Advertisement

The PMA Scale: A Measure of Physicians’ Motivation to Adopt Medical Devices

Open AccessPublished:January 24, 2017DOI:https://doi.org/10.1016/j.jval.2016.12.002

      Abstract

      Background

      Studies have often stated that individual-level determinants are important drivers for the adoption of medical devices. Empirical evidence supporting this claim is, however, scarce. At the individual level, physicians’ adoption motivation was often considered important in the context of adoption decisions, but a clear notion of its dimensions and corresponding measurement scales is not available.

      Objectives

      To develop and subsequently validate a scale to measure the motivation to adopt medical devices of hospital-based physicians.

      Methods

      The development and validation of the physician-motivation-adoption (PMA) scale were based on a literature search, internal expert meetings, a pilot study with physicians, and a three-stage online survey. The data collected in the online survey were analyzed using exploratory factor analysis (EFA), and the PMA scale was revised according to the results. Confirmatory factor analysis (CFA) was conducted to test the results from the EFA in the third stage. Reliability and validity tests and subgroup analyses were also conducted.

      Results

      Overall, 457 questionnaires were completed by medical personnel of the National Health Service England. The EFA favored a six-factor solution to appropriately describe physicians’ motivation. The CFA confirmed the results from the EFA. Our tests indicated good reliability and validity of the PMA scale.

      Conclusions

      This is the first reliable and valid scale to measure physicians’ adoption motivation. Future adoption studies assessing the individual level should include the PMA scale to obtain more information about the role of physicians’ motivation in the broader adoption context.

      Keywords

      Introduction

      Decisions regarding whether to adopt a certain technology are among the most important medical and administrative decisions made in health care systems in general and in hospitals in particular [
      • Greenberg D.
      • Peterburg Y.
      • Vekstein D.
      • et al.
      Decisions to adopt new technologies at the hospital level: insights from Israeli medical centers.
      ] The process of adoption is, however, complex and influenced by many interacting factors [
      • Greenhalgh T.
      • Robert G.
      • Macfarlane F.
      • et al.
      Diffusion of innovations in service organizations: systematic review and recommendations.
      ,
      • Robert G.
      • Greenhalgh T.
      • MacFarlane F.
      • et al.
      Adopting and assimilating new non-pharmaceutical technologies into health care: a systematic review.
      ]. Factors that the literature has identified as being relevant to the adoption process have been categorized into four levels: the environmental, the organizational, the individual, and the innovation level. Thus far, research has predominantly focused on the organizational level [
      • Reyes P.M.
      • Li S.
      • Visich J.K.
      Accessing antecedents and outcomes of RFID implementation in health care.
      ]. The individual level has received less attention even though adoption decisions in hospitals are made by (groups of) individuals and underlie subjective influences [
      • Wernz C.
      • Zhang H.
      • Phusavat K.
      International study of technology investment decisions at hospitals.
      ,
      • Gold H.T.
      • Pitrelli K.
      • Hayes M.K.
      • et al.
      Decision to adopt medical technology: case study of breast cancer radiotherapy techniques.
      ]. Especially physicians largely impact the adoption of innovations in hospitals because they serve as initiators, supporters, and decision makers [
      • Greenberg D.
      • Peterburg Y.
      • Vekstein D.
      • et al.
      Decisions to adopt new technologies at the hospital level: insights from Israeli medical centers.
      ,
      • Wernz C.
      • Zhang H.
      • Phusavat K.
      International study of technology investment decisions at hospitals.
      ,
      • Gold H.T.
      • Pitrelli K.
      • Hayes M.K.
      • et al.
      Decision to adopt medical technology: case study of breast cancer radiotherapy techniques.
      ,
      • Barasa E.W.
      • Molyneux S.
      • English M.
      • et al.
      Setting healthcare priorities in hospitals: a review of empirical studies.
      ]. Physicians’ motivation has often been cited as crucial to adoption although empirical evidence supporting this assumption is rather scarce [
      • Greenhalgh T.
      • Robert G.
      • Macfarlane F.
      • et al.
      Diffusion of innovations in service organizations: systematic review and recommendations.
      ,
      • Godin G.
      • Bélanger-Gravel A.
      • Eccles M.
      • et al.
      Healthcare professionals’ intentions and behaviours: a systematic review of studies based on social cognitive theories.
      ,
      • Wisdom J.P.
      • Chor K.H.B.
      • Hoagwood K.E.
      • et al.
      Innovation adoption: a review of theories and constructs.
      ]. This notion is even more important considering that adoption decisions are also made in the absence of medical evidence for a medical technology [
      • Berliner E.
      Adopting medical technology.
      ]. In general, the motivation to adopt innovations—also described as innovativeness—is a multidimensional construct, and the sources of this motivation are manifold [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ,
      • Goldsmith R.E.
      The validity of a scale to measure global innovativeness.
      ,
      • Hoffmann S.
      • Soyez K.
      A cognitive model to predict domain-specific consumer innovativeness.
      ]. Defining the motivation construct entails complexity, which likely explains the lack of consistent and standard measurement methods [
      • Wisdom J.P.
      • Chor K.H.B.
      • Hoagwood K.E.
      • et al.
      Innovation adoption: a review of theories and constructs.
      ]. A reliable and valid measurement scale is an option to systematically incorporate motivation into future adoption research [
      • Goldsmith R.E.
      The validity of a scale to measure global innovativeness.
      ]. To increase the applicability of a scale, one should ensure that it can be used across different medical technologies and specialties. Nevertheless, the attributes of medical technologies differ, which is also reflected in adoption decisions. For example, medical devices feature functional aspects (i.e., size, form, usability, and learning efforts) that are relevant to the physician and that do not exist in or are not equally important for other health technologies such as pharmaceuticals. Normally, if such differences are not considered, the validity of a single scale decreases.
      Therefore, the objective of this study was to develop and validate a scale to measure physicians’ individual motivation to adopt medical devices excluding other medical technologies and components influencing adoption decisions that are not located on an individual level (e.g., objective medical evidence or financial criteria). The goal was to develop a scale that is applicable across medical devices and specialties. The scale was developed to target hospital-based physicians. The data used to develop and validate the scale were obtained from the literature, face-to-face and telephone interviews conducted with physicians in a pilot study, and a subsequent online survey of physicians used by the National Health Service (NHS) England. The data were analyzed using exploratory factor analysis (EFA) and confirmatory factor analysis (CFA).

