The introduction of care robots as a leadership challenge in home care facilities in Finland

Abstract Aims This paper analyses the factors that influence home care employees’ intention to introduce robots. Background The introduction of different kinds of care robots is a topical issue in elderly care now and in the near future. Methods Cross‐sectional research conducted through a questionnaire. The survey data (N = 162) were collected in five locations around Finland in 2019. The analysis was carried out by regression analysis, Sobel test and by Hayes’ bootstrapping method. Results The results show that self‐efficacy is pivotal in the willingness to introduce care robots. Employees’ age increases the enthusiasm to introduce robots but reduces self‐efficacy. Work engagement does not correlate with self‐efficacy or behavioural intention related to the introduction of care robots. Conclusions The present paper reveals the significance of attitudes, cognitive factors and age in the adoption of care robots in home care facilities. Practical implications It is important to pay attention to supporting the employees’ sense of technology management and the construction of a robot‐positive atmosphere when introducing care robots, and the development of skills of older employees and employees with a lower educational level should be supported.


| Cognitive approach to the introduction of care robots
This study approaches issues of nursing leadership from the perspective of the adoption of care robots, employing a framework of cognitive social psychology. Especially, the concept of behavioural intention, which is an individual's willingness to exert effort to perform the target behaviour, plays an important role in models that explain the adoption of healthcare technology (Holden & Karsh, 2010) and care robots (Alaiad & Zhou, 2014;Rantanen et al., 2018;Vichitkraivin & Naenna, 2019). According to the theory of planned behaviour, behavioural intention is affected by self-efficacy (perceived behavioural control), subjective norms and attitudes (Ajzen, 1991). Bandura's (1982) concept of self-efficacy refers to a person's sense of control over a behaviour. In the context of the adoption of care robotics, the concept represents a person's belief that he/ she will learn to use the robots as desired. According to Latikka et al., (2019), robot use self-efficacy is associated with the acceptance of using robots. There is a significant relationship between nurses' self-efficacy and their age, occupation and educational level (Zaki, 2016); however, Lin (2016) noticed that there were no significant differences between groups who were trained and not trained according to items related to self-efficacy.
A subjective norm is a person's perception of the degree to which other important people approve or disapprove of the target behaviour (Ajzen, 1991;Holden & Karsh, 2010). The relationship of subjective norms and the adoption of care robots have been little studied, but preliminary results suggest an existence of dependence between them (Rantanen et al., 2018). In contrast, in studies related to the acceptance of IT technology in health care, results have been contradictory (Holden & Karsh, 2010). Nomura et al., (2006) examined negative attitudes and fears concerning robots. Attitudes have been studied from the perspective of practical care and the promotion of the independent living of elderly people. In these studies, care robots are considered to be useful in certain tasks in home health care, such as measuring vital signs, setting medication reminders, communication and in case of a fall or other emergency (Alaiad & Zhou, 2014;Broadbent et al., 2012;Mast et al., 2012).
A person's attitude towards his or her own work can also be assumed to be a relevant factor for various innovative development processes. Thus, in this study, a research design based on Bandura´s (1982) concept of self-efficacy and Ajzen's (1991) theory of planned behaviour has been complemented by Schaufeli et al.,'s (2006) concept of work engagement which means a positive state of emotion and motivation characterized by vigour, dedication and absorption.
This has connections, for example to job performance, innovativeness, health and age (Schaufeli et al., 2006;Hakanen, 2009). In nursing, patient safety requires the engagement of nurses in their practice that emerges from settings of autonomy and results in safer, cost-effective patient outcomes (Bargagliotti, 2011). Relatedly, there is a positive relationship between age, years of nursing experience and work engagement (Keyko et al., 2016).

