The AI Healthcare Preference Scale<\/strong> (Kerstan, Bienefeld, & Grote, 2024) is a psychometric instrument designed to evaluate individuals’ behavioral and evaluative preferences regarding the allocation of clinical tasks between artificial intelligence (AI)<\/a> systems and human physicians. Developed in response to the accelerated integration of algorithmic diagnostic and therapeutic tools into contemporary medicine, the scale quantifies the degree to which health consumers prefer human-led, collaborative, or autonomous machine-led care. Grounded theoretically in the classic levels of automation<\/em> (LOA) taxonomies developed by Endsley and Kaber (1999)<\/a>, the instrument comprises six items administered via an electronic questionnaire. Each item is scored on a 5-point bipolar continuum ranging from 1 (only physician<\/em>) to 5 (only AI<\/em>), with intermediate anchors denoting variable degrees of joint human-machine agency.<\/p>\n Psychometric evaluation among adult participants demonstrated robust internal consistency, yielding a Cronbach’s alpha of .75. Exploratory factor analysis (EFA) demonstrated an essentially unidimensional structural organization, with all items loading decisively onto a single underlying factor of AI preference. Subsequent confirmatory factor analysis (CFA) incorporated within a broader nomological network confirmed the scale’s distinctiveness, achieving acceptable goodness-of-fit indices (χ² = 1008.46, df = 547, p < .001, χ²\/df = 1.84, CFI = 0.91, TLI = 0.90, RMSEA = 0.04, SRMR = 0.06). The AI Healthcare Preference Scale serves as a standardized empirical diagnostic tool for researchers, health policy analysts, and organizational psychologists investigating patient acceptance, algorithm aversion, and the transition toward hybrid clinical decision architectures<\/span>.<\/p>\n Artificial Intelligence, Healthcare Preferences, Levels of Automation, Human-AI Interaction, Algorithm Aversion, Medical Decision-Making, Patient Acceptance, Psychometrics, Clinical Trust, Digital Health<\/p>\n The AI Healthcare Preference Scale was developed by researchers at the ETH Zurich<\/strong> (Swiss Federal Institute of Technology), based in Zurich, Switzerland, within the Department of Management, Technology, and Economics (D-MTEC), specifically the Chair of Work and Organizational Psychology:<\/p>\n Correspondence Address:<\/strong> Sophie Kerstan, ETH Zurich, Department of Management, Technology, and Economics, Work and Organizational Psychology, Weinbergstrasse 56\/58, 8092 Zurich, Switzerland. Email: skerstan@ethz.ch.<\/p>\n The deployment of machine learning algorithms, computer vision, and large language models across medical specialties has precipitated a paradigm shift in healthcare delivery. Despite demonstrable algorithmic parity or superiority in diagnostic imaging, oncological triage, and risk stratification, implementation frequently falters due to socio-psychological resistance, colloquially termed algorithm aversion<\/em>. The primary objective of the AI Healthcare Preference Scale<\/strong> is to provide an empirical assessment of user inclinations regarding who should execute critical healthcare tasks: a human physician, an autonomous algorithm, or an integrated hybrid collaborative team.<\/p>\n Historically, research evaluating technological readiness in medicine relied on generic instruments, such as the Technology Acceptance Model<\/a> (TAM) or general trust scales. However, these tools frequently obscure the nuanced reality of contemporary clinical workflows, where automation is not a binary present-or-absent condition but exists across a dynamic spectrum of shared responsibility. Kerstan, Bienefeld, and Grote (2024) developed this instrument to quantify the comparative agency that end-users are willing to surrender to automated systems across distinct stages of the clinical pathway.<\/p>\n From an applied and translational perspective, the scale offers multiple strategic applications:<\/p>\n The focal psychological construct captured by the instrument is Artificial Intelligence Healthcare Preference<\/strong>. Rather than capturing a general personality trait, this construct represents a domain-specific, evaluative attitudinal disposition that reflects an individual’s preferred allocation of cognitive, diagnostic, and prescriptive authority between a human medical professional and an automated computational system.<\/p>\n This construct is characterized by three core psychometric facets embedded within its unidimensional operationalization:<\/p>\n At the low end of the spectrum (scale anchors 1–2), the construct reflects pronounced human physician preference, rooted in interpersonal trust, beliefs in human empathy, and fear of algorithmic inflexibility. At the midpoint (anchor 3), the construct reflects an egalitarian integration of both human and artificial competencies. At the high end (scale anchors 4–5), it reflects technocentric optimization, wherein the respondent prioritizes algorithmic consistency, speed, and evidence-based precision over human practitioner involvement.<\/p>\n The AI Healthcare Preference Scale is conceptually anchored in the classic engineering psychology framework of Levels of Automation (LOA)<\/strong> formulated by Endsley and Kaber (1999)<\/a>, alongside subsequent elaborations by Parasuraman, Sheridan, and Wickens (2000). The LOA paradigm posits that automation cannot be treated as a monolithic, all-or-nothing phenomenon; rather, task execution within complex socio-technical systems is distributed along a continuum ranging from purely manual human execution to complete computer autonomy.