      Theoretical Background

      To our knowledge, there are no theories or conceptual frameworks that were specifically designed to explain the influence of physicians’ motivation on technology adoption in a health care context. Nevertheless, related theories exist that try to explain how technological or behavioral adoption occurs on an individual level, such as the innovation diffusion theory [
      • Rogers E.M.
      ], the technology acceptance model (TAM) [
      • Davis F.D.
      A Technology Acceptance Model for Empirically Testing New End-User Information Systems: Theory and Results.
      ,
      • Venkatesh V.
      • Davis F.D.
      A theoretical extension of the Technology Acceptance Model: four longitudinal field studies.
      ,
      • Venkatesh V.
      • Bala H.
      Technology Acceptance Model 3 and a research agenda on interventions.
      ], the theory of reasoned action (TRA) [
      • Fishbein M.
      • Ajzen I.
      Belief, Attitude, Intention, and Behavior:.
      ], or the theory of planned behavior (TPB) [
      • Ajzen I.
      From intentions to actions: a theory of planned behavior.
      ]. Although these theories do not focus on individual motivation per se, they include elements such as social influence, subjective norms, or perceived usefulness that are related to the motivation concept. The relevance of social and functional aspects for adoption decisions is also acknowledged in the consumer innovativeness and motivation literature [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ,
      • Goldsmith R.E.
      The validity of a scale to measure global innovativeness.
      ,
      • Bartels J.
      • Reinders M.J.
      Consumer innovativeness and its correlates: a propositional inventory for future research.
      ,
      • Venkatraman M.P.
      The impact of innovativeness and innovation type on adoption.
      ,
      • Sheth J.N.
      • Newman B.I.
      • Gross B.L.
      Why we buy what we buy—a theory of consumption values.
      ]. In addition, authors of the innovativeness literature identify cognitive and hedonic aspects as being important drivers of one’s motivation to adopt [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ,
      • Venkatraman M.P.
      The impact of innovativeness and innovation type on adoption.
      ].
      The behavioral adoption theories and the innovativeness and motivation literature have common ground and offer general elements that can serve as a theoretical starting point to study physicians’ motivation to adopt medical technologies. Nevertheless, a useful scale has to operationalize these general elements with respect to the specific characteristics of physicians’ adoption situations in health care. In health care, physicians are concerned with decisions to adopt technologies that are used for the treatment of patients. This is a major difference compared with adoption decisions in a consumer context in which the adopted technologies are used for oneself. Therefore, it is necessary to incorporate an additional component capturing physicians’ motivation to improve patients’ benefit by adopting medical technologies [
      • Teplensky J.D.
      • Pauly M.V.
      • Kimberly J.R.
      • et al.
      Hospital adoption of medical technology: an empirical test of alternative models.
      ]. Similar attempts have been made in studies that have tried to operationalize TAM, TRA, or TPB in a health care context but they have suffered from a lack of standardization [
      • Holden R.J.
      • Karsh B.T.
      The Technology Acceptance Model: its past and its future in health care.
      ,
      • Yarbrough A.K.
      • Smith T.B.
      Technology acceptance among physicians—a new take on TAM.
      ]. Furthermore, these studies have subsumed patient benefit under perceived usefulness (i.e., in the form of increased quality of care or efficiency) and have not perceived the need to implement a separate dimension [
      • Holden R.J.
      • Karsh B.T.
      The Technology Acceptance Model: its past and its future in health care.
      ]. For our scale we have built on the adoption and innovativeness literature and thus have incorporated social, functional, hedonic, and cognitive elements as well as physicians’ orientation toward patients.

      Methods

       Initial Item Collection

      Because no specific scale assessing the adoption of medical devices is currently available, we used the validated, multidimensional innovativeness scale developed by Vandecasteele and Geuens as the starting point of our item collection [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ,
      • Goldsmith R.E.
      The validity of a scale to measure global innovativeness.
      ]. This scale, which was developed to analyze the consumer-product relationship, includes rather general dimensions that we assumed are also relevant in health care settings. The scale describes the concept of innovativeness through four dimensions, that is, functional, hedonic, social, and cognitive, that are based on goals and values underlying human actions [
      • Ford M.E.
      • Nichols C.W.
      A taxonomy of human goals and some possible applications.
      ,
      • Schwartz S.H.
      Universals in the content and structure of values: theoretical advances and empirical tests in 20 countries.
      ]. In addition to extracting items from this scale, we searched the adoption and innovativeness literature for related concepts [
      • Venkatraman M.P.
      The impact of innovativeness and innovation type on adoption.
      ,
      • Sheth J.N.
      • Newman B.I.
      • Gross B.L.
      Why we buy what we buy—a theory of consumption values.
      ,
      • Holden R.J.
      • Karsh B.T.
      The Technology Acceptance Model: its past and its future in health care.
      ,
      • Yarbrough A.K.
      • Smith T.B.
      Technology acceptance among physicians—a new take on TAM.
      ,
      • Tellis G.J.
      • Yin E.
      • Bell S.J.
      Global consumer innovativeness: cross-country differences and demographic commonalities.
      ,
      • Kennedy M.T.
      • Fiss P.C.
      Institutionalization, framing, and diffusion: the logic of TQM adoption and implementation decisions among US hospitals.
      ] and generated new items when appropriate. The items identified in our initial collection were further discussed and refined in meetings held with five health economists. The refinement consisted of adding and excluding items, changing item wording, and structuring the item selection. After the refinement process, our scale comprised six dimensions (i.e., Functional, Conformity, Power, Hedonic, Patient Benefit, and Cognitive) including 34 items.

       Pilot Study

      To increase the validity of the initial item selection, we conducted a pilot study comprising face-to-face and telephone interviews with 19 physicians with different medical specialties. During the in-depth interviews, the comprehensibility, relevance, completeness, wording, and order of the selected items were evaluated. All physicians confirmed the comprehensibility, relevance, and completeness of the proposed dimensions. The wording and order of the items were, however, revised on the basis of physicians’ feedback, and some items were deleted or consolidated so that the initial scale distributed online comprised 28 items (see Appendix A in Supplemental Materials found at doi:10.1016/j.jval.2016.12.002).

       Data Collection

      Data collection was carried out using an online survey that was administered via EFS survey software (QuestBack, Oslo, Norway). The objective was to develop a scale that can be used in a medical context and is flexible enough to be completed by physicians with different specialties who had been involved in decisions to adopt medical devices.
      Invitation letters were sent via email to physicians listed in the Named Clinician Report (as of June 2011) of the NHS Digital, which is an executive nondepartmental public body of the Department of Health. The Department of Health is responsible for England. Email addresses were constructed on the basis of names and health care provider information obtained from the report. The physicians were contacted in three consecutive stages, and the observations collected in each stage were analyzed using EFA. After each stage, we revised the scale according to the results. In stage 1, all physicians were invited to participate in the survey. In stage 2, reminders were sent to the physicians. Finally, in stage 3, all physicians, including those who had already responded to the survey, were again asked to participate in the final version of the survey. Overall, the data collection phase ran from August 12, 2015, to September 21, 2015.

       Measurement

      Invited physicians were asked to think of real adoption decisions that they had been involved in. This method ensured that the collected data were based on different kinds of medical devices and reflected realistic situations. All items of the six postulated motivation dimensions were measured on a five-point Likert scale. Respondents reported their agreement with the item statements using the following response options: strongly disagree, disagree, neutral, agree, and strongly agree. In addition, we measured the relative importance of the dimensions by asking the participants to allocate 100 points in total to the six dimensions.
      Filter questions were included at the beginning of the survey to determine whether physicians had been actively involved in decisions to adopt medical devices. Respondents who were not involved in such decisions were excluded. We also asked the physicians to provide an estimate of how strong their influence on the adoption decision is. The corresponding item ranged from 1 (no influence on the adoption decision) to 6 (very strong influence on the adoption decision). Additional questions about respondent and hospital characteristics (e.g., position in hospital hierarchy, medical specialty, and hospital size) were included at the end of the survey.