| Significance of sociodemographic factors
From the perspective of care robot adoption and deployment intentions, certain background variables also play a key role. Flandorfer (2012) stated that the effect of age, gender, technological experience and education is significant in the acceptance of assistive robots. However, studies on the effects of age have been contradictory (cf. Backonja et al., 2017;Flandorfer, 2012;Syrdal et al., 2013).
According to Ezer et al., (2009), age did not have a significant effect on the characteristics that participants attributed to their robot, when technology and robot experience were accounted for, and their study suggests younger and older adults with comparable experience with technology will have similar expectations of robots as performance-oriented machines. According to Kuo et al. (2009), differences related to gender and educational level were shown to be far less important, although people with a lower educational level expressed more negative feelings towards robots than people who had achieved a higher degree of education. Thus, previous studies suggest that certain cognitive factors and factors related to employees' backgrounds have a significant impact on their willingness to adopt a new technology, which raises the question of how these factors are taken into account in the management of the technological change process.

| General perspectives on managing the robot deployment process
Nursing leaders are pivotal when implementing care robots in home care, and it is also crucial to involve employees in the process (De Brún et al., 2019). Jones-Roberts (2008) noticed that employees engaged in the development of a project appreciated the opportunity to create new approaches. However, when solutions had already been developed by other employees, they did not achieve the same level of commitment or ownership. Nursing leadership needs to conquer obstacles and cope with challenges when facing changes in nursing (Holm & Severinsson, 2014), and nurse leaders need to be aware of the fact that they have an influential role concerning the implementation of new practices and facilitating a willingness to adapt to new technologies (Bianchi et al., 2018). It is, therefore, critical to ensure that employees work at full capacity and with the support of the leadership, who, in turn, are supported by technology (Ganann et al., 2019). Gifford et al., (2013) found that individual level barriers were related to nurses' skills and attitudes when implementing new practices. In their study, the perception that there was no reason to change was the second most identified issue related to nurses' beliefs that implementing the recommendation would not have an influence on patient care or outcomes (Gifford et al., 2013). Maurits et al., (2015 found that employers should focus primarily on the appreciation of nursing staff by senior leadership. Given that an ageing population poses challenges for aged care, leadership is one important aspect in meeting those challenges (Backman et al., 2017). Thus, nursing leaders should be aware of the different factors which may have impact on employee's willingness to adapt new innovations like care robots, and according to previous research, behavioural intention, self-efficacy, subjective norm, attitudes, work engagement and sociodemographic factors may influence how employees accept and adapt to change or new innovations.

| Aim and hypotheses
This study focusses on Finnish Registered Nurses, licensed vocational nurses and other social and healthcare workers. The aim of the study was to analyse employees' intentions to introduce care robots in home care.
First, studies based on the theory of planned behaviour (Ajzen, 1991) show that self-efficacy, attitudes and perceived subjective norm are important factors in the adoption of new health technologies (e.g. Alaiad & Zhou, 2014;Holden & Karsh, 2010).
Second, previous studies suggest that age, level of education and experience with technology are important factors in technology acceptance (e.g. Ezer et al., 2009;Kuo et al., 2009Flandorfer, 2012Syrdal et al., 2013;Backonja et al., 2017). Third, previous studies show that work engagement is important for innovativeness (Schaufeli et al., 2006;Hakanen, 2009). Based on these results, this study asks How do the factors of self-efficacy, attitudes, perceived subjective norm, work engagement, age, level of education and experience with technology affect the willingness of home care employees to introduce care robots in their work?
In addition to direct effects, indirect effects related to selfefficacy are considered. The first hypothesis based on Ajzen´s theory of planned behaviour is. H1: Home care employees' behavioural intention depends on self-efficacy related to the use of care robots, attitudes towards the usefulness of robots in various home care tasks and perceived norms in the work community.
The second and third hypotheses are related to the significance of background variables.
H2: Employees' age is associated with behavioural intention and self-efficacy related to the use of care robots in home care. H3: Employees' educational level and previous experience with using welfare and health technology are associated with behavioural intention and self-efficacy related to the use of care robots in home care.
The study further examines the effect of work engagement on behavioural intention and self-efficacy H4: The level of employees' work engagement is associated with behavioural intention and self-efficacy related to the use of robots in home care.