<\/p>\n According to Endsley and Kaber (1999), system operations can be broken down into four structural stages: (1) information acquisition, (2) information analysis, (3) decision selection, and (4) action implementation. Within each stage, varying levels of agency can be assigned to either the human or the machine. Kerstan et al. (2024) translated these engineering principles into the healthcare delivery context by establishing a five-point behavioral response architecture:<\/p>\n By mapping scale responses to this structured hierarchy, the scale bridges human-computer interaction (HCI) theory with social-cognitive theories of trust, risk perception, and decision science.<\/p>\n Empirical validation of the AI Healthcare Preference Scale was executed by Kerstan, Bienefeld, and Grote (2024) across adult samples recruited via the Prolific research platform. The authors assessed multiple aspects of validity:<\/p>\n The AI Healthcare Preference Scale demonstrates robust internal consistency. Across its initial validation sample of adult participants in the United States, Kerstan et al. (2024) observed a Cronbach’s alpha of .75<\/strong>. For a brief, six-item instrument measuring an evolving attitudinal construct, a coefficient of .75 indicates sound internal consistency, comfortably exceeding the standard .70 threshold recommended for academic research instruments (Nunnally & Bernstein, 1994).<\/p>\n Item-total correlations across the six items remained uniform, demonstrating that each item reliably taps into the core latent dimension without excessive item redundancy. The modest length of the scale balances measurement precision with minimal participant burden, mitigating survey fatigue in extended online experimental protocols.<\/p>\n The structural dimensionality of the scale was systematically explored using both exploratory and confirmatory analytic frameworks:<\/p>\n All six items contributed significantly to their designated latent variable (p < .001), corroborating the structural validity and dimensional parsimony of the measure<\/span>.<\/p>\n The technical parameters and administration profile of the scale are structured as follows:<\/p>\n The AI Healthcare Preference Scale was published in 2024<\/strong>. The instrument was developed under the auspices of academic research conducted at ETH Zurich and published in the peer-reviewed journal Risk Analysis<\/em> (Society for Risk Analysis \/ John Wiley & Sons, Inc.).<\/p>\n Researchers intending to use the scale for non-commercial academic research, pedagogical purposes, or public policy analysis should refer to the original publication (https:\/\/doi.org\/10.1111\/risa.14216<\/a>) or contact the corresponding author, Sophie Kerstan (skerstan@ethz.ch), for formal permission and to verify any applicable institutional repository conditions. Commercial applications, clinical software integrations, or proprietary deployments require explicit licensing permission from the copyright holders.<\/p>\n The AI Healthcare Preference Scale (Kerstan, Bienefeld, & Grote, 2024) is a 6-item psychometric measure evaluating patient and user preferences for human physician care versus artificial intelligence automation based on Endsley & Kaber’s levels of automation framework.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","template":"","categories":[109,579,2],"tags":[20398,183,4456,7831,24919,24918,6,2474],"class_list":["post-8338","psych_scale","type-psych_scale","status-publish","hentry","category-health-psychology","category-human-computer-interaction","category-psychometrics","tag-algorithm-aversion","tag-artificial-intelligence","tag-digital-health","tag-healthcare","tag-levels-of-automation","tag-patient-preferences","tag-psychometrics","tag-trust-in-automation"],"_links":{"self":[{"href":"https:\/\/ashadocs.org\/index.php?rest_route=\/wp\/v2\/psych_scale\/8338","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ashadocs.org\/index.php?rest_route=\/wp\/v2\/psych_scale"}],"about":[{"href":"https:\/\/ashadocs.org\/index.php?rest_route=\/wp\/v2\/types\/psych_scale"}],"author":[{"embeddable":true,"href":"https:\/\/ashadocs.org\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ashadocs.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=8338"}],"version-history":[{"count":0,"href":"https:\/\/ashadocs.org\/index.php?rest_route=\/wp\/v2\/psych_scale\/8338\/revisions"}],"wp:attachment":[{"href":"https:\/\/ashadocs.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=8338"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ashadocs.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=8338"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ashadocs.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=8338"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}Keywords<\/h2>\n
Authors<\/h2>\n
\n
Purpose<\/h2>\n
\n
Psychological Construct<\/h2>\n
\n
Theoretical Framework<\/h2>\n
\n
Validity<\/h2>\n
\n
Reliability<\/h2>\n
Factor Analysis<\/h2>\n
\n
\n
Instrument \/ Measurement Tool<\/h2>\n
\n
\n
Permissions & Fee and Test Year<\/h2>\n
References<\/h2>\n
\n
Items of the Scale<\/h2>\n