       Data Preparation and Analysis

      Before we conducted the EFA, we tested sampling adequacy using the Kaiser-Meyer-Olkin (KMO) criterion [
      • Poggi A.
      Job satisfaction, working conditions and aspirations.
      ,
      • Williams B.
      • Brown T.
      • Onsman A.
      Exploratory factor analysis: a five-step guide for novices.
      ]. KMO values range between 0 and 1, and values above 0.9 are considered as marvelous [
      • Kaiser H.F.
      Index of factorial simplicity.
      ]. For the EFA, we chose to perform principal axis factoring because our aim was to check whether the shared variance of our items can be explained by unobservable constructs (i.e., latent hypothetical constructs or factors) rather than to simply reduce the number of items on the basis of complete variance [
      • Sainfort F.
      The use of factor analysis techniques.
      ,
      • Conway J.M.
      • Huffcutt A.I.
      A review and evaluation of exploratory factor analysis practices in organizational research.
      ,
      • Park H.S.
      • Dailey R.
      • Lemus D.
      The use of exploratory factor analysis and principal components analysis in communication research.
      ]. Although an EFA is not an exact method of testing theoretical hypotheses, it allows us to draw conclusions about the construct validity of self-report scales [
      • Williams B.
      • Brown T.
      • Onsman A.
      Exploratory factor analysis: a five-step guide for novices.
      ,
      • O’Leary-Kelly S.W.
      • Vokurka R.J.
      The empirical assessment of construct validity.
      ]. The number of factors extracted was based on our preceding theoretical considerations, the KMO criterion (all eigenvalues >1), a corresponding scree plot, and parallel analysis [
      • Conway J.M.
      • Huffcutt A.I.
      A review and evaluation of exploratory factor analysis practices in organizational research.
      ,
      • Park H.S.
      • Dailey R.
      • Lemus D.
      The use of exploratory factor analysis and principal components analysis in communication research.
      ]. As recommended in the literature, promax (oblique) rotation was applied to the principal axis factoring results to achieve better interpretability of the results [
      • Conway J.M.
      • Huffcutt A.I.
      A review and evaluation of exploratory factor analysis practices in organizational research.
      ,
      • Henson R.K.
      • Roberts J.K.
      Use of exploratory factor analysis in published research—common errors and some comment on improved practice.
      ,
      • Reise S.P.
      • Waller N.G.
      • Comrey A.L.
      Factor analysis and scale revision.
      ]. Promax rotation allows factors to be correlated, an assumption that cannot be overlooked [
      • Park H.S.
      • Dailey R.
      • Lemus D.
      The use of exploratory factor analysis and principal components analysis in communication research.
      ,
      • Fabrigar L.R.
      • Wegener D.T.
      • MacCallum R.C.
      • et al.
      Evaluating the use of exploratory factor analysis in psychological research.
      ]. To test the scale’s internal consistency reliability, we calculated Cronbach alpha [
      • Tavakol M.
      • Dennick R.
      Making sense of Cronbach’s alpha. Int.
      ]. Although items that load onto a specific factor should show high intercorrelations, that is, high internal consistency, items across different factors should exhibit low correlations [
      • DeVon H.A.
      • Block M.E.
      • Moyle-Wright P.
      • et al.
      A psychometric toolbox for testing validity and reliability.
      ,
      • Huysamen G.K.
      Coefficient alpha: unnecessarily ambiguous; unduly ubiquitous.
      ]. We applied the described methods to the data collected in each stage of our survey. After each stage, we deleted or reworded items with factor loadings lower than 0.32 or which loaded onto two or more factors [
      • Worthington R.L.
      • Whittaker T.A.
      Scale development research—a content analysis and recommendations for best practices.
      ].
      A CFA was conducted on the results from the final EFA to test the postulated relationships between latent constructs and items [
      • Jackson D.L.
      • Gillaspy J.A.
      • Purc-Stephenson R.
      Reporting practices in confirmatory factor analysis: an overview and some recommendations.
      ]. It was conducted using data from stage 3 of our collection phase. The model was estimated using maximum likelihood. The chi-square value, the root mean square error of approximation (RMSEA), the standardized root mean square residual (SRMR), the Tucker-Lewis index (TLI), and the comparative fit index (CFI) were used to evaluate model fit [
      • Jackson D.L.
      • Gillaspy J.A.
      • Purc-Stephenson R.
      Reporting practices in confirmatory factor analysis: an overview and some recommendations.
      ,
      • Schmitt T.A.
      Current methodological considerations in exploratory and confirmatory factor analysis.
      ]. We calculated several different goodness-of-fit indices because there is substantial discussion about the reliability of the indices and their threshold values to evaluate model fit [
      • Fabrigar L.R.
      • Wegener D.T.
      • MacCallum R.C.
      • et al.
      Evaluating the use of exploratory factor analysis in psychological research.
      ,
      • Worthington R.L.
      • Whittaker T.A.
      Scale development research—a content analysis and recommendations for best practices.
      ,
      • Jackson D.L.
      • Gillaspy J.A.
      • Purc-Stephenson R.
      Reporting practices in confirmatory factor analysis: an overview and some recommendations.
      ].
      A good model fit is achieved if the null hypothesis of a likelihood ratio test on the basis of the chi-squared distribution is rejected [
      • Fabrigar L.R.
      • Wegener D.T.
      • MacCallum R.C.
      • et al.
      Evaluating the use of exploratory factor analysis in psychological research.
      ]. RMSEA values lower than 0.05 indicate good model fit, values from 0.05 to 0.08 acceptable fit, and values from 0.08 to 0.1 marginal fit [
      • Fabrigar L.R.
      • Wegener D.T.
      • MacCallum R.C.
      • et al.
      Evaluating the use of exploratory factor analysis in psychological research.
      ]. SRMR values lower than 0.1 point to an acceptable model fit [
      • Worthington R.L.
      • Whittaker T.A.
      Scale development research—a content analysis and recommendations for best practices.
      ]. The CFI and TLI values should be larger than 0.95 [
      • Hu L.T.
      • Bentler P.M.
      Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives.
      ]. Furthermore, we assessed the reliability of our scale by calculating composite reliability scores [
      • Werts C.E.
      • Linn R.L.
      • Joreskog K.G.
      Intraclass reliability estimates—testing structural assumptions.
      ,
      • Geldhof G.J.
      • Preacher K.J.
      • Zyphur M.J.
      Reliability estimation in a multilevel confirmatory factor analysis framework.
      ] and its validity by analyzing convergent and discriminant validity on the basis of a comparison of the average variance extracted (AVE) and the squared correlation values [
      • DeVon H.A.
      • Block M.E.
      • Moyle-Wright P.
      • et al.
      A psychometric toolbox for testing validity and reliability.
      ,
      • Fornell C.
      • Larcker D.F.
      Evaluating structural equation models with unobservable variables and measurement error.
      ]. Covergent validity is evident if the amount of variance explained by the latent construct is greater than the variance based on measurement error, that is, if the AVE value is higher than 0.5 [
      • Fornell C.
      • Larcker D.F.
      Evaluating structural equation models with unobservable variables and measurement error.
      ]. Discriminant validity is evident if the AVE values are higher than the squared correlation values [
      • Fornell C.
      • Larcker D.F.
      Evaluating structural equation models with unobservable variables and measurement error.
      ]. Together, convergent and discriminant validity are strong indicators of construct validity [
      • DeVon H.A.
      • Block M.E.
      • Moyle-Wright P.
      • et al.
      A psychometric toolbox for testing validity and reliability.
      ,
      • Thompson B.
      • Daniel L.G.
      Factor analytic evidence for the construct validity of scores: a historical overview and some guidelines.
      ]. Content validity was addressed through the pilot study [
      • DeVon H.A.
      • Block M.E.
      • Moyle-Wright P.
      • et al.
      A psychometric toolbox for testing validity and reliability.
      ]. All data analyses were conducted using STATA 13 SE (StataCorp LP, College Station, TX).