| Settings
The study is a cross-sectional research, which investigates the adoption of care robots. The research design is based on Bandura´s (1982) concept of self-efficacy and Ajzen's (1991) theory of planned behaviour and a previous Finnish study the adoption of care robots (Rantanen et al., 2018). The design has been complemented by the concept of work engagement, and a detailed analysis of background variables (age, educational level and previous experiences with using welfare and health technology) and their mediation effects.

| Sample
The study aimed to comprise a comprehensive regional sample. Therefore, municipalities from the Helsinki metropolitan area and other regions in Southern Finland, Eastern Finland, Western Finland and Northern Finland were included in the sample.
Adequate regional coverage was achieved by looking at five municipalities (Table 1). One larger city (200,000 inhabitants), two medium-sized towns (50,000-100,000 inhabitants) and two rural municipalities participated in the study. All home care employees in these municipalities (a total of 1,128 people) were included in the sample. In addition to home visiting employees, Registered Nurses, physiotherapists and other social and healthcare workers working in the offices of home care facilities were included in the sample, as well as supervisors who were involved in client matters. Administrative supervisors and office workers were excluded from the sample.
The total number of respondents was 162, but due to missing values in some of the study variables (Table 2), the analytic sample size ranged from 152-162, depending on the analysis being carried out. The sample size can be considered adequate from the point of view of the research design and the method used. The main method of the study is linear regression analysis, and the number of independent variables is at most seven.

| Instrument
The questionnaire contained a total of 72 questions. Background questions were related to gender, age, locality, education and experiences with using health and welfare technology. Most of the questions were Likert-scale statements (1 = totally disagree to 5 = totally agree). Measures related to the concepts of Ajzen (1991) and Bandura's (1982) theories were based on a study conducted  Table 2). The questions were based on an earlier Finnish measure (Rantanen et al., 2017(Rantanen et al., , 2018 and related to promoting safety (four items, α = 0.865 in the original study), practical assistance (four items, α = 0.824 in the original study) and guidance (three items,

| Data collection
The survey data were collected in five locations around Finland in

TA B L E 2 Sum variables
were contacted in each municipality and practical arrangements for administering the survey were agreed upon. The information letter and a link to the electronic questionnaire were sent to all municipal home care employees' email addresses by the supervisors.
Employees responded to the questionnaire during their working hours, while they were at the home care facility's office.

| Analysis
The data were analysed using SPSS Statistics. First, the data were revised, and two incomplete forms were rejected. The sum variables were formed from the individual items by averaging, and their internal consistency was examined by Cronbach's alpha coefficient.
The normality of the distributions was investigated graphically and by skewness and kurtosis coefficients. The distributions were close to normal. Thus, it was decided to carry out further analyses in a parametric manner. Although the issue is controversial, the use of parametric methods for sum variables based on Likertscale items is considered acceptable by many researchers (Clason & Dormody, 1994;Norman, 2010). The distribution of UWES was negatively skewed, and the scales of its items were converted to 5-point by combining response options "never" and "a few times a year," as well as options "once a month" and "a few times a month." The actual statistical analyses were carried out using the Pearson Product-moment coefficient and linear regression analysis with the Enter method. Before regression analyses, the normality of the residual distributions and the linearity condition were checked graphically, and the multicollinearity between the independent variables was investigated by the VIF coefficients.
The effect of age was also examined by one-way analysis of variance, so that the respondents were divided into four age groups.
Before the analyses, the assumption of homogeneity of variances was checked by Levene's test. Mediation effect was analysed by the Sobel test.
The significance of previous experience was investigated nonparametrically using, at first, Spearman's Rho coefficient. For regression analysis, three dummy variables were generated by combining response options 1, 2 and 3 into 0 and options 4 and 5 into 1. The effect of an employee's educational level was analysed by t test and by regression analysis using a dummy variable (1 = high education degree, 0 = no high education degree). Indirect effects of previous experience and educational level were analysed by Hayes' Bootstrapping method.
Mediation analysis was made using PROCESS macro 3.3 for SPSS, and 10,000 bootstrap samples were used. The indirect effect was considered significant when the upper and lower bounds (95% confidence intervals) did not contain a value of zero.

