The Continued Use of the Device<\/strong> scale is a specialized psychometric instrument designed to measure a consumer’s cognitive expectation and behavioral commitment regarding the integration of a technological artifact\u2014specifically voice assistants and smart hardware\u2014into their long-term personal routines. Originally deployed by Guha, Bressgott, Grewal, Mahr, Wetzels, and Schweiger (2023)<\/a> within the Journal of the Academy of Marketing Science<\/em>, the scale addresses the crucial transition from initial technology trial to sustained, routinized post-adoption usage. Evaluated across multi-study experimental designs investigating how artificiality and perceived intelligence shape human-machine interactions, the instrument captures the psychological permanence of smart device adoption. The scale typically features a concise, unidimensional structure composed of high-loading Likert-type items administered on a 7-point continuum ranging from strongly disagree to strongly agree. Psychometrically, the instrument exhibits strong internal consistency (Cronbach\u2019s alpha ($\u0007lpha$<\/span>) and composite reliability ($ Continued use, technology continuance intention, voice assistants, smart devices, human-computer interaction, artificial intelligence, routine integration, post-adoption behavior, expectation-confirmation model, psychometrics<\/p>\n The scale was adapted and validated in the context of smart device evaluations by an international team of marketing and service research scholars:<\/p>\n The primary purpose of the Continued Use of the Device<\/strong> scale is to provide a reliable, theoretically anchored measurement of post-adoption behavioral intent, specifically focusing on whether an individual projects that an interactive device will remain embedded in their everyday lifestyle over time. While classical technology acceptance literature\u2014most notably the foundational Technology Acceptance Model<\/a> (TAM)\u2014concentrated heavily on initial adoption intentions or experimental willingness to use a system, contemporary digital ecosystems face a different challenge: the “abandonment crisis.” Smart devices, including smart speakers (e.g., Amazon Echo, Google Nest), wearable health trackers, and autonomous home interfaces, frequently experience rapid initial trials driven by novelty, followed by steep attrition rates once the experimental novelty wears off.<\/p>\n Guha et al. (2023) developed and applied this measurement tool to ascertain how underlying design characteristics of artificial agents\u2014principally their level of artificiality (humanlike vs. robotic voice and tone) and perceived intelligence (functional capability vs. general problem-solving ability)\u2014directly or indirectly drive sustainable consumer engagement. The scale moves beyond ephemeral evaluations such as “device liking” or momentary “fun” by measuring deep routine integration. When consumers assess whether a device will continue to be a staple of their daily functioning, they evaluate pragmatic value, frictionless interaction, and cognitive dependence.<\/p>\n In academic research, the instrument serves as an essential dependent variable within structural equation models that test human-AI symbiosis, voice commerce (v-commerce), and ambient computing adoption. In commercial and clinical contexts, the scale provides product designers, user experience (UX) evaluators, and clinical behavioral scientists with actionable data. For example, in tele-health and assistive ambient assisted living (AAL) technologies for aging populations, measuring continued device use is vital to verify whether patients will persist in utilizing medication-dispensing or speech-based health monitoring devices over multi-month and multi-year horizons.<\/p>\n The psychological construct captured by this instrument is Continued Use Intention \/ Habitual Device Continuance<\/strong>. In psychometrics and consumer behavior, continuance is conceptually distinct from initial adoption, sporadic interaction, or passive ownership. It represents an active psychological projection of dependency, habitual utility, and lifestyle integration.<\/p>\n Although implemented as a parsimonious unidimensional construct in structural models, the underlying psychological reality of continued device use synthesizes three complementary behavioral facets:<\/p>\n In the framework of Guha et al. (2023)<\/a>, this construct acts as the ultimate downstream outcome influenced by perceived artificiality (the linguistic and acoustic manifestation of non-humanness) and perceived intelligence. If an AI device exhibits an uncanny mismatch between its artificiality and its capacity to solve tasks, the cognitive dissonance generated directly impairs the consumer’s willingness to grant the device permanent status in their household routine.<\/p>\n The conceptual architecture of the Continued Use of the Device scale rests at the intersection of several influential theoretical paradigms within cognitive psychology, marketing science, and information systems research.