      Results

       Descriptive Results

      In total, we were able to identify the email addresses of 13,596 individuals listed in the Named Clinician Report; nevertheless, approximately one-third of the email addresses were obsolete at the time we distributed the questionnaire. A total of 230 questionnaires were completed in the first stage (completed after 14 days; see results of the first EFA in Appendix A in Supplemental Materials), 99 in the second stage (completed after 19 days; see results of the second EFA in Appendix B in Supplemental Materials found at doi:10.1016/j.jval.2016.12.002), and 128 in the final stage (completed after another 14 days), resulting in a total of 457 questionnaires received. After excluding physicians who had not been involved in the adoption of medical devices or who did not have influence on the adoption decision, 365 completed questionnaires remained. For the three stages of our analysis, the item used to measure respondents’ influence on the adoption decision had a mean value of 4.91 ± 0.80, 4.73 ± 0.89, and 4.98 ± 0.82, respectively. This result is equivalent to a strong influence. In the third stage, the sample almost exclusively consisted of physicians who were older (i.e., 46–55 years or 56–65 years) and male (see Table 1). All the following results presented refer to the final stage of the survey because it included the revised and final version of the physician-motivation-adoption scale (PMA scale). Results for stages 1 and 2 are presented in Appendices A and B, respectively, in Supplemental Materials.
      Table 1Descriptive results
      CovariateStage 1 (%)Stage 2 (%)Stage 3 (%)Overall (%)
      Profession
       Physician95979596
       Other5354
      Age (y)
       36–4513211515
       46–5549485450
       56–6536273032
       >652412
      Sex, male84838785
      Hospital position
       Consultant100100100100
       Specialty registrar0000
       Foundation doctor0000
      Medical specialty
       Surgery26242224
       Gynecology11111111
       Cardiology5596
       Pediatrics7887
       Ophthalmology31376
       Others (17 different)483904445
      Type of trust
       NHS acute trust42404442
       NHS foundation trust56595657
       Other2101
      Hospital size
       <50 beds2101
       50–200 beds6846
       201–500 beds24293428
       >500 beds69616265
      n
      n refers to the number of completed questionnaires after excluding physicians who had not adopted medical devices and physicians who had no influence on adoption decisions.
      (absolute values)
      18875102365
      No. of items to no. of observations ratio6.712.593.78
      NHS, National Health Service.
      low asterisk n refers to the number of completed questionnaires after excluding physicians who had not adopted medical devices and physicians who had no influence on adoption decisions.

       Exploratory and Confirmatory Factor Analyses

      The KMO value was 0.76, which is referred to as “middling” (values higher than 0.7 are deemed to be acceptable) [
      • Kaiser H.F.
      Index of factorial simplicity.
      ,
      • DeVon H.A.
      • Block M.E.
      • Moyle-Wright P.
      • et al.
      A psychometric toolbox for testing validity and reliability.
      ]. In addition, none of the single items had a value lower than 0.5.
      We extracted six factors on the basis of the KMO criterion (all eigenvalues >1), the scree plot (bend between the sixth and seventh factors), and the parallel analysis (see Fig. 1) [
      • Conway J.M.
      • Huffcutt A.I.
      A review and evaluation of exploratory factor analysis practices in organizational research.
      ,
      • Reise S.P.
      • Waller N.G.
      • Comrey A.L.
      Factor analysis and scale revision.
      ,
      • Fabrigar L.R.
      • Wegener D.T.
      • MacCallum R.C.
      • et al.
      Evaluating the use of exploratory factor analysis in psychological research.
      ]. The total variance explained by the six factors was 94.39%. After promax rotation, the structure of the factor loadings was confirmed, and no substantial cross-loadings were found (see Table 2).
      Fig. 1
      Fig. 1Parallel analysis based on 50 random data sets.
      Table 2Overview of factor loadings
      ItemsFactor 1Factor 2Factor 3Factor 4Factor 5Factor 6
      F1: Reliability0.07–0.02–0.010.140.610.01
      F2: Time saving0.040.05–0.170.200.66–0.03
      F3: Practicability0.03–0.140.04–0.050.850.00
      F4: Facilitation–0.040.02–0.05–0.150.75–0.04
      Con1: Expectations0.020.210.040.70–0.050.01
      Con2: Advice0.06–0.020.020.86–0.06–0.03
      Con3: Utilization–0.04–0.150.090.830.060.01
      Con4: Majority opinion–0.050.04–0.010.550.050.10
      P1: Recognition0.760.17–0.050.06–0.050.04
      P2: Career0.880.060.010.07–0.02–0.10
      P3: Opinion leader0.950.00–0.09–0.040.040.04
      P4: Decision makers0.84–0.110.090.070.090.04
      P5: Future earnings0.710.030.000.09–0.08–0.03
      P6: Pioneer0.84–0.020.10–0.200.060.03
      H1: Passion0.190.410.24–0.010.130.07
      H2: Fun0.070.820.030.00–0.08–0.08
      H3: Emotions0.050.79–0.090.09–0.020.01
      H4: Excitement0.050.800.05–0.13–0.050.00
      PB1: Increased effort–0.070.110.18–0.090.220.60
      PB2: Time-consuming0.13–0.07–0.110.11–0.140.70
      PB3: Routine processes0.08–0.21–0.110.00–0.200.72
      PB4: Learning efforts–0.140.080.020.060.120.64
      PB5: Convince colleagues–0.020.080.09–0.140.070.54
      Cog1: Analytical mind–0.080.080.810.13–0.07–0.04
      Cog2: Intellectual challenge0.030.050.870.05–0.08–0.02
      Cog3: Improve skills0.10–0.150.83–0.020.11–0.10
      Cog4: Logical thinking–0.05–0.020.75–0.02–0.060.14
      Average weight of dimension in %, mean ± SD2.96 ± 4.095.87 ± 7.847.24 ± 7.485.55 ± 7.8227.29 ± 16.5251.09 ± 22.89
      Cronbach α0.940.850.880.810.780.75
      Note. Factor 1: Power; Factor 2: Hedonic; Factor 3: Cognitive; Factor 4: Conformity, Factor 5: Functional; Factor 6: Patient Benefit. Bold numbers: mutual factors loadings on hypothesized underlying factor.
      CFA confirmed the postulated relationships between items and latent constructs. All standardized factor loadings were higher than 0.4 and significant (P < 0.01). CFA results and goodness-of-fit indicators are presented in Table 3. The chi-squared statistic was 507.27 and the null hypothesis of the corresponding likelihood ratio test was not rejected. The RMSEA value for our model was 0.08, indicating at least marginal fit. The SRMR value for our model was 0.09, indicating an acceptable model fit. The CFI and TLI values were 0.87 and 0.86, respectively, indicating suboptimal model fit.
      Table 3CFA results and goodness-of-fit indicators
      ItemsFactor loadingsSEP valueError varianceCR
      Functional
       F1: Reliability0.550.080.000.700.80
       F2: Time saving0.600.070.000.64
       F3: Practicability0.930.050.000.14
       F4: Facilitation0.710.060.000.49
      Conformity
       Con1: Expectations0.760.050.000.420.83
       Con2: Advice0.940.040.000.11
       Con3: Utilization0.710.060.000.50
       Con4: Majority opinion0.490.080.000.76
      Power
       P1: Recognition0.850.030.000.270.94
       P2: Career0.950.020.000.10
       P3: Opinion leader0.900.020.000.18
       P4: Decision makers0.810.040.000.34
       P5: Future earnings0.750.050.000.44
       P6: Pioneer0.770.040.000.41
      Hedonic
       H1: Passion0.640.070.000.590.86
       H2: Fun0.880.030.000.22
       H3: Emotions0.790.050.000.38
       H4: Excitement0.790.050.000.37
      Patient Benefit
       PB1: Increased effort0.750.070.000.440.75
       PB2: Time-consuming0.490.100.000.76
       PB3: Routine processes0.500.100.000.75
       PB4: Learning efforts0.690.070.000.52
       PB5: Convince colleagues0.620.080.000.62
      Cognitive
       Cog1: Analytical mind0.820.040.000.330.88
       Cog2: Intellectual challenge0.930.030.000.14
       Cog3: Improve skills0.760.050.000.42
       Cog4: Logical thinking0.700.060.000.52
      Goodness-of-fit indicators
       χ2507.27 (P value 0.00)
       RMSEA0.08 (90% CI 0.067–0.092)
       CFI0.87
       TLI0.86
       SRMR0.09
      CI, confidence interval; CFA, confirmatory factor analysis; CIF, comparative fit index; CR, composite reliability; RMSEA, root mean square error of approximation; SE, standard error; SRMR, standardized root mean square residual; TLI, Tucker-Lewis index.