| E THI C AL ISSUE S
The study was conducted according to the instructions of the The

| Respondents
Respondents' average age was 43 years, and the majority of respondents (94%) were women. Most respondents (71%) were licensed vocational nurses, and 14% were Registered Nurses. Half of the respondents had worked in home care facilities at least 10 years.
The distributions of respondents' gender, profession and region are close to national values (Table 1).

| The effect of self-efficacy, attitudes and subjective norms
Employees' behavioural intention to introduce care robots to home care facilities correlates strongly with self-efficacy, perceived subjective norms and attitudes. These three independent variables based on the theory of planned behaviour have a significant correlation with each other. In contrast, work engagement does not correlate significantly with any other sum variable. A person's age, in turn, relates to self-efficacy, their subjective norms and work engagement (Table 3).
According to regression model 1 (Table 4), self-efficacy, attitude and subjective norms explain behavioural intention. The effect of self-efficacy is the strongest, but all these coefficients are significant (p < .001). The model explains behavioural intention very well (R 2 = .70) ( Table 4).

| Significance of age
Behavioural intention and age do not correlate significantly, r(160) = .005, p = .946, and there is no significant difference in behavioural intention between different age groups, F(3, 158) = 0.087, p = .967. In contrast, age and self-efficacy correlate significantly with each other r(160) = −0.214, p = .006), and there are significant differences between age groups in self-efficacy (F(3, 158) = 3.41, p = .019). According to regression model 2 (Table 4), the effect of employees' age is significant. Furthermore, according to regression model 1 in Table 5, the effect of age on self-efficacy is significant.
Mediation analysis reveals that age has both a direct effect (p < .001) and an indirect effect on behavioural intention, and selfefficacy is the mediating variable here (-z = 2.76, p = .006). The direct effect of age is positive, the indirect effect is negative, and the total effect is not significant when the effects cancel each other out ( Figure 1).

| Significance of educational level
According to the results, employees' willingness to introduce robots depends on a person's educational level. Behavioural in-  (Table 4). This is explained by the fact that self-efficacy related to the use of robots, t(157) = 2.856, p = .005, is stronger in the group of graduates. According to regression analysis (Table 5), the effect of educational level on self-efficacy is significant. Indeed, it seems that educational level has an indirect effect on behavioural intention through self-efficacy. According to bootstrapping analysis, the indirect effect coefficient was 0.332 (95% CI = 0.129-0.532).
Educational level has also little direct effect on behavioural intention (c´ = 0.207, p = .047), but the indirect effect is stronger (62% of total effect) ( Figure 1).

| Significance of employees' previous experience with using welfare and health technology
Employees' previous experience with using welfare and health technology is significantly associated with the willingness to introduce care ro- According to regression analysis, none of the dummy variables related to experience have a significant effect on the intention to introduce care robots (Table 4), and only the existence of good experiences has a significant effect on self-efficacy ( of previous experience (c´ = 0.013, p = .738) or good experience (c´ = 0.066, p = .224) are not significant (Figure 1).
According to regression analysis, work engagement does not explain variation in behavioural intention (Table 4) or self-efficacy (Table 5).

