275 Organizacija, V olume 54 Issue 4, November 2021 Research Papers 1 Received: 24th May 2021; revised: 23th August 2021; accepted: 23th September 2021 Are we Ready to Use Microchip Implants? An International Cross- sectional Study Anja ŽNIDARŠIČ 1 , Alenka BAGGIA 1 , Antonín PAVLÍČEK 2 , Jakub FISCHER 2 , Maciej ROSTAŃSKI 3 , Borut WERBER 1 1 Faculty of Organizational Sciences, University of Maribor, Kranj, Slovenia, anja.znidarsic@um.si (AZ), alenka.baggia@um.si (AB), borut.werber@um.si (BW) (corresponding author) 2 Faculty of Informatics and Statistics, Prague University of Economics and Business, Prague, Czech Republic, antonin.pavlicek@vse.cz (AP), fischerj@vse.cz (JF) 3 Faculty of Computer Science, Academy of Business in Dabrowa Gornicza, Dąbrowa Górnicza, Poland, mrostanski@gmail.com (MR) Background and purpose: Despite their clear relevance to human life, microchip implants are still widely viewed as negative, threatening our privacy and raising growing concerns about our health. This paper aims to investigate the important factors influencing people’s perception of microchip implants and their willingness to use them for different purposes. Methodology: The cross-sectional study was conducted in three European countries and the data were analysed using the group Structural Equation Modeling approach. Only complete answers to the online survey questionnaire items were used representing a convenience sample of 804 respondents. Results: The results show that perceived ease of use, usefulness and perceived trust are significant predictors of intention to use microchip implants. Perceived trust is influenced by privacy and technology safety. Concerns about painful procedures and other health concerns reduce the perceived usefulness of microchip implants. Apart from the predictor health concerns, the results were similar in all countries. Conclusion: Based on the presented results, researchers interested in investigating the actual use of microchip implants can establish a solid foundation for their research. The results may assist policy makers in developing the regulations to ensure the safe use of microchip implants and allow for a higher level of security. As a follow-up, in- vestigation of changes in the acceptance of microchip implants following the threat of a global pandemic is proposed. Keywords: Microchip implant, Near field communication, Behavioural intentions, Structural equation model, Tech- nology acceptance model DOI: 10.2478/orga-2021-0019 1 Introduction Changes in industry opened the door to a variety of emerging technologies, such as wearable Internet, coop- erating to coordinating machines (Internet of Things), technologies implanted in the human body, and others. These innovative technologies are capable of helping in unpredictable critical situations that occur anywhere in the world, for example digital tracking of patients and identi- fying their contacts. Another successful example are mobile or wearable de- vices based on Radio Frequency Identification (RFID) that use electromagnetic fields to transmit data. Although these devices are revolutionizing healthcare and medicine (Virk- ki et al., 2017), they are still vulnerable to loss and theft. RFID microchip implants (MI) do not have these short- comings. MIs are widely used for healthcare applications such as monitoring (Basham, 2014), enhancement medi- cal devices, and other therapeutic purposes (K. Michael 276 Organizacija, V olume 54 Issue 4, November 2021 Research Papers & Michael, 2013). In addition to healthcare applications, MIs have also been shown to be effective in minimizing errors and personally identifiable medical information (Mohamed, 2020). These novel approaches could make a positive contribution by enhancing the security and safety of people in extreme situations (Sarwar et al., 2019). On the other hand, MI could pose some health risks (e.g., re- jection, allergic reaction) or threats to privacy and security (Rodriguez, 2019). The research on MI adoption found in the literature fo- cuses on either students (Smith, 2008), specific age groups (Achille et al., 2012), small business owners (K. Michael et al., 2017), or even a population with various disabilities (Mohamed, 2020). In this paper, we aim to identify and explore the factors that influence the population’s inten- tions to use MIs. The MI under consideration is a passive NFC RFID device which does not require power supply and can be read only from a short distance. To the best of our knowledge, research on the adoption of MIs from the users’ perspective has not been performed on a wider basis and reported in the literature. Based on a prior pilot study (Werber et al., 2018), an international cross-section- al study was conducted in four European countries. The main contribution of this paper is a model of the factors that influence end-user’s behavioural intention to use MIs, followed by the analysis of the relationships be- tween the constructs of the research model and the analysis of national differences in behavioural intention to use MIs. 2 Literature review The general use of RFID has been researched for al- most two decades. In recent years, we have witnessed a breakthrough in the use of RFID in healthcare and med- icine (Virkki et al., 2017). Whereas active RFID tags re- quire a power source, passive tags draw their energy from the radio wave of an RFID reader, so no power is need- ed (Gaffney & Gopini, 2020). One of the applications of RFID technology is Near Field Communication (NFC) mi- crochips that can be implanted in the human body. These MIs come inside a glass tube and can be read only from a short distance. MIs have also been used in healthcare for prosthetic, monitoring, and enhancement medical devices, to combat diseases such as epilepsy, Parkinson’s disease and severe depression (K. Michael & Michael, 2013), and to impair cancer cells (Lai et al., 2016). The use of MIs helps to min- imize errors in the collection of important medical infor- mation (Mohamed, 2020). In addition to use in health care, there are reports of actual cases where MIs have been used to support intervention in natural disasters (Sarwar et al., 2019). 2.1 Technology acceptance Various methods have been used to research the ac- ceptance of RFID technology and MIs in particular. The Technology Acceptance Model (TAM) is widely used to determine the level of technology acceptance. TAM model anticipates two basic factors that influence the behavioural intention to use technology: perceived usefulness and per- ceived ease of use (Davis, 1989). Researchers have used other approaches to determine the intention for use new technologies. Katz & Rice (2009) defined their own scales to determine the potential for RFID use in healthcare, whereas some authors also identified age as an important predictor of technology use (Gauttier, 2019). 2.2 Microchip implant acceptance Due to their specifics, the willingness to use MIs de- pends not only on perceived usefulness and perceived ease of use, but also on other factors mainly related to health and privacy issues. When implanting a foreign object in the human body, health is always the first consideration. Despite the increasing popularity of MI, the potential health risks have not been adequately researched to ensure the safety of its use. Various problems have been cited in the literature, ranging from the risks of movement in the body, possible effect on emotional behaviour, allergies, ef- fect on the nervous system, and pain during the insertion of the MI (Fram et al., 2020). According to Albrecht (2010), MIs could potentially lead to malignancies, whereas Lai et al. (2016) have found the possibility of treating cancer cells with MIs. However, most of this research is based on microchips for animals, while there is limited evidence on the safety of MI in humans (Fram et al., 2020). The first specific study of RFID adoption in healthcare from an end-user perspective found that physical place- ment (without actual insertion into the body) did not ap- pear to raise public concern, with the exception of a small minority (Katz & Rice, 2009). Public interest in RFID was strongest for emergency intervention services. Research on the acceptance of MIs showed that they were treated positively (Smith, 2008), and the acceptance of MIs for life-saving purposes was highest (Rotter et al., 2008). In addition, studies have found that the willingness to adopt MIs is slowly increasing (Perakslis et al., 2014), al- though the perception of MIs as secure technology varies according by country of residence and generational fac- tors (Perakslis & Michael, 2012). Carr (2020) believes that MIs can be a solution to reduce contacts and risks af- ter pandemic outbreaks. MIs have been used for various non-therapeutic purposes, initially for personal interest only (K. Michael & Michael, 2013), but more recently in the workplace, for example to access a secured worksta- tion (Fram et al., 2020). According to K. Michael et al. 277 Organizacija, V olume 54 Issue 4, November 2021 Research Papers (2017), there are numerous reasons for rejecting MIs for employee identification, where data protection and securi- ty reasons certainly cannot be ignored (Rodriguez, 2019). Chipping employees in the workplace raises even more ethical issues and challenges (Gauttier, 2019). The situ- ation during pandemics has shown, how quickly privacy rights can disappear when confronted with health and safe- ty concerns and therefore it is crucial to draft the employee microchipping legislation (Turner, 2020). 