       Reliability and Validity Tests and Subgroup Analysis

      The values for Cronbach alpha ranged from 0.74 to 0.94. This suggests that the factors have good internal consistency reliability (see Table 2) [
      • DeVon H.A.
      • Block M.E.
      • Moyle-Wright P.
      • et al.
      A psychometric toolbox for testing validity and reliability.
      ,
      • Streiner D.L.
      Starting at the beginning: an introduction to coefficient alpha and internal consistency.
      ]. Composite reliability values derived from the CFA results ranged from 0.75 to 0.94, indicating good composite reliability [
      • Bagozzi R.P.
      • Yi Y.
      Specification, evaluation, and interpretation of structural equation models.
      ].
      Furthermore, different aspects of validity, including discriminant, convergent, and content validity, were evaluated. AVE and squared correlation values of the latent constructs are presented in Table 4. The AVE values of all the factors were higher than their squared correlation values, and therefore we can assume discriminant validity. In addition, all AVE values except the AVE value for the Patient Benefit factor were higher than 0.5, which indicates convergent validity of the remaining factors. The physicians who participated in the pilot study confirmed the relevance of the suggested items and factors; that is, they provided evidence for content validity [
      • O’Leary-Kelly S.W.
      • Vokurka R.J.
      The empirical assessment of construct validity.
      ].
      Table 4AVE and squared correlation values on the basis of CFA results
      DimensionsFunctionalConformityPowerHedonicPatient BenefitCognitive
      Functional0.51
      Conformity0.000.55
      Power0.010.120.71
      Hedonic0.000.090.330.61
      Patient Benefit0.030.010.000.010.38
      Cognitive0.020.020.090.270.070.65
      Note. AVE estimates are displayed in boldface on the diagonal; squared correlation values are presented below the diagonal.
      AVE, average variance extracted; CFA, confirmatory factor analysis.
      Subgroup analyses using EFA were conducted for three subgroups on the basis of the data from the final validation round. Respondents who indicated that they provided their answers on the basis of implantable devices (43 observations) and diagnostic devices (59 observations) formed the first subgroup and second subgroup, respectively. In addition, a third subgroup analysis was specifically carried out on observations from physicians (6 of the 102 observations in stage 3 were derived from other medical personnel).
      Again, all subgroup analyses resulted in six-factor solutions, and the loading patterns were nearly identical to the base case. As the number of observations decreased, cross-loadings could be observed for three items in the first subgroup analysis (i.e., implantable devices). No cross-loadings were found in the second and third subgroup analyses, confirming the structure of the PMA scale. For all subgroup analyses, the internal consistency reliability, as measured by Cronbach alpha, was satisfactory to excellent. (All these results are available from the authors on request.)

       The PMA Scale and Factor Weightings

      We termed the six resulting factors “Functional,” “Conformity,” “Power,” “Hedonic,” “Patient Benefit,” and “Cognitive.” Considering the relative importance of the factors, the Patient Benefit factor had the highest ranking, with 51.09 ± 22.89 percentage points, whereas the “Power” factor had the lowest ranking, with 2.96 ± 4.09 percentage points (see Table 2). The final 27-item PMA scale is presented in Table 5.
      Table 5PMA scale with the final item catalog
      Functional
      I supported the adoption of the medical innovation because in contrast to alternative technologies …
      F1… it is more reliable.
      F2… it saves time.
      F3… it is more practical.
      F4… it facilitates procedures.
      Conformity
      I supported the adoption of the medical innovation because …
      Con1… my colleagues in my organization expect me to do so.
      Con2… my peers advise me to do so.
      Con3… all of my peers are using it.
      Con4… I seek conformity with the majority opinion on medical innovations.
      Power
      I supported the adoption of the medical innovation to …
      P1… achieve professional recognition.
      P2… advance my career.
      P3… become established as a key opinion leader.
      P4… have an impact on decision makers.
      P5… increase my personal future earnings.
      P6… be recognized as a pioneer.
      Hedonic
      My adoption decisions are influenced by the fact that …
      H1… I am passionate about using the state of the art.
      H2… working with the latest medical innovations is fun.
      H3… using the latest technologies stimulates my emotions.
      H4… working with the latest medical innovations makes my work exciting.
      Patient Benefit
      If I was convinced that a certain medical innovation would benefit patients, I would adopt it …
      PB1… even if this would mean increased effort on my part.
      PB2… even if it was disproportionately time-consuming.
      PB3… even if it interfered with routine processes of my work.
      PB4… even if it demanded high learning efforts on my part.
      PB5… even if I had to convince colleagues with a conflictive opinion.
      Cognitive
      I think about the adoption of medical innovations that …
      Cog1… satisfy my analytical mind.
      Cog2… challenge me intellectually.
      Cog3… improve my medical and intellectual skills.
      Cog4… demand logical thinking.
      PMA, physician-motivation-adoption.