| D ISCUSS I ON
The results are consistent with the theory of planned behaviour (Ajzen, 1991) and previous studies on the adoption of new technology in health care (i.e. Alaiad & Zhou, 2014;Holden & Karsh, 2010).
The first hypothesis that "home care employees' behavioural intention depends on self-efficacy related to the use of care robots, attitudes towards the usefulness of robots in various home care tasks and perceived norms in the work community" is supposed.
Consistent with previous studies, employees' self-efficacy (Rantanen et al., 2018;Latikka et al., 2019) and the social influence of familiar people (Alaiad & Zhou, 2014) are important factors in terms of the adoption of care robots in home care.
Previous studies on the effect of age on attitudes towards robots and the willingness to use robots have been contradictory (Flandorfer, 2012;Syrdal et al., 2013). The hypothesis that "employees' age is associated with behavioural intention and self-efficacy related to the use of care robots in home care" is supposed. However, the dependence is very complex. Ezer et al. (2009)  nursing leaders should take into consideration that self-efficacy is pivotal in the willingness to adopt care robots and that there are significant differences between age groups in self-efficacy when they become involved in the introduction of robots (De Brún et al., 2019).
The third hypothesis "employees' educational level and previous experience with using welfare and health technology are associated with behavioural intention and self-efficacy related to the use of care robots in home care" is partially supposed. The finding is consistent with previous studies (Ezer et al., 2009;Flandorfer, 2012 In contrast, the hypothesis that "the level of employees' work engagement is associated with behavioural intention and selfefficacy related to the use of robots in home care" is not supposed. Employees' general enthusiasm for the work is not connected to willingness to introduce robots. This is a surprise, as previous studies suggest that work engagement is linked to innovation (Schaufeli et al., 2006), and it could be assumed that vigour and innovation provide a starting point for job development through the introduction of robots.

| Methodological strengths and limitations
The questionnaire was based on validated high-reliability measures  (Papadopoulos and Koulouglioti, 2018).

| CON CLUS ION
It is notable that nurse leaders have an influential role in the implementation of new practices and in facilitating the willingness to adapt to new technologies by supporting technology use (Bianchi et al., 2018).
Therefore, it is crucial for nursing leaders to take into consideration employees' age, education and personal qualities, such as their level of self-efficacy (Gifford et al., 2013). Lin (2016) noticed that the higher the general self-efficacy and the self-efficacy about nursing skills learning, the lower the anxiety levels towards nursing skills assessment subjects. Behavioural intention is significantly stronger in the group of graduates than among other employees. This is an important finding concerning home care nursing leadership because in Finland the majority of those who work in home care do not possess a higher education degree. The result is similar with previous studies, and there is a statistically high significant relation between nurses' self-efficacy and their age, occupation and educational level (Zaki, 2016).

| IMPLI C ATI ON S FOR N UR S ING MANAG EMENT
The introduction of robotics technology means changes to many work-related practices, and thus, it is important to involve the entire work community. Our research recommends that it is also important to pay attention to supporting the employees' sense of technology management and constructing a robot-positive atmosphere when introducing care robots. The development of skills of older employees and employees with a lower level of education should also be supported. Change can be facilitated by various pilot projects where employees gain experience in adopting the technology. In contrast, the personnel's general enthusiasm for work, and support for it, is less central in this process of change.

ACK N OWLED G M ENTS
The authors wish to thank the home care employees who participated in the study.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
Author elects to not share data. A care robot can help with medication (e.g. giving medicine, recognising medicine, observing medicine use).

O RCI D
A care robot can contribute to the safety of an older person living at home (e.g. by informing relatives and/or healthcare workers of sudden change in the state of health).
A care robot can remind an older person to take their medicine, of week days, of meetings.
A care robot can guide an older person with the use of phone or bank-related issues.
A care robot can guide an older person in physical exercise.
A care robot can ease the anxiety and loneliness of an elderly person.
A care robot can increase the independence of everyday life in older people.
A robot can increase the meaning of an elderly person's life.
Work engagement (Scale: Never, a few times a year, once a month, a few times a month, once a week, a few times a week, every day).
At my work, I feel bursting with energy.
At my job, I feel strong and vigorous.
I am enthusiastic about my job.
My job inspires me.
When I get up in the morning, I feel like going to work.
I feel happy when I am working intensely.
I am proud on the work that I do.
I am immersed in my work.
I get carried away when I'm working.
I have a lot of experience with using welfare and healthcare technology.

Experience (good) (Likert scale).
I have good experience with using welfare and healthcare technology.

Experience (bad) (Likert scale).
I have experience that welfare and healthcare technology does not work as desired.