2.3 Structural equation modelling Structural equation modelling approach is mainly used to test the hypotheses in the technology acceptance mod- el (Beaujean, 2014). The minimum sample size for per- forming SEM has been discussed several times. Proposals range from 150 to 400 when there are three or more meas- ured items per latent variable (Hair et al., 2019) or 250 to avoid rejection of the model due to the combination of rules for fit indices (Hu & Bentler, 1999). Schumacker & Lomax (2010) propose the analysis performed according to the standard two-stage approach at SEM, the first step being the validation of the meas- urement model. The Confirmatory Factor Analysis (CFA) is performed to determine how well the measured items reflect the theoretical latent variables and to examine the construct validity of the measurement model, which is examined through convergent validity and discriminant validity. When examining convergent validity, one needs to examine that the estimates of standardized factor load- ings do not exceed 0.5 (or even 0.7), Composite Reliabil- ity (CR) for each latent variable exceeds 0.7, and Average Variance Extracted (A VE) for each latent variable exceeds 0.5 (Koufteros, 1999). In a second step, SEM is used to test the structural re- lationships between the latent variables. The unstandard- ized B and standardized path coefficients β (relationships between the latent variables), z-values (ratio of β to its standard error), and the significance level are calculated. For each endogenous latent variable, a coefficient of deter- mination (R^2) is calculated, representing the percentage of the explained variance of the variable by the set of its predictors. The overall fit of the measurement and structural mod- el are assessed based on a set of fit indices: • The value of the comparative fit index (CFI) should be at least 0.9 to indicate adequate model fit (Koufteros, 1999). • The root mean square error of approximation (RMSEA) value should be below 0.06 (Teo & Zhou, 2014), or between 0.06 and 0.08 to be in- terpreted as mediocre (MacCallum et al., 1996). • The standardised root mean square residual (SRMR) should be less than 0.05, however val- ues as high as 0.08 are deemed acceptable (Hu & Bentler, 1999). • Some goodness-of-fit (GFI) indices are affected by the complexity of the model (e.g., CFI, but not RMSEA) (Cheung & Rensvold, 2002). Therefore, generally accepted criterion (e.g., CFI = 0.90) in complex models should be judged with caution. Multigroup Confirmatory Factor Analysis (MG-CFA) and multigroup Structural Equation Modelling (MG-SEM) is then used to complement the general two-step procedure when we have multiple groups. Using MG-CFA and MG- SEM we can assess the measurement invariance (MInv), concerning the comparison of the same measurement model in different groups, and compare the effects or con - structs’ means across groups, which concerns the analysis of the moderating role of a categorical variable that forms groups in a specified SEM (Miceli & Brabaranelli, 2016). Before making meaningful comparisons of survey results across groups, researchers should ensure that re- spondents from different groups have ascribed similar meaning to survey items (Cheung & Lau, 2011). MInv as- sesses the psychometric equivalence of a construct across groups or over time (Putnick & Bornstein, 2016), while measurement noninvariance suggests that a construct has a different structure and/or meaning to different groups. MInv is usually tested using configural invariance, weak invariance, and strong invariance, sometimes these are fol- lowed by strict invariance (Beaujean, 2014). The configural invariance tests whether the model configuration (all constructs have the same pattern of free and fixed parameters) is the same among all groups in a multigroup context (Putnick & Bornstein, 2016). For weak invariance, the item loadings must be the same across groups, for strong invariance the intercepts of indicators must be the same across groups, while for strict invariance also error variances must be constrained to be equal across groups. The results for each invariance test are explained by the change of several alternative fit indices (AFI) since χ^2 tends to be oversensitive to small, unimportant devia- tions from a perfect model in large samples (Chen, 2007). Change in CFI (ΔCFI), SRMR (ΔSRMR), and RMSEA (ΔRMSEA) were used to assess model fit. Cheung & Rensvold (2002) proposed the use of a criterion of -0.01 change in CFI of two nested models. Whereas Chen (2007) suggested that a criterion of a -0.01 for ΔCFI is paired with ΔRMSEA of 0.015 and SRMR of 0.030 (for metric invari- ance) or 0.015 (for scalar or residual invariance). The rules of thumb for AFI and ΔAFI might not gener- alize to the wide range of SEMs encountered in practice, models with only negligible mis-specifications should not be rejected, and researchers should not rely on a single rule-of-thumb cut-off for any (Δ)AFI (Jorgensen et al., 2018). Traditionally, configural invariance is assessed by evaluating the overall fit of the configural model, whereas (Jorgensen et al., 2018) proposed a permutation test, espe- 278 Organizacija, V olume 54 Issue 4, November 2021 Research Papers cially when evaluating configural invariance with small to moderate sample sizes. As pointed out by Jorgensen et al. (2018) configural models frequently do not fit the data per - fectly, but the use of the permutation test of configural in- variance can prevent inflated type I errors when the model fits only approximately well. The idea of the permutation test is that the variable of group membership is randomly shuffled (several times) and the model is fitted to that data. In the permutation test, the proportion of the statistics (e.g. χ^2, CFI, RMSEA) that are more extreme than the ob- served statistics (of the original model), is calculated. This is a one-tailed p-value that approximates the probability of obtaining statistics under investigation (e.g. χ^2), as poor as the observed one, if the invariance across all groups holds true. If p<α , H0 has to be rejected. The permutation test could be applied for both badness of fit measures (e.g. χ^2, RMSEA,...) or goodness of fit indices (e.g. CFI) (see e.g. Jorgensen et al. (2018) for further information). 