      Discussion

      The EFA and the CFA revealed that physicians’ motivation to adopt medical devices can be measured using the PMA scale, which consists of six distinct dimensions. These dimensions are theoretically meaningful, and we interpret them as follows.
      The first dimension, which we termed Functional, is similar to the “functional” dimension of the scale developed by Vandecasteele and Geuens [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ] and to the constructs “perceived usefulness” and “perceived ease of use” of the TAM [
      • Davis F.D.
      A Technology Acceptance Model for Empirically Testing New End-User Information Systems: Theory and Results.
      ,
      • Venkatesh V.
      • Davis F.D.
      A theoretical extension of the Technology Acceptance Model: four longitudinal field studies.
      ,
      • Venkatesh V.
      • Bala H.
      Technology Acceptance Model 3 and a research agenda on interventions.
      ]. It captures the increase in utility or functionality related to the adoption of a medical device. Hence, the underlying motivation for adopting medical devices is to gain a functional or technical advantage.
      The second dimension, which we termed Patient Benefit, describes physicians’ superordinate motivation to improve patients’ well-being. This dimension should not be confused with actual medical evidence. It was designed to capture physicians’ individual perceptions of any beneficial medical aspects of a medical device. These perceptions do not have to be in line with actual medical evidence in the sense of randomized controlled trial data. Instead, medical evidence needs to be considered in addition to the dimensions of the PMA scale when studying adoption decisions. To avoid physicians’ trivial confirmation of the importance of the Patient Benefit dimension, the corresponding items imply a trade-off between the benefits and the drawbacks of adoption. This was done to measure how much value physicians assign to the perceived patient benefit of a certain medical device. For instance, if a physician is strongly convinced of the beneficial aspects of a medical device, this physician would probably try to foster the adoption of that device even in the context of potential adoption hurdles (e.g., learning efforts and interference with routine processes). Our findings indicate that physicians consider the Patient Benefit dimension to be highly important in the adoption context (see factor weightings). In line with this result, Teplensky et al. [
      • Teplensky J.D.
      • Pauly M.V.
      • Kimberly J.R.
      • et al.
      Hospital adoption of medical technology: an empirical test of alternative models.
      ] considered patient orientation as physicians’ predominant motivation for adoption.
      The third dimension, Power, includes motives such as superiority, self-determination, and recognition. These motives describe physicians’ desire to improve their own situation and gain personal advantages [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ,
      • Schwartz S.H.
      Universals in the content and structure of values: theoretical advances and empirical tests in 20 countries.
      ]. The basic notion is that the adoption of a medical device can lead to a more successful career. A related concept is evident in the “social” dimension of the scale developed by Vandecasteele and Geuens [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ].
      The fourth dimension, Conformity, describes the desire to conform to peers and colleagues. It is assumed that adoption behavior reflects physicians’ desire to align with the norm or to meet expectations [
      • Venkatesh V.
      • Davis F.D.
      A theoretical extension of the Technology Acceptance Model: four longitudinal field studies.
      ,
      • Schwartz S.H.
      Universals in the content and structure of values: theoretical advances and empirical tests in 20 countries.
      ]. In this case, the adoption of a medical device could be used to close the gap with peers who have or have not made an adoption decision. A similar concept can be found in the TAM [
      • Venkatesh V.
      • Davis F.D.
      A theoretical extension of the Technology Acceptance Model: four longitudinal field studies.
      ].
      The fifth dimension, Hedonic, describes a personal desire to obtain the latest products and is related to positive feelings of enjoyment and satisfaction triggered by the adoption of a medical device. A similar concept in the field of consumer research can be found in the scale developed by Vandecasteele and Geuens [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ].
      The sixth dimension, Cognitive, captures physicians’ desire for cognitive stimulation and to expand their own cognitive limits. It is assumed that medical devices can be adopted to satisfy a desire for cognitive stimulation. This dimension is also similar to the “cognitive” dimension of the scale developed by Vandecasteele and Geuens [
      • Vandecasteele B.
      • Geuens M.
      Motivated consumer innovativeness: concept, measurement, and validation.
      ,
      • Ford M.E.
      • Nichols C.W.
      A taxonomy of human goals and some possible applications.
      ].
      Although a rather small number of observations were available for each round of validation, the overall characteristics of the sample were favorable. In particular, our sample consisted of responses from actual decision makers who indicated that they had a strong influence on adoption decisions. Moreover, all responding physicians were consultants and predominantly worked in large hospitals with more than 500 beds. Accordingly, our results are based on a sample from the desired target population, that is, physicians involved in the adoption of medical devices.
      Regarding the PMA scale, the EFA revealed a uniform structure of factor loadings without cross-loadings, allowing a clear interpretation of the latent constructs. The CFA confirmed the structure of the PMA scale. Model fit was acceptable regarding RMSEA and SRMR, but suboptimal regarding chi-squared, CFI, and TLI. It is known that model fit assessment is difficult given the small sample size [
      • Hu L.T.
      • Bentler P.M.
      Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives.
      ], and thus we believe that further goodness-of-fit assessments with larger sample size will be necessary in future studies. Reliability based on composite reliability was again favorable, with values exceeding 0.7. Convergent and discriminant validity were also confirmed except convergent validity for the Patient Benefit dimension. The AVE value for the Patient Benefit dimension was lower than 0.5, which means that the average amount of variance of the items explained by the Patient Benefit dimension was less than 50%. It was, however, higher than the squared correlation value, indicating that the items measure a distinct dimension. It is possible that further item deletion will solve this problem (and increase model fit); nevertheless, we think that this should be done using new data with more observations.

       Limitations

      This study is subject to limitations. The sample size used was rather small. This is especially true for the samples collected in stages 2 and 3. Although a general rule for a minimum sample size has yet not been defined [
      • Williams B.
      • Brown T.
      • Onsman A.
      Exploratory factor analysis: a five-step guide for novices.
      ], authors recommend a sample size of at least 100 observations or recommend a certain number of items to number of participants ratio [
      • Reise S.P.
      • Waller N.G.
      • Comrey A.L.
      Factor analysis and scale revision.
      ,
      • Fabrigar L.R.
      • Wegener D.T.
      • MacCallum R.C.
      • et al.
      Evaluating the use of exploratory factor analysis in psychological research.
      ,
      • Hair J.F.
      • Anderson R.E.
      • Tatham R.L.
      • et al.
      Multivariate Data Analysis.
      ]. We believe that the moderate model fit based on CFA was influenced by the small sample size in stage 3 and we advise authors to run a CFA and test the model’s goodness of fit before using the PMA scale. With new data it should also be possible to test whether deleting items increases model fit and convergent validity of the Patient Benefit dimension.
      A social desirability bias may have influenced the Power and Patient Benefit dimensions. From a social or moral perspective, it is generally not perceived desirable for a physician to admit that personal gain was the motivation for adoption (e.g., for career reasons). Therefore, we expected a tendency toward lower values of agreement for the power dimension. The bias concerning the Patient Benefit dimension should be in the opposite direction because helping patients is viewed as a physician’s primary goal. Because decisions to adopt a device are made by physicians who act as representatives for the actual consumer (i.e., the patients), physicians’ personal goals are expected to play a lesser role in adoption decisions than would the goals of the affected individuals. We assumed that the single items within the dimensions are equally affected by the biases. Therefore, the validity of the scale should not have been affected because the structure of the items, rather than the raw values, is important.
      It is possible that our study was prone to self-selection bias. Physicians with a distinct willingness to help patients might also be more likely to participate in a survey study. Therefore, physicians with a different attitude might have been under-represented in our study. Although this is a rather general limitation that applies to various types of survey research, we recommend that researchers interpret the values of the items conservatively with the understanding that the true values might be higher (e.g., Power) or lower (e.g., Patient Benefit). Nevertheless, again, the validity of the PMA scale should not have been affected by self-selection bias.
      Our study was validated using questionnaires completed by physicians employed by the NHS England. Researchers should be aware that our results might differ from those based on data collected in another population of physicians. For example, the typical career paths of physicians in other countries could vary, potentially affecting the mean value of the Power dimension. Furthermore, different health care systems could alter the decision autonomy of physicians. Nevertheless, we are convinced that heterogeneity does not impact the structure of our scale because it does not impact the loading patterns of items or the existence of motivation dimensions. Instead, heterogeneity is more likely to influence the relative importance of the items and dimensions and the relative impact of physicians’ motivation to adopt on adoption decisions. Therefore, we advise researchers to conduct a CFA before administering the PMA scale in other populations of physicians [
      • Costello A.
      • Osborne J.
      Best practices in exploratory factor analysis: four recommendations for getting the most from your analysis.
      ].

