
categorization processes theories (Turner, 1985). These theories state that in age-heterogeneous groups age can lead to sub grouping. As discussed in the introduction, when age becomes a factor or social categorization, and people classify themselves within a group of the same age category, sub grouping occurs. In line with these theories we would expect that our mean is high because in our sample we have high age diversity in combination with the fact that age is a salient factor of group categorization. Our sample does indeed have a greater standard deviation (SD = .97) than the standard deviations in the study of Kunze et al. (2011) (SD = 0.66), and Kunze et al. (2013) (SD = 00.49), which would indicate higher age diversity in our sample compared to the other studies. In the introduction we hypothesized that the IS implementation context could be a context in which age becomes a salient factor (Kanfer & Ackerman, 2004), and the high mean on perception of negative age-discrimination climate would support that hypothesis. However, we were not able to statistically examine this because we could not control for age diversity within the organization and did not have other organizations, without an IS implementation context, to compare our results with. Subsequent studies should examine the role of age stereotypes, age diversity and age salience in relationship to the perception of negate age-discrimination climate. Furthermore, they should investigate whether age indeed becomes more salient in the context of IS implementation by comparing different organizations.
The second multiple regression analyzed the relationship of the independent variables with the dependent variable intention to use the system. The variance of the intention to use the system was solely significantly explained by experience with the system. These findings are not in line with
which has proven its statistical significance (Khechine et al., 2016). In many studies performance expectancy, effort expectancy, and social influence have shown to directly relate to the intention to use the system (Venkatesh et al., 2016). However, in our study we do not see any significant relationships of these variables. There are multiple possible explanations for this. First of all our sample has an extremely high mean of intention to use the system (5.96/7), which might indicate that we were too late in the adoption process. Venkatesh et al. (2013) stated that in the post-
adoption phase, it is more about continuance intention instead of initial intention (Venkatesh, Thong, Chan, Hu, & Brown, 2011). Because we did not expect large difference in the number of months of experience with the system we only measured initial intention. However, it might be that the intention to use the system was not the appropriate dependent variable to measure at this stage in the implementation for most participants that filled in the survey. Second, our sample probably suffers with a larger selection bias. We had a low response rate, which might have led to the fact that only very active users of the system filled in the questionnaire, which biased the results (Furnham, 1986). Overall, this are probably the main causes that led to too little variance in the dependent variable intention to use the system in our sample, which explains that the expected UTAUT variables were not significant. Further explanation of the possible biases is explained in the limitations sector.
Moreover, we expected to find a significant relationship between the perception of a negative age- discrimination climate and the individual level of intention to adopt the IS system. Our hypothesis was drawn from previous research that showed that a negative perception of age-discrimination climate can lead to negative organizational outcomes and negative personal outcomes (Kunze et al., 2011; 2013; Liebermann et al., 2013). We did not find a significant result for this relationship. Also this might be explained by the little variance of the intention to use the system due to the timing in the adoption process in combination with a selection bias. However, although we did not find many significant results, we did find interesting trends that with more research could potentially lead to new insights regarding to age and the age-discrimination climate in relationship to IS adoption behavior. This is explained below. First we examine the possible moderating variables on intention to use the system, second we examine the possible moderating variables on the perception of negative age-discrimination climate.
To investigate the relationship between the perception of negative age-discrimination climate and intention to use the system further we first examined the role of age on the intention to use the system. Previous research has found it difficult to prove the exact effect of age as a moderator in the UTAUT model. Venkatesh et al. (2003) originally hypothesized that performance expectancy had a greater influence on intention to use the system for younger users. These thought were based on