3 Methods On the basis of previous studies (e.g. M. G. Michael & Michael, 2010; Perakslis et al., 2014), we constructed a model of factors influencing the behavioural intentions to use MIs based on TAM, as shown in Figure 1. The ex- tended model based on TAM adopts the components of TAM: Perceived Ease of Use (PEU), Perceived Useful- ness (PU), and Behavioural Intention to Use (BIU). We added constructs of personal factors to the basic model of TAM: Perceived Trust (PT), Privacy Right (PR), Privacy Threat (PTh) and Health Concerns (HC). Three predictor variables have been added to the model: Age, Technology Safety, and Painful Procedure. The items of basic TAM constructs were defined ac- cording to previous research: • Perceived Ease of Use (PEU): Items availabili- ty, and multifunctionality were adopted from the original TAM model (Davis, 1989), whereas the items on the option to be lost or stolen were in- cluded on the basis of the pilot study (Werber et al., 2018). • Perceived Usefulness (PU): In addition to five items adopted from (Katz & Rice, 2009), the items on organ donation information and saving lives under different conditions were added based on the pilot study (Werber et al., 2018). • Behavioural Intention to Use (BIU): Based on the pilot study, items considered intention to use for healthcare purposes, for identification purposes, for shopping and payment, and for everyday use at home. The intention to use in case GPS posi- tioning and tracking was not possible was added. Age was included as a predictor variable to BIU construct based on previous research on the influ- ence of age on the adoption of new technologies (Gauttier, 2019). • Additional constructs of personal factors were constructed on the basis of earlier research as fol- lows: • Health Concerns (HC): HC concerns refer to pos- sible health risks of MI derived from previous re- search (e.g. Albrecht, 2010). The variable Painful Procedure (PP) was adopted based on the basis of claims that pain or damage is associated with the insertion of MI (M. G. Michael & Michael, 2010). • Perceived Trust (PT): Items concerning the per- ceived trust that the state and other institutions will ensure security have been derived from (Smith, 2008). The variable Technology Safety (TS) associated with PT has been adopted as a predictor variable in PT construct from (Perakslis et al., 2014). • Privacy Right (PR): Items concerning the level of privacy were adopted from Lockton & Rosenberg (2005). • Privacy Threat (PTh): The construct was includ- ed in the model on the basis of previous research (Bansal et al., 2015), assuming that it has a neg- ative impact on the perceived right to privacy. Items of HC, PP, PT, PR, PTh, TS and PEU, were measured on a 5-point scale of agreement (“strongly disa- gree” to “strongly agree”), whereas the items of PU were measured on a 5-point scale of acceptability (“very bad idea” to “very good idea”). Five items on BIU were meas- ured with yes/no options. To study the relationships among our constructs, nine hypotheses presented in Table 1 were postulated. In addi- tion, positive or negative impacts between constructs are depicted. The relationships among constructs, together with the type of the relationship (positive or negative), are present- ed in Figure 1 as a proposed MI acceptance model. To test the hypotheses of the proposed structural mod- el, we updated the pre-developed questionnaire (Werber et al., 2018). Two scales on Privacy Right and Privacy Threat (Katz & Rice, 2009) and an item on Technology safety (Perakslis et al., 2014) were added. 3.1 Data collection The convenience sampling approach was utilized in MIs acceptance research. Respondents were invited to participate in an online survey through various channels, ranging from the researcher’s social networks to media posts. The survey was conducted in 2016 and 2017 in four countries: Slovenia, the Czech Republic, Poland, and Croatia. The introduction to the survey, included the de- scription that the research was about passive NFC RFID microchip implants that do not require power supply and 279 Organizacija, V olume 54 Issue 4, November 2021 Research Papers Table 1: Research hypotheses and their relationships Research hypotheses Relationship Type H1: Privacy threat will have a significant positive effect on privacy right PTh  PR Positive H2: Privacy right will have a significant negative effect on perceived trust PR  PT Negative H3: Technology safety will have a significant positive effect on perceived trust TS  PT Positive H4: Painful procedure will have a significant positive effect on health concerns PP  HC Positive H5: Health concerns will have a significant negative effect on perceived usefulness HC  PU Negative H6a: Perceived ease of use will have a significant positive effect on perceived usefulness PEU  PU Positive H6b: Perceived ease of use will have a significant positive effect on behavioural intention to use PEU  BIU Positive H7a: Perceived Trust will have a significant positive effect on perceived usefulness PT  PU Positive H7b: Perceived Trust will have a significant positive effect on behavioural intention to use PT  BIU Positive H8: Perceived usefulness will have a significant positive effect on behavioural intention to use PU  BIU Positive H9: Age will have a significant negative effect on behavioural intention to use Age  BIU Negative Figure 1: The proposed acceptance model for behavioural intentions to use microchip implant 280 Organizacija, V olume 54 Issue 4, November 2021 Research Papers can be read only from a short distance. Possible uses of MIs were also listed. The questionnaire was distributed to a convenience sample of the general population. Only complete answers to the questionnaire items were used, namely 250 (31.1%) from Slovenia, 339 (42.2%) from the Czech Republic, and 215 (26.7%) from Poland. The Croatian subsample (146) was excluded from the analyses due to the minimal sample size requirements for structural equation modeling (SEM). 3.2 Statistical methods The measurement model shown in Figure 1 describes the relationships between the observed measured items and the unobserved latent variables. The data obtained from the survey were analysed using the SEM approach. The analysis was performed according to the stand- ard two-stage approach at SEM (Schumacker & Lomax, 2010), the first step being the validation of the measure- ment model, whereas in the second step, structural rela- tionships between latent variables were tested. The overall fit of the measurement and structural model were assessed based on a set of fit indices. MG-CFA and MG-SEM were used to complement the general two-step procedure due to the multiple groups representing samples from three countries. All analyses, CFA, MInv, and SEM, were performed with the R-package lavaan (Rosseel, 2014) and semTools (Jorgensen et al., 2020). In the following section, the re- sults are presented according to the described analysis pro- cedure. 4 Results In this section the results of 804 responses with com- plete answers to the items included in the research model (88.16% of 912 responses) are presented. The sample con- sists of 51.56% women and 48.44% men, detailed distribu- tion according to countries is presented in Table 2. The majority of respondents are employed (67.75%) or students (20.88%). The status of respondents according to countries is presented in Table 3. The mean age of the respondents is 37.5 years with standard deviation 13.91 years. The age distribution ac- cording to countries is presented in Table 4. Table 2: Gender distribution according to country Table 3: Status of respondents according to country Table 4: Descriptive statistics (means (M) and standard deviations (SD)) for age of respondents according to country Men Women Slovenia 40.96% 59.04% the Czech Republic 47.49% 52.51% Poland 58.69% 41.31% Student Employed Unemployed Retired Slovenia 33.73% 53.82% 5.62% 6.83% the Czech Republic 20.48% 68.25% 3.26% 8.01% Poland 6.54% 83.18% 5.14% 5.14% M SD Slovenia 35.7 11.8 the Czech Republic 38.0 14.2 Poland 38.7 11.8 281 Organizacija, V olume 54 Issue 4, November 2021 Research Papers 4.1 Descriptive statistics of the questionnaire items The percentages of respondents who are willing to in- sert a MI for different usages are shown in Table 5. The highest percentage of the respondents (47.76%) would insert an MI for health care purposes and the lowest per- centage of the respondents (20.65%) would have MI for shopping and payment. The variable “number of MI uses” was calculated as the sum of five dichotomous variables on different MI uses and its mean value was 1.53 and SD = 1.85 (Table 6). Descriptive statistics for 25 continuous variables in- cluded in the research model are presented in Table 6. On average, the highest agreement among respondents was in the case of PR construct for the variables discussing the right to control your information (M = 4.50,SD = 0.73), and that no one should be able to collect or disclose your personal information without your consent (M = 4.46,SD = 0.83). On average, the largest disagreement is obtained for the statement about the security and protection of hu- man rights (M = 2.53,SD = 1.23). 