       Recommendations for the Application of the PMA Scale

      The PMA scale was designed to capture physicians’ individual motivation to adopt medical devices. This implies that the PMA scale alone is not sufficient to comprehensively explain adoption decisions. Other factors such as medical evidence, financial criteria, or manufacturer support are also important components of adoption decisions and need to be considered in addition to the PMA scale. Therefore, it is possible that factors such as medical evidence superimpose other factors such as physicians’ motivation regarding their impact on adoption decisions. Nevertheless, to get more insights about these relationships and the relative importance of different factors, further research that includes the PMA scale is needed.

       Implications for Future Research

      Several aspects of the dimensions of the PMA scale must be discussed in future research. First, it should be further explored whether a high or low agreement with a motivational dimension increases the likelihood of adoption of a medical device. For example, during our pilot study most physicians emphasized that an increase in patients’ benefit is their primary motivation for adopting medical devices. The same physicians also stated to be very hesitant to adopt medical devices in general and to strongly rely on medical evidence. These statements do not contradict each other. Instead, this might indicate that physicians who place more weight on patients’ benefit when considering the adoption of a medical device might be more risk-averse and might therefore have a lower likelihood of adoption unless strong and unequivocal medical evidence for using the medical device becomes available.
      Second, future research should explore whether physicians who agree with the Hedonic and Power dimensions tend to be early adopters and whether these physicians have a higher likelihood of adopting medical devices in general. It should also be clarified whether these assumptions hold in general or are specific to certain medical devices.
      Third, researchers should also investigate whether physicians’ motivation to adopt changes over a device’s life cycle. For example, one might assume that hedonic or cognitive motivations are present only in the case of “new” innovations and that these can “wear off” with time.
      Finally, research should explore whether physicians’ motivation overlaps with other factors that influence the adoption decision (e.g., medical evidence, financial aspects, and manufacturer activities) and how its relative importance can be assessed.

      Conclusions

      Although physicians’ motivation has been widely recognized as an important factor in the adoption of medical devices, it was a challenge to measure the latent construct because validated scales were not available [
      • Greenhalgh T.
      • Robert G.
      • Macfarlane F.
      • et al.
      Diffusion of innovations in service organizations: systematic review and recommendations.
      ]. The PMA scale is a consistent and valid tool for systematically measuring physicians’ motivation to adopt different medical devices. It consists of six dimensions, which we interpreted theoretically and termed Functional, Conformity, Power, Hedonic, Patient Benefit, and Cognitive. The implementation of the scale in future research may help to improve our understanding of the role of individual factors in the adoption of medical devices. Moreover, the PMA scale can help to improve our understanding of the role of motivation in group decision making and to expand our knowledge of adoption and subsequent diffusion processes in hospitals.
      Taken together, the PMA scale fills an important gap in the literature and can be used to increase the current knowledge of physicians’ thought processes and adoption decisions.

      Acknowledgments

      We are grateful for all the important comments and the feedback that the interviewed physicians provided during the pilot study. We also thank all the survey respondents for their participation. In addition, we thank Martina Stoppel and Mathias Baumann for their administrative support of our study as well as Vera Winter for her valuable comments on our article.