4.2 Measurement model evaluation Construct validity aims to determine how well a set of measured items reflects the theoretical latent variable they are designed to measure. Construct validity is examined with the evaluation of convergent validity and discrimi- nant validity. Convergent validity First, the initial overall measurement model (M1), disregarding the countries was evaluated. The results of measurement model development and the fit indices are presented in Table 7. The item PEU3 (λ= 0.441) was re- moved from the model, due to its standard loading below 0.5. Standardized factor loadings for items in the (final) overall model (M2) exceed a threshold of 0.5 for conver- gent validity. In addition, 86% (18) of them exceed even a stricter threshold of 0.7. The A VE for six constructs exceeds a threshold of 0.5 for convergent validity (Table 8). The A VE for PEU is slightly below 0.5 (A VE = 0.47), but its CR is higher than 0.6 (CR = 0.73) the convergent validity of the construct is still adequate (Fornell & Larcker, 1981). The obtained re- sults prove the convergent validity for the set of latent var- iables and corresponding items in the measurement model, therefore all included items are significantly related to the specified latent variable. Discriminant validity The discriminant validity of the M2 (overall) measure- ment model was examined through the comparison of the square root of A VE of each latent variable to the correla- tions between the latent variables (Table 8). The correla- tions among the constructs (and three measured variables included in the structural model) are given in the right panel of Table 8, where the diagonal elements correspond to the values of the square root of A VE. The values of the square root of A VE for each construct surpass the corre- sponding correlations between constructs. Therefore, dis- criminant validity can be inferred for all latent variables. Internal consistency of the questionnaire was assessed to determine the extent to which the measured items with- in the construct were related to each other. Cronbach’s al- pha coefficients (Table 8) for six constructs ranged from 0.79 to 0.92 indicating high internal reliability (Hair et al., Table 5: Willingness to insert microchip implant (N=804) Would you insert MI: PPR a ...for health care purposes (identification, storage of medical data, information on organ donations, etc.)? 47.76% ...for identification purposes (ID card, passport, driving license, etc.)? 31.22% ...for shopping and payment purposes (debit cards, credit cards, profit cards, etc.)? 20.65% ...for everyday use at home (unlocking of house or apartment, car, computer, mobile phone, etc.)? 22.76% ...if you had been assured that GPS tracking and positioning would not be possible? 30.60% a PPR – Percentages of positive responses 282 Organizacija, V olume 54 Issue 4, November 2021 Research Papers Construct Questionnaire item M SD Privacy Right (PR) No one should be able to gather or disclose your personal information without your consent. (PR1) 4.46 0.83 People should have the right to control their personal information. (PR2) 4.50 0.73 Privacy Threat (PTh) Organizations and agencies ask you for too much personal information. (PTh1) 3.93 0.94 The present use of computers is an actual threat to personal privacy in the country. (PTh2) 3.69 1.02 I am concerned about threats to my privacy in the country today. (PTh3) 3.49 1.11 Technology safety (TS) MIs technology is safe enough to be used in humans. (TS1) 2.66 1.01 Health Concerns (HC) MIs can be threatening to my health because of the possibility of movement in my body. (HC1) 2.94 1.08 MIs may affect my emotional behaviour (control of human behaviour, etc.). (HC2) 2.70 1.18 MIs can be threatening to my health because of possible allergies. (HC3) 3.29 1.09 MIs can be threatening to my health because of their impact on the nervous system. (HC4) 3.06 1.11 Painful Procdure (PP) Implanting MI is a painful procedure. 2.84 1.02 Perceived Trust (PT) The state will ensure the security and the protection of human rights (security of identity documents, passport, identity theft, tracking via GPS, no records should be archived without the consent of the person observed). (PT1) 2.53 1.23 Banks will provide security (payment, discretion of operation, transactions, etc.). (PT2) 2.68 1.22 The healthcare system will provide security (personal data, medical data, information on treat- ments, organ donation, etc.). (PT3) 3.05 1.26 Perceived Usefulness (PU) MIs could be used for - monitoring the health of the user. (PU1) 3.71 1.04 - warning about potential health problems or complications. (PU2) 3.80 1.02 - storing a user’s medical info to be used in an emergency. (PU3) 3.80 1.02 - personalized health info. (PU4) 3.39 1.15 - storing information about organ donation. (PU5) 3.43 1.14 - saving life (e.g. unconsciousness, cardiac pacemaker, insulin dispenser). (PU6) 3.85 1.00 Perceived Ease of Use (PEU) MIs are always available. (PEU1) 3.55 0.99 MIs cannot be lost. (PEU2) 3.78 0.96 MIs cannot be stolen (high-security protection). (PEU3) 3.08 1.18 MIs can integrate multiple functions at the same time. (PEU4) 4.01 0.80 Behavioural Intention to Use (BIU) Number of different subcutaneous microchip uses. 1.53 1.85 Table 6: Descriptive statistics with means (M) and standard deviations (SD) of items for model constructs (N = 804) 283 Organizacija, V olume 54 Issue 4, November 2021 Research Papers Table 7: Results of measurement model development and model fit indices Model χ 2 df CFI SRMR RMSEA 90% CI M1 – initial overall model 1010.560 194 0.916 0.052 0.072 M2 – overall model with removed PEU3 863.804 174 0.927 0.049 0.070 Final model (M2) for each country MSI –Slovenia 339.061 174 0.950 0.052 0.062 MCZ - Czech Republic 589.404 174 0.877 0.067 0.084 MPO - Poland 359.205 174 0.940 0.051 0.070 Table 8: Cronbach’s Alpha, Composite reliability (CR), average variance extracted (AVE), square root of AVE on the diagonal (marked in grey) and correlations among constructs Correlations Cronbach’s CR AVE PR PTh TS PP HC PT PU PEU Age BIU Alpha PR 0.858 0.862 0.758 0.871 PTh 0.793 0.797 0.568 0.464 0.754 TS a / / / -0.049 -0.206 / PP a / / / -0.014 0.032 -0.154 / HC 0.838 0.855 0.597 0.098 0.278 -0.503 0.466 0.773 PT 0.877 0.877 0.706 -0.111 -0.274 0.384 -0.004 -0.310 0.840 PU 0.921 0.921 0.660 0.030 -0.152 0.394 -0.090 -0.391 0.577 0.813 PEU 0.614 0.727 0.474 0.175 -0.056 0.276 -0.155 -0.315 0.406 0.509 0.689 Age a / / / 0.002 -0.001 -0.046 0.038 0.131 -0.142 -0.174 -0.026 / BIU a / / / -0.086 -0.238 0.368 -0.074 -0.391 0.541 0.471 0.334 -0.140 / a Measured variables TS, PP , Age, and BIU are included into the table only to compare square root of AVE of a construct with correlations to other constructs and measured variables. Cronbach’s Alpha, CR, and AVE are not applicable for measured variables. 2019). The coefficient for PEU is slightly lower but still acceptable (Cronbach’s alpha = 0.61). The overall model fit The overall fit of the final measurement model (M2) was assessed based on a set of commonly used fit indi- ces (Table 7). The χ^2 was 863.81 with 174 degrees of freedom. Both, CFI and SRMR indicate a good model fit (CFI = 0.927 SRMR = 0.049). RMSEA is equal to 0.070 and the upper bound of RMSEA 90% confidence interval (0.066,0.075) is lower than 0.08 suggesting a good model fit (MacCallum et al., 1996). According to the set of the calculated fit indices, we conclude that the measurement model fits the sample data well. We tested whether our final model fits each country’s subsample. The model fits all of the subgroups well, with SRMR values from 0.051 to 0.67, RMSEAs of 0.062 to 0.084, CFIs of 0.88 to 0.95 (see Table 7). When examined separately, the M2 fits each subgroup well. Therefore, we can proceed with testing measurement invariance across groups. 4.3 Multiple-group analyses: testing measurement invariance across countries The measurement invariance tests are performed us- ing the hierarchical ordering of nested models (Putnick & Bornstein, 2016) starting with the evaluation of the config- ural invariance and following by weak, strong, and strict 284 Organizacija, V olume 54 Issue 4, November 2021 Research Papers invariance. The results of the model fits are shown in Table 9. Test of configural invariance A permutation test, based on 1000 repetitions, revealed no evidence against the null hypothesis of configural invar- iance using either χ^2 (p = 0.557), CFI (p = 0.721), SRMR (p=0.375) or RMSEA (p = 0.557) and its 90% confidence interval (p = 0.557 for the lower bound and upper bound as criterion). This indicates that configural invariance is supported. Test of weak invariance To test for weak invariance, the factor loadings were constrained to be equal across groups. Because the weak invariance model (M4) is nested within the baseline con- figural model (M3), a χ^2 difference test was performed. Since χ^2-test of two nested models is oversensitive to small, unimportant deviation from a perfect model in a large sample (Chen, 2007; Cheung & Rensvold, 2002), we report the differences in alternative fit indices (ΔAFI). ΔCFI (-0.002), ΔREMSA (-0.001 , and ΔSRMR (0.002) between the configural and weak models (Table 9). Test of strong invariance To test for strong invariance, in addition to factor load- ings also intercepts were constrained to be equal across groups. The ΔAFIs of strong invariance model (M5) ac- cording to M4 were as follows: ΔCFI (0.020), EMSA (0.006), and ΔSRMR (0.006). Since ΔCFI is above the prescribed level, there is some evidence that the intercepts are not completely invariant across the three considered groups. When intercept of measured item PT3 was freely estimated accros groups, the partial strong variance (model M5a) was established. However, there was a significant CFI difference (ΔCFI = -0.026) between the partial strong model and the strict model (M6). These results suggest a greater lack of fit when constraining also error variances to be invariant across groups. Since partial strong measurement invari- ance was supported, we can proceed to the evaluation of the structural model (Putnick & Bornstein, 2016). Table 9: Testing measurement invariance across countries Model (Model comparison) χ 2 (Δχ 2 ) df CFI (ΔCFI) SRMR (ΔSRMR) RMSEA (ΔRMSEA) RMSEA 90% CI M3 - configural invariance 1287.67 522 0.922 0.056 0.074 0.069; 0.079 M4 - weak invariance (M3) 1330.32 (42.7) 552 (30) 0.920 (-0.002) 0.058 (0.002) 0.073 (-0.001) 0.068; 0.078 M5 – strong invariance (M4) 1563.50 (233.2) 582 (30) 0.900 (-0.020) 0.064 (0.006) 0.079 (0.006) 0.075; 0.084 M5a – strong partial invari- ance (M3) 1446.28 (116.0) 580 (28) 0.911 (-0.009) 0.062 (0.004) 0.075 (0.002) 0.070; 0.079 M6 - strict invariance (M5a) 1742.68 (296.4) 622 (42) 0.885 (-0.026) 0.064 (0.002) 0.082 (0.007) 0.077; 0.087 4.4 Testing structural model After assessing the model fit of the overall measure- ment model and its partial strong invariance, a structural model was evaluated. According to the research model (Figure 1), four measured variables and 11 structural paths were added to the six constructs described in the previous section and in accordance with the proposed hypotheses. First, the overall model was evaluated. Examining the overall structural model The overall structural model (SM1) fit was good. The following criteria were determined: χ^2=1548.63, df=264, CFI=0.88, and RMSEA=0.078 (90% CI; 0.074,0.082). Since the main aim of this research is group comparison, 285 Organizacija, V olume 54 Issue 4, November 2021 Research Papers the detailed results of the overall model are not presented here. Although the research hypotheses were supported in the overall model, it is not clear whether these hypotheses hold across different countries. For example, would the impact of health concerns on perceived usefulness remain significant for all three countries? To determine whether the structural relationships are invariant, it is essential to establish a structural model invariance. Measurement invariance of the structural model The fit of the partial strong invariance model (SM1) (Table 10) was good: χ^2=2426.11, df=822, CFI=0.86, and RMSEA=0.083 (90% CI; 0.079,0.087). The fit of the structural model (SM2), where also structural coefficients were constrained to be equal across groups, was as follows: χ^2=2480.64, df=872, CFI=0.85, and RMSEA=0.083 (90% CI; 0.079,0.087). The χ^2-test (p=0.0001) of the two nested models suggests that models SM1 and SM2 are sig- nificantly different, meaning that some paths vary across groups. To determine whether the structural paths are invar- iant across three groups, the individual structural coeffi- cients were successively restricted to be equal across three groups and nested models were compared. More precisely, model SM1 and a model in which a particular path coeffi- cient of interest was specified as invariant were compared at a time (Table 10). Since we are comparing three groups, such an approach ensures that the χ^2 difference test has 2 degrees of freedom, whereby any observed χ^2 differenc- es greater than 5.99 being statistically significant at a 5% significance level. Results of model comparisons are presented in Table 10. The following paths were found to be different across groups at 5% significance level: • PTh → PR (SM1a), • PP → HC (SM1d), • HC → PU (SM1e), • PT → PU (SM1h). Those coefficients were freely estimated across groups in the final model (SM3). 4.5 Results of the final structural model The fit of the final model (Table 10) was good: χ^2=2441.77, df=864, CFI=0.85, and RMSEA=0.084 (90% CI; 0.079,0.086). Table 11 shows the results for the unstandardized coefficients (B), standardized coefficients (β) and corresponding z-values, which reflect the relation- ships among the latent variables in terms of magnitude and statistical significance. Due to relatively large number of tests, the adjusted p-values using false discovery rate method (Benjamini & Hochberg, 1995) were calculated. A graphical overview of the (un)confirmed hypotheses is shown in Figure 2. For each endogenous construct the co- efficient of determination (R^2) was also calculated (Table 12). The theory of TAM suggests that there are positive ef- fects of PEU and PU on BIU (in this case H6b and H8). The results confirmed that both hypotheses are confirmed in all three countries at a 5% significance level and that the structural coefficients do not differ statistically significant - ly across countries. We found that all constructs togeth- er explain 30.6% of the total variance of BIU in Poland, 33.5% in Slovenia, and 26.0% in the Czech Republic. Another relationship that is usually predicted in TAM applications is the positive impact of PEU on PU. In our model, this relationship is described by hypothesis H6a, which is confirmed at a 5% significance level in all three countries (and the magnitude does not differ statistically significantly across countries). Four constructs and three external variables were add- ed to the original TAM model, as shown in Figure 1. Hy- pothesis H1, which indicates a positive impact of PTh on PR was confirmed in all three countries, although the mag- nitude of the impact is different. The effect of PTh on PR was significantly stronger in Slovenia (β=0.568) and the weakest in the Czech Republic (β=0.394). The same magnitude of negative impact of PR to PT (H2) was confirmed in all three countries. The negative impact of age on the BIU (H9) was not confirmed in any country. The constructs with a positive impact on the basic components of TAM are HC and PT, where HC impact PU (H5), whereas PT impacts PU and BIU (H7a, H7b). Two hypotheses (H7a, H7b) were confirmed in all three coun- tries, but the magnitudes of the effects differ. Hypothesis H5 was not confirmed in the Czech Republic. The variable PU has two significant predictors (HC and PT) that can explain between 31.4% and 47.0% of its total variance in Poland and the Czech Republic, respectively. The positive impact of the PP external variable on HC was proposed with hypothesis H4 and confirmed in all three countries. The magnitude of the effect was highest for the Poles (β=0.596) and lowest for the Czechs (β=0.347). Similarly, the same positive impact of the external variable TS on PT (H3) was confirmed in all three countries. The two predictors of PT (PR and TS) explain at least 15.3% of the total variance of H3. 