      Supplemental Materials

      References

        • Greenberg D.
        • Peterburg Y.
        • Vekstein D.
        • et al.
        Decisions to adopt new technologies at the hospital level: insights from Israeli medical centers.
        Int J Technol Assess Health Care. 2005; 21: 219-227
        • Greenhalgh T.
        • Robert G.
        • Macfarlane F.
        • et al.
        Diffusion of innovations in service organizations: systematic review and recommendations.
        Milbank Q. 2004; 82: 581-629
        • Robert G.
        • Greenhalgh T.
        • MacFarlane F.
        • et al.
        Adopting and assimilating new non-pharmaceutical technologies into health care: a systematic review.
        J Health Serv Res Policy. 2010; 15: 243-250
        • Reyes P.M.
        • Li S.
        • Visich J.K.
        Accessing antecedents and outcomes of RFID implementation in health care.
        Int J Prod Econ. 2012; 136: 137-150
        • Wernz C.
        • Zhang H.
        • Phusavat K.
        International study of technology investment decisions at hospitals.
        Ind Manage Data Syst. 2014; 114: 568-582
        • Gold H.T.
        • Pitrelli K.
        • Hayes M.K.
        • et al.
        Decision to adopt medical technology: case study of breast cancer radiotherapy techniques.
        Med Decis Making. 2014; 34: 1006-1015
        • Barasa E.W.
        • Molyneux S.
        • English M.
        • et al.
        Setting healthcare priorities in hospitals: a review of empirical studies.
        Health Policy Plan. 2015; 30: 386-396
        • Godin G.
        • Bélanger-Gravel A.
        • Eccles M.
        • et al.
        Healthcare professionals’ intentions and behaviours: a systematic review of studies based on social cognitive theories.
        Implement Sci. 2008; 3: 36
        • Wisdom J.P.
        • Chor K.H.B.
        • Hoagwood K.E.
        • et al.
        Innovation adoption: a review of theories and constructs.
        Adm Policy Ment Health. 2014; 41: 480-502
        • Berliner E.
        Adopting medical technology.
        Med Decis Making. 2014; 34: 948-950
        • Vandecasteele B.
        • Geuens M.
        Motivated consumer innovativeness: concept, measurement, and validation.
        Int J Res Mark. 2010; 27: 308-318
        • Goldsmith R.E.
        The validity of a scale to measure global innovativeness.
        J Appl Bus Res. 2011; 7: 89-97
        • Hoffmann S.
        • Soyez K.
        A cognitive model to predict domain-specific consumer innovativeness.
        J Bus Res. 2010; 63: 778-785
        • Rogers E.M.
        Diffusion of Innovations. 5th ed. Free Press, New York, NY2003
        • Davis F.D.
        A Technology Acceptance Model for Empirically Testing New End-User Information Systems: Theory and Results.
        Sloan School of Management, Massachussetts Institute of Technology, Cambridge, MA1986
        • Venkatesh V.
        • Davis F.D.
        A theoretical extension of the Technology Acceptance Model: four longitudinal field studies.
        Manage Sci. 2000; 46: 186-204
        • Venkatesh V.
        • Bala H.
        Technology Acceptance Model 3 and a research agenda on interventions.
        Decis Sci. 2008; 39: 273-315
        • Fishbein M.
        • Ajzen I.
        Belief, Attitude, Intention, and Behavior:.
        An Introduction to Theory and Research. Addison-Wesley, Reading, MA1975
        • Ajzen I.
        From intentions to actions: a theory of planned behavior.
        in: Kuhl J. Beckmann J. Action Control: From Cognition to Behavior. Springer-Verlag, New York, NY1985
        • Bartels J.
        • Reinders M.J.
        Consumer innovativeness and its correlates: a propositional inventory for future research.
        J Bus Res. 2011; 64: 601-609
        • Venkatraman M.P.
        The impact of innovativeness and innovation type on adoption.
        J Retailing. 1991; 67: 51-67
        • Sheth J.N.
        • Newman B.I.
        • Gross B.L.
        Why we buy what we buy—a theory of consumption values.
        J Bus Res. 1991; 22: 159-170
        • Teplensky J.D.
        • Pauly M.V.
        • Kimberly J.R.
        • et al.
        Hospital adoption of medical technology: an empirical test of alternative models.
        Health Serv Res. 1995; 30: 437-465
        • Holden R.J.
        • Karsh B.T.
        The Technology Acceptance Model: its past and its future in health care.
        J Biomed Inform. 2010; 43: 159-172
        • Yarbrough A.K.
        • Smith T.B.
        Technology acceptance among physicians—a new take on TAM.
        Med Care Res Rev. 2007; 64: 650-672
        • Ford M.E.
        • Nichols C.W.
        A taxonomy of human goals and some possible applications.
        in: Ford M.E. Ford D.H. Humans as Self-Constructing Systems: Putting the Framework to Work. Erlbaum, New York, NY1987
        • Schwartz S.H.
        Universals in the content and structure of values: theoretical advances and empirical tests in 20 countries.
        in: Zanna M.P. Advances in Experimental Social Psychology. Academic Press, Inc., San Diego, CA1992
        • Tellis G.J.
        • Yin E.
        • Bell S.J.
        Global consumer innovativeness: cross-country differences and demographic commonalities.
        J Int Mark. 2009; 17: 1-22
        • Kennedy M.T.
        • Fiss P.C.
        Institutionalization, framing, and diffusion: the logic of TQM adoption and implementation decisions among US hospitals.
        Acad Manage J. 2009; 52: 897-918
        • Poggi A.
        Job satisfaction, working conditions and aspirations.
        J Econ Psychol. 2010; 31: 936-949
        • Williams B.
        • Brown T.
        • Onsman A.
        Exploratory factor analysis: a five-step guide for novices.
        J Emerg Prim Health Care. 2010; 8: 1-13
        • Kaiser H.F.
        Index of factorial simplicity.
        Psychometrika. 1974; 39: 31-36
        • Sainfort F.
        The use of factor analysis techniques.
        Med Decis Making. 2002; 22: 451-452
        • Conway J.M.
        • Huffcutt A.I.
        A review and evaluation of exploratory factor analysis practices in organizational research.
        Organ Res Methods. 2003; 6: 147-168
        • Park H.S.
        • Dailey R.
        • Lemus D.
        The use of exploratory factor analysis and principal components analysis in communication research.
        Hum Commun Res. 2002; 28: 562-577
        • O’Leary-Kelly S.W.
        • Vokurka R.J.
        The empirical assessment of construct validity.
        J Oper Manage. 1998; 16: 387-405
        • Henson R.K.
        • Roberts J.K.
        Use of exploratory factor analysis in published research—common errors and some comment on improved practice.
        Educ Psychol Meas. 2006; 66: 393-416
        • Reise S.P.
        • Waller N.G.
        • Comrey A.L.
        Factor analysis and scale revision.
        Psychol Assess. 2000; 12: 287-297
        • Fabrigar L.R.
        • Wegener D.T.
        • MacCallum R.C.
        • et al.
        Evaluating the use of exploratory factor analysis in psychological research.
        Psychol Methods. 1999; 4: 272-299
        • Tavakol M.
        • Dennick R.
        Making sense of Cronbach’s alpha. Int.
        J Med Educ. 2011; 2: 53-55
        • DeVon H.A.
        • Block M.E.
        • Moyle-Wright P.
        • et al.
        A psychometric toolbox for testing validity and reliability.
        J Nurs Scholarsh. 2007; 39: 155-164
        • Huysamen G.K.
        Coefficient alpha: unnecessarily ambiguous; unduly ubiquitous.
        SA J Ind Pyschol. 2006; 32: 34-40
        • Worthington R.L.
        • Whittaker T.A.
        Scale development research—a content analysis and recommendations for best practices.
        Couns Psychol. 2006; 34: 806-838
        • Jackson D.L.
        • Gillaspy J.A.
        • Purc-Stephenson R.
        Reporting practices in confirmatory factor analysis: an overview and some recommendations.
        Psychol Methods. 2009; 14: 6-23
        • Schmitt T.A.
        Current methodological considerations in exploratory and confirmatory factor analysis.
        J Psychoeduc Assess. 2011; 29: 304-321
        • Hu L.T.
        • Bentler P.M.
        Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives.
        Struct Equ Modeling. 1999; 6: 1-55
        • Werts C.E.
        • Linn R.L.
        • Joreskog K.G.
        Intraclass reliability estimates—testing structural assumptions.
        Educ Psychol Meas. 1974; 34: 25-33
        • Geldhof G.J.
        • Preacher K.J.
        • Zyphur M.J.
        Reliability estimation in a multilevel confirmatory factor analysis framework.
        Psychol Methods. 2014; 19: 72-91
        • Fornell C.
        • Larcker D.F.
        Evaluating structural equation models with unobservable variables and measurement error.
        J Mark Res. 1981; 18: 39-50
        • Thompson B.
        • Daniel L.G.
        Factor analytic evidence for the construct validity of scores: a historical overview and some guidelines.
        Educ Psychol Meas. 1996; 56: 197-208
        • Streiner D.L.
        Starting at the beginning: an introduction to coefficient alpha and internal consistency.
        J Pers Assess. 2003; 80: 99-103
        • Bagozzi R.P.
        • Yi Y.
        Specification, evaluation, and interpretation of structural equation models.
        J Acad Mark Sci. 2012; 40: 8-34
        • Hair J.F.
        • Anderson R.E.
        • Tatham R.L.
        • et al.
        Multivariate Data Analysis.
        Prentice-Hall, Englewood Cliffs, NJ1995
        • Costello A.
        • Osborne J.
        Best practices in exploratory factor analysis: four recommendations for getting the most from your analysis.
        Pract Assess Res Eval. 2005; 10: 1-9