286 Organizacija, V olume 54 Issue 4, November 2021 Research Papers Table 10: Testing measurement invariance of structural coefficients across countries Structural model (SM) (Model comparison) χ 2 (Δχ 2 ) df p CFI (ΔCFI) SRMR (ΔSRMR) RMSEA (ΔRMSEA) RMSEA 90% CI SM1 – partial strong invariance 2426.11 850 / 0.854 0.127 0.083 0.079; 0.087 SM2 – structural coef- ficients (SM2) 2480.64 (54.33) 872 (22) 0.0001 0.851 (-0.003) 0.131 (0.004) 0.083 (0.000) 0.079; 0.087 Constrained individual paths to be equal across groups: SM1a: PTh -> PR (SM1) 2432.87 (6.76) 852 (2) 0.0340 0.854 (0.000) 0.127 (0.000) 0.083 (0.000) 0.079; 0.087 SM1b: PR -> PT (SM1) 2427.69 (1.56) 852 (2) 0.4589 0.853 (-0.001) 0.129 (0.002) 0.083 (0.000) 0.079; 0.087 SM1c: TS -> PT (SM1) 2426.60 (0.49) 852 (2) 0.7815 0.854 (0.000) 0.127 (0.000) 0.083 (0.000) 0.079; 0.087 SM1d: PP -> HC (SM1) 2435.13 (9.02) 852 (2) 0.0110 0.853 (-0.001) 0.128 (0.001) 0.083 (0.000) 0.079; 0.087 SM1e: HC -> PU (SM1) 2439.56 (13.45) 852 (2) 0.0012 0.853 (-0.001) 0.128 (0.001) 0.083 (0.000) 0.080; 0.087 SM1f: PEU -> PU (SM1) 2426.13 (0.02) 852 (2) 0.9890 0.854 (0.000) 0.127 (0.000) 0.084 (0.001) 0.079; 0.087 SM1g: PEU -> BIU (SM1) 2426.23 (13.45) 852 (2) 0.9413 0.854 (0.000) 0.127 (0.000) 0.083 (0.000) 0.079; 0.087 SM1h: PT -> PU (SM1) 2439.66 (13.55) 852 (2) 0.0011 0.853 (-0.001) 0.129 (0.002) 0.083 (0.000) 0.080; 0.087 SM1i: PT -> BIU (SM1) 2431.96 (5.85) 852 (2) 0.0536 0.854 (0.000) 0.129 (0.002) 0.083 (0.000) 0.079; 0.087 SM1j: PU -> BIU (SM1) 2427.39 (1.28) 852 (2) 0.5276 0.854 (0.000) 0.129 (0.002) 0.083 (0.000) 0.079; 0.087 SM1k: Age -> BIU (SM1) 2431.09 (4.98) 852 (2) 0.0830 0.854 (0.000) 0.127 (0.000) 0.083 (0.000) 0.079; 0.087 SM3 – final model (SM1) 2441.77 (15.66) 864 (14) 0.3349 0.854 (0.000) 0.128 (0.001) 0.084 (0.000) 0.079; 0.086 287 Organizacija, V olume 54 Issue 4, November 2021 Research Papers Table 11: Summary of hypotheses testing for the structural model across countries Hypothesis & Path Expected Sign (Constrained across groups) Country B β z p Adjusted p Confirmed? H1 PTh→PR + (No) SI 0.769 0.568 7.631*** 0.000 0.000 Yes CZ 0.441 0.394 5.670*** 0.000 0.000 Yes PO 0.583 0.463 6.257*** 0.000 0.000 Yes H2 PR→PT - (Yes) SI -0.139 -0.086 -2.719** 0.007 0.008 Yes CZ -0.103 Yes PO -0.113 Yes H3 TS→PT + (Yes) SI 0.390 0.387 11.183*** 0.000 0.000 Yes CZ 0.422 Yes PO 0.375 Yes H4 PP→HC + (No) SI 0.377 0.489 8.030*** 0.000 0.000 Yes CZ 0.260 0.347 6.048*** 0.000 0.000 Yes PO 0.505 0.596 9.480*** 0.000 0.000 Yes H5 HC→PU - (No) SI -0.226 -0.239 -4.053*** 0.000 0.000 Yes CZ -0.038 -0.036 -0.713 0.476 0.476 No PO -0.295 -0.264 -4.052*** 0.000 0.000 Yes H6a PEU→PU + (Yes) SI 0.480 0.383 8.744*** 0.000 0.000 Yes CZ 0.332 Yes PO 0.374 Yes H6b PEU→BIU + (Yes) SI 0.253 0.094 2.208* 0.027 0.030 Yes CZ 0.074 Yes PO 0.098 Yes H7a PT→PU + (Yes) SI 0.308 0.396 6.723*** 0.000 0.000 Yes CZ 0.509 0.599 11.020*** 0.000 0.000 Yes PO 0.264 0.308 4.803*** 0.000 0.000 Yes H7b PT→BIU + (Yes) SI 0.755 0.455 10.908*** 0.000 0.000 Yes CZ 0.378 Yes PO 0.436 Yes H8 PU→BIU + (Yes) SI 0.385 0.181 4.351*** 0.000 0.000 Yes CZ 0.163 Yes PO 0.191 Yes H9 Age→BIU - (Yes) SI -0.007 -0.060 -1.845 0.065 0.069 No CZ -0.057 No PO 0.046 No 288 Organizacija, V olume 54 Issue 4, November 2021 Research Papers Table 12: Coefficients of determination (R 2 ) Construct SI CZ PO PR 0.323 0.155 0.214 HC 0.239 0.120 0.355 PT 0.157 0.189 0.153 PU 0.368 0.470 0.314 BI 0.335 0.260 0.306 Figure 2: The final acceptance model for behavioural intention to use microchip implant 5 Discussion and conclusions With today’s access to information, people are becom- ing more and more aware of new technologies and their widespread applications. When we think about the use of new technologies, we are confronted with a variety of at- titudes, from technophile to traditionalist to conservative. Despite many concerns about privacy (Rodriguez, 2019) and possible reactions to a foreign body in the human body (Albrecht, 2010), the applications of MI in healthcare are very successful (e.g. pacemakers, drug administration, prostheses). Although some people have already decided to use MI for non-therapeutic purposes (Fram et al., 2020), research on the adoption of MI by individual users, fac- tors that influence attitudes towards MI should be investi- gated. Although the general attitude towards implants has changed in recent years, some health, religious or personal concerns still limit the general use of MI. The aim of this study was to identify the drivers and barriers to the adoption of MI in the general population. It complements previous research on the adoption of MI by providing insight into the attitudes of an individual, the actual user of MI. The cross-sectional study, which was 289 Organizacija, V olume 54 Issue 4, November 2021 Research Papers conducted in three European countries, showed similar at- titudes towards the introduction of MIs. In general, people would be most willing to use MI for healthcare purposes (47.8%), whereas, similar to Franks & Smith (2021), willingness for other purposes was signif- icantly lower. For example, 31.2% of respondents would be willing to use MI for identification, 20.6% for shopping and payment, and 22.8% for everyday use at home. Perceived ease of use, usefulness, and trust were iden- tified as the most important predictors of intention to use MIs. In addition to their influence on intention to use, per- ceived trust and ease of use, influence perceived useful- ness. In contrast to these predictors, health concerns act as a negative predictor of perceived usefulness in two out of three countries. According to these results, the public is aware of the variety of useful applications of MI but has no confidence in research on the safety of the technology. We anticipated that due to the variety of body modifica- tions, such as piercing, tattoos and even plastic surgery, which have become popular in recent years, the problem of pain when inserting foreign bodies will no longer be an obstacle. Nevertheless, according to the results of this study, health concerns are still anticipated by the fact that the insertion of a MI is a painful procedure. Privacy right perception can be predicted by the indi- vidual’s attitude towards threats to privacy. Furthermore, concerns about privacy rights act as a negative predictor of perceived trust. Given that people have become ac- customed to wearable devices or smartphones and have agreed to be tracked by them in order to take advantage of their benefits, we can therefore say that this is a surprising result. In line with the lifestyle changes mentioned above, our expectation that the perceived trust is anticipated by the perception of the security of the technology of MI, was confirmed. It is interesting to note that age is not a predictor of intention to use MIs in any of three countries studied. We expected a negative relationship between age and intention to use MIs. Table 5 shows that, the willingness to use MI is much higher when it is used for health purposes. Since older people have more health issues they might be more prone to use MIs for specific use in healthcare (e.g. MI in peacemaker) than we expected. Therefore, this specific use would need to be researched in detail. So far, we cannot confirm that older people are less likely to use new tech- nologies. The results show that there are barriers related to pri- vacy issues that affect trust in MI. On the other hand, the safety of the technology has a significant positive impact on trust. Moreover, perceived trust in the technology of MI influences the decision to accept MI. In general, we can conclude that attitudes towards the acceptance of MIs are similar in all European countries considered. The results of this research could be useful for other research areas, especially for the healthcare industry, where the use of MI could contribute the most. The bot- tom line is that we are not yet ready to use MI. We could use it if it would benefit our health status. There are still many health and privacy issues to be addressed in order to achieve greater adoption of this technology in our daily lives. Although the current situation surrounding the Cov- id-19 pandemic has likely had a significant impact on indi- vidual perceptions of new technologies, the results of this study have made an important contribution to research on MI by providing an insight into perceptions of use from the end-user perspective. It is expected that willingness to adopt will increase as more applications of MI become available, not only in healthcare but also in daily life. 5.1 Practical implications MI has the potential to become an inevitable part of our lives in the near future. Not only its applications in healthcare, but also its everyday use could significantly change our lives. The recent situation has revealed even more potential MI applications in preventing the spread of pandemics. Identifying the key factors that influence attitudes towards MI is essential for organizations aiming to promote MI and support its widespread adoption. The research shows that despite the ease of use of MI, there is still too many fears about the privacy and safety of this technology. It is therefore necessary for public authorities to ensure standards and legislation that enable the safe use of MIs. Although several research papers show the present use of MIs by hobbyists and in certain work environments, there is still no research on the effect on the human body. Thus, before proposing a general use of MIs, research on health issues should be shifted from animals to humans so that individuals have enough information to make a rea- sonable decision about the use of MI. This study also highlighted the lack of knowledge about this technology. Despite the fact that MI was intro- duced as a passive device that cannot be tracked from dis- tance, the respondents were afraid of being tracked. On the other hand, most of them use mobile or wearable devices without proper security settings or even publish their lo- cation and status on social media and other platforms. It would therefore be crucial to educate the potential users about the real benefits and weaknesses of MIs before us- ing them in a particular setting. The research also found, that people would be more willing to use MI if it would benefit their health. Thus, if MIs with health benefits (e.g. measuring blood pressure or sugar) were offered to the public, they would be more readily accepted than identi- fication-only devices. 290 Organizacija, V olume 54 Issue 4, November 2021 Research Papers 5.2 Limitations and further research Unlike other technologies, MIs are quite specific, so there are many arguments for and against their adoption. The variables included in the proposed model were defined based on previous research and a pilot study. We do not exclude the possibility that other external variables could influence attitudes towards the adoption of MIs (e.g. (Mo- hamed, 2020) including religious concerns, which we be- lieve are not a major issue in the EMEA region where this study was conducted). The cost of implantation and use of MIs was also not considered in this study. Due to the small sample size, the Croatian subsample of this study with 146 responses (131 complete surveys) was not included in the analyses. Although we found some similarities with studies per- formed on other continents (Franks & Smith, 2021; Peraks- lis & Michael, 2012), further research should include other regions or continents to obtain an overall picture of public acceptance of MI and identify the factors that influence the diversity of attitudes. In addition, other factors such as religious views, conspiracy mentality or online activi- ties could also be considered. Particular attention should be paid to changes in attitudes due to the current extreme health problems associated with the virus COVID-19. Acknowledgements This research was supported by the Slovenian Re- search Agency; Program No. P5-0018 – Decision Sup- port Systems in Digital Business and Prague University of Economics and Business, IG Agency, grant number F4/27/2019. Literature Achille, R., Perakslis, C., & Michael, K. (2012). Eth- ical Issues to Consider for Microchip Implants in Humans. Ethics in Biology, Engineering and Medi- cine, 3, 75–86. https://doi.org/10.1615/EthicsBiolo- gyEngMed.2013007009 Albrecht, K. (2010). Microchip-induced tumors in labora- tory rodents and dogs: A review of the literature 1990- 2006. 2010 IEEE International Symposium on Tech- nology and Society, 337–349. https://doi.org/10.1109/ISTAS.2010.5514622 Bansal, G., Zahedi, F. M., & Gefen, D. (2015). Do context and personality matter? Trust and privacy concerns in disclosing private information online. Information & Management, 53(1), 1–21. https://doi.org/10.1016/j. im.2015.08.001 Basham, R. (2014). Surveilling the Elderly: Emerging Demographic Needs and Social Implications of RFID Chip Technology Use. In M. G. Michael & K. Michael (Eds.), Uberveillance and the Social Implications of Microchip Implants: Emerging Technologies (pp. 169– 185). IGI Global. https://doi.org/10.4018/978-1-4666- 4582-0.ch007 Beaujean, A. A. (2014). Latent variable modeling using R: A step-by-step guide. In Latent variable modeling us- ing R: A step-by-step guide. Routledge/Taylor & Fran- cis Group. https://doi.org/10.4324/9781315869780 Benjamini, Y ., & Hochberg, Y . (1995). Controlling the False Discovery Rate: A Practical and Powerful Ap- proach to Multiple Testing. Journal of the Royal Sta- tistical Society. Series B (Methodological), 57(1), 289–300. http://www.jstor.org/stable/2346101 Carr, N. K. (2020). As Society Strives for Reduced Contact during the Pandemic, How Can Human Microchipping Help? Villanova Law Review, 65, 46–60. https://digi- talcommons.law.villanova.edu/vlr/vol65/iss6/3 Chen, F. F. (2007). Sensitivity of Goodness of Fit Indexes to Lack of Measurement Invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), 464– 504. https://doi.org/10.1080/10705510701301834 Cheung, G. W., & Lau, R. S. (2011). A Direct Comparison Approach for Testing Measurement Invariance. Orga- nizational Research Methods, 15(2), 167–198. https:// doi.org/10.1177/1094428111421987 Cheung, G. W., & Rensvold, R. B. (2002). Evaluating goodness-of-fit indexes for testing measurement in- variance. In Structural Equation Modeling (V ol. 9, Is- sue 2, pp. 233–255). Lawrence Erlbaum. https://doi.org/10.1207/S15328007SEM0902_5 Davis, F. D. (1989). Perceived Usefulness , Perceived Ease Of Use , and User Acceptance. MIS Quarterly, 13(3), 319–339. https://doi.org/10.2307/249008 Fornell, C., & Larcker, D. F. (1981). Evaluating Structur- al Equation Models with Unobservable Variables and Measurement Error. Journal of Marketing Research, 18(1), 39–50. https://doi.org/10.2307/3151312 Fram, B. R., Rivlin, M., & Beredjiklian, P. K. (2020). On Emerging Technology: What to Know When Your Pa- tient Has a Microchip in his Hand. Journal of Hand Surgery, 45(7), 645–649. https://doi.org/10.1016/j. jhsa.2020.01.008 Franks, C., & Smith, R. (2021). Changing perceptions of biometric technologies. In Research Report no. 20. Canberra: Australian Institute of Criminology. https:// doi.org/10.52922/rr78146 Gaffney, M., & Gopini, G. (2020). Attitude Towards Chip Implant Devices for Individual Enhancement Purpos- es on Example of Estonian Population . [University of Tartu]. https://web-proxy.io/proxy/dspace.ut.ee/bit- stream/handle/10062/67048/gaffney_matthew_and_ gopini_gopichand.pdf?sequence=4&isAllowed=y Gauttier, S. (2019). “I” ve got you under my skin’ – the role of ethical consideration in the (non-) acceptance of in- 291 Organizacija, V olume 54 Issue 4, November 2021 Research Papers sideables in the workplace. Technology in Society, 56, 93–108. https://doi.org/10.1016/j.techsoc.2018.09.008 Hair, J. F., Black, W. C., Babin, B. J., & Anderson, R. E. (2019). Multivariate Data Analysis (8th ed.). Cengage Learning. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Con- ventional criteria versus new alternatives. Struc- tural Equation Modeling, 6(1), 1–55. https://doi. org/10.1080/10705519909540118 Jorgensen, T. D., Kite, B. A., Chen, P.-Y ., & Short, S. D. (2018). Permutation randomization methods for test- ing measurement equivalence and detecting differen- tial item functioning in multiple-group confirmatory factor analysis. In Psychological Methods (V ol. 23, Issue 4, pp. 708–728). American Psychological Asso- ciation. https://doi.org/10.1037/met0000152 Jorgensen, T. D., S, P., Schoemann, A. M., Rosseel, Y ., Miller, P., Quick, C., Garnier-Villarreal, M., Selig, J., Boulton, A., Preacher, K., Coffmann, D., Rhemtulla, M., Robitzsch, A., Enders, C., Arslan, R., Clinton, B., Panko, P., Merkle, E., Chesnut, S., … Ben-Shachar, M. S. (2020). Useful Tools for Structural Equation Model- ing. https://cran.r-project.org/web/packages/semTools/ semTools.pdf Katz, J. E., & Rice, R. E. (2009). Public views of mobile medical devices and services: A US national survey of consumer sentiments towards RFID healthcare technology. International Journal of Medical Infor- matics, 78, 104–114. https://doi.org/10.1016/j.ijmed- inf.2008.06.001 Koufteros, X. A. (1999). Testing a model of pull produc- tion: a paradigm for manufacturing research using structural equation modeling. Journal of Operations Management, 17(4), 467–488. https://doi.org/10.1016/ S0272-6963(99)00002-9 Lai, H. C., Chan, H. W., & Singh, N. P. (2016). Effects of radiation from a radiofrequency identification (RFID) microchip on human cancer cells. International Jour- nal of Radiation Biology, 92(3), 156–161. https://doi.org/10.3109/09553002.2016.1135264 Lockton, V ., & Rosenberg, R. S. (2005). RFID: The next serious threat to privacy. Ethics and Information Technology, 7(4), 221–231. https://doi.org/10.1007/ s10676-006-0014-2 MacCallum, R. C., Browne, M. W., & Sugawara, H. M. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods, 1(2), 130–149. https://doi.org/10.1037/1082- 989X.1.2.130 Miceli, G. N., & Brabaranelli, C. (2016). Structural Equa- tions Modeling: Theory and Applications in Strategic Management. In G. B. Dagnino & C. M. C (Eds.), Research Methods for Strategic Management (pp. 98– 136). Routledge/Taylor & Francis Group. https://doi. org//10.1155/2012/263953 Michael, K., Aloudat, A., Michael, M. G., & Perakslis, C. (2017). You Want to do What with RFID? Perceptions of radio-frequency identification implants for employ- ee identification in the workplace. IEEE Consum- er Electronics Magazine, 6(3), 111–117. https://doi. org/10.1109/MCE.2017.2684978 Michael, K., & Michael, M. G. (2013). The future prospects of embedded microchips in humans as unique identifiers: the risks versus the rewards. Me- dia, Culture & Society, 35(1), 78–86. https://doi. org/10.1177/0163443712464561 Michael, M. G., & Michael, K. (2010). Toward a State of Überveillance [Special Section Introduction]. IEEE Technology and Society Magazine, 29(2), 9–16. https:// doi.org/10.1109/MTS.2010.937024 Mohamed, M. A. (2020). Modeling of Subcutaneous Im- plantable Microchip Intention of Use. In T. Ahram, W. Karwowski, A. Vergnano, F. Leali, & R. Taiar (Eds.), Intelligent Human Systems Integration 2020 (pp. 842– 847). Springer International Publishing. https://doi. org//10.1007/978-3-030-39512-4_128 Perakslis, C., & Michael, K. (2012). Indian Millennials: Are microchip implants a more secure technology for identification and access control? Proceedings of the 2012 IEEE Conference on Technology and So- ciety in Asia, T and SA 2012, 2012, 1–9. https://doi. org/10.1109/TSAsia.2012.6397977 Perakslis, C., Michael, K., Michael, M. G., & Gable, R. (2014). Perceived Barriers for Implanting Microchips in Humans : A Transnational Study. In Norbert Wiener in the 21st Century (21CW). https://doi.org/10.1109/ NORBERT.2014.6893929 Putnick, D. L., & Bornstein, M. H. (2016). Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Developmental Review, 41, 71–90. https://doi.org/ https://doi.org/10.1016/j.dr.2016.06.004 Rodriguez, D. A. (2019). Chipping in at work: Privacy con- cerns related to the use of body microchip (“RFID”) implants in the employer-employee context. Iowa Law Review, 104(3), 1581–1611. https://ilr.law.uiowa.edu/ assets/Uploads/ILR-104-3-Rodriguez.pdf Rosseel, Y . (2014). The lavaan tutorial. In The lavaan tuto- rial. http://lavaan.ugent.be/tutorial/tutorial.pdf Rotter, P., Daskala, B., & Compano, R. (2008). RFID im- plants: Opportunities and and challenges for identifying people. IEEE Technology and Society Magazine, 27(2), 24–32. https://doi.org/10.1109/mts.2008.924862 Sarwar, N., Khan, F. N., Ali, A., Rafique, H., Hussain, I., & Irshad, A. (2019). Microchip with Advance Human Monitoring Technique and RFTS. In I. S. Bajwa, F. Kamareddine, & A. Costa (Eds.), Intelligent Technolo- gies and Applications (pp. 560–570). Springer Singa- pore. https://doi.org//10.1007/978-981-13-6052-7_48 Schumacker, R. E., & Lomax, R. G. (2010). A beginner’s guide to structural equation modeling, 3rd ed. In A be- 292 Organizacija, V olume 54 Issue 4, November 2021 Research Papers ginner’ s guide to structural equation modeling, 3rd ed. Routledge/Taylor & Francis Group. Smith, C. (2008). Human microchip implantation. Journal of Technology Management and Innova- tion, 3(3), 151–156. https://doi.org/10.4067/S0718- 27242008000100015 Teo, T., & Zhou, M. (2014). Explaining the intention to use technology among university students: a structur- al equation modeling approach. Journal of Comput- ing in Higher Education, 26(2), 124–142. https://doi. org/10.1007/s12528-014-9080-3 Turner, W. (2020). Chipping Away at Workplace Privacy: The Implantation of RFID Microchips and Erosion of Employee Privacy. Washington University Journal of Law & Policy, 61, 275–298. https://openscholarship.wustl.edu/law_journal_law_ policy/vol61/iss1/18/ Virkki, J., Wei, Z., Liu, A., Ukkonen, L., & Björninen, T. (2017). Wearable passive E-textile UHF RFID tag based on a slotted patch antenna with sewn ground and microchip interconnections. International Journal of Antennas and Propagation, 2017, 27–29. https://doi. org/10.1155/2017/3476017 Werber, B., Baggia, A., & Žnidaršič, A. (2018). Factors Affecting the Intentions to Use RFID Subcutaneous Microchip Implants for Healthcare Purposes. Organi- zacija, 51(2), 121–133. https://doi.org/10.2478/orga- 2018-0010 Anja Žnidaršič is an Associate Professor of Quantitative Methods at the Faculty of Organizational Sciences, University of Maribor, Slovenia. Her main research interests are social network analysis, micro- enterprises, information-communication technology, students’ performance in methodological courses, and technology adoption. Alenka Baggia is an Assistant Professor of Information Systems at the Faculty of Organizational Sciences, University of Maribor. Her main research interests are digital literacy, technology acceptance, green information systems, and software quality. Antonín Pavlíček is Senior Assistant at Department of Informatics and Analytical Methods at University College of Business in Prague. His main research interests include new media, social media, information management and information security. Jakub Fischer is the Dean of the Faculty of Informatics and Statistics of the Prague University of Economics and Business. He focuses his research on the field of social and economic statistics and national accounting. Maciej Rostański is an Assistant Professor at the Computer Science Faculty of WSB Academy in Dąbrowa Górnicza, Poland. His main research interests include cloud computing, cyber security and Internet of Things. Borut Werber is an Assistant Professor of Information Systems at the Faculty of Organizational Sciences, University of Maribor, Slovenia. His main research interests are micro-enterprises, information- communication technology, and novel technologies. Smo pripravljeni uporabiti podkožni mikročip? Mednarodna presečna raziskava Ozadje in namen: Kljub očitni pomembnosti podkožnih mikročipov za naše življenje, jih v večini primerov ljudje še vedno obravnavamo kot negativne. Prevladuje mnenje, da ogrožajo našo zasebnost in lahko vplivajo na naše zdrav - je. V prispevku preučujemo pomembne dejavnike, ki vplivajo na dojemanje podkožnih mikročipov in pripravljenost posameznika, da bi podkožni mikročip uporabil za različne namene. Metodologija: Presečno študijo smo izvedli v treh evropskih državah, podatke pa smo analizirali s pomočjo mode - liranja strukturnih enačb. V analizo smo vključili priložnostni vzorec 804 anketirancev, ki so v celoti izpolnili spletni vprašalnik. Rezultati: Rezultati kažejo, da lahko na osnovi konstruktov zaznana enostavnost uporabe, uporabnost in zaznano zaupanje napovemo pripravljenost za uporabo podkožnega mikročipa. Na zaznano zaupanje vplivata zasebnost in varnost tehnologije. Zaznano uporabnost podkožnega mikročipa zmanjšuje zaskrbljenost zaradi bolečih postopkov vstavljanja in skrbi glede zdravja. Razen vpliva konstrukta skrb za zdravje, so rezultati podobni v vseh državah. Zaključki: Prikazani rezultati lahko služijo kot dobra osnova za nadaljnje raziskave glede dejanske uporabe pod - kožnih mikročipov. Rezultati lahko snovalcem zakonodaje pomagajo pri oblikovanju usmeritev in predpisov, ki bodo zagotovili varno uporabo podkožnih mikročipov in zagotovili višjo stopnjo varnosti. V nadaljevanju bi bilo smiselno raziskati ali se je pripravljenost za uporabo podkožnih mikročipov spremenila po izkušnji z globalno pandemijo. Ključne besede: Podkožni mikročip, Komunikacija kratkega dosega, Vedenjske namere, Model strukturne enačbe, Model sprejetja tehnologije