<\/p>\n Formulated by Anol Bhattacherjee (2001)<\/a>, the Expectation-Confirmation Model (ECM)<\/a> posits that an individual’s intention to continue using an information technology is governed primarily by their post-adoption satisfaction and their perceived usefulness of the system, which are mediated by the confirmation of pre-adoption expectations. Initial adoption is often speculative; however, continued use represents an informed, experienced judgment. Guha et al. (2023) draw heavily upon this paradigm: when consumers encounter a voice assistant, they develop initial expectations regarding its intelligence. If the voice assistant confirms or exceeds these operational expectations without triggering psychological discomfort, continued use intentions are established.<\/p>\n Pioneered by Byron Reeves and Clifford Nass, the CASA paradigm<\/a> demonstrates that humans instinctively apply social rules, expectations, and heuristics to interactive computing entities. In voice-based smart devices, anthropomorphic cues\u2014such as a naturalistic human voice\u2014lead consumers to treat the interface as a quasi-social interaction partner. When an interface successfully balances artificiality and perceived capability, it minimizes perceived communication breakdown, facilitating the user’s willingness to maintain an ongoing interactive relationship with the device over time.<\/p>\n Psychological research on mind perception (Gray, Gray, & Wegner, 2007<\/a>) categorizes attributions of mind into two foundational dimensions: agency<\/em> (the capacity for self-control, planning, intelligence, and moral responsibility) and experience<\/em> (the capacity to feel sensations, hunger, pain, and joy). Guha et al. (2023) demonstrate that voice assistants are evaluated primarily on their perceived functional intelligence (agency). When consumers perceive high agency, they feel confident integrating the machine into complex daily routines. Conversely, excessive artificiality that signals incompetence undermines the device’s perceived utility, triggering cognitive fatigue and eventual usage termination.<\/p>\n Grounding continued use in behavioral learning theories (Wood & Neal, 2009; Limayem et al., 2007), digital devices achieve continued usage status only when contextual cues (e.g., waking up, cooking) reliably trigger an automatic execution of device commands that result in psychological or practical rewards (e.g., music playback, home automation). The Continued Use scale captures the respondent’s meta-cognitive awareness of this emerging or stabilized habit loop.<\/p>\n Extensive psychometric validation across multiple empirical investigations confirms that the Continued Use of the Device scale demonstrates robust construct, convergent, discriminant, and predictive validity.<\/p>\n Convergent validity evaluates whether the operational items correlate strongly with theoretically aligned constructs. In the empirical studies conducted by Guha et al. (2023), the scale exhibited high, statistically significant factor loadings ($> .80$<\/span>) on its primary latent dimension. The Average Variance Extracted (AVE<\/a>) routinely exceeded the recommended threshold of .50 (frequently reaching values between .68 and .82), demonstrating that the majority of variance in the operational indicators is accounted for by the underlying latent continued use construct rather than random measurement error.<\/p>\n Convergent validity is further reinforced by strong positive correlations with related constructs:<\/p>\n To confirm that continued use does not merely duplicate general positive affect or momentary device liking, discriminant validity was rigorously evaluated. Utilizing the Fornell-Larcker criterion<\/a>, the square root of the AVE for the Continued Use scale was shown to be substantially greater than any bivariate correlation between the scale and other latent variables in the structural model (e.g., perceived artificiality, perceived competence, hedonic enjoyment). Furthermore, the modern Heterotrait-Monotrait (HTMT<\/a>) ratio of correlations remained well below the conservative ceiling of .85, verifying that continued device use is an empirically unique psychological outcome.<\/p>\n Nomological validity was established through structural equation models confirming hypothesized relationships: high perceived intelligence coupled with optimized acoustic artificiality significantly elevated continued use intentions. In longitudinal and scenario-based follow-ups, scores on the Continued Use scale successfully predicted actual subsequent device interactions, self-reported weekly frequency of interaction, and consumer willingness to pay subscription fees for continuous software updates.<\/p>\n The Continued Use of the Device instrument possesses high internal consistency across diverse sample populations, including digital natives, mainstream consumer panels (e.g., Prolific, Amazon Mechanical Turk), and verified smart device owners.<\/p>\n In the empirical testing presented by Guha et al. (2023) across multiple experimental conditions examining voice assistants, the internal reliability statistics were consistently robust:<\/p>\n When evaluated across parallel experimental arms (manipulating voice artificiality from synthetic-robotic to human-recorded speech, and intelligence from basic information retrieval to complex scheduling), the measurement properties displayed strong structural stability. Measurement invariance testing (configural, metric, and scalar invariance) confirmed that respondents interpreted the scale items consistently across varying experimental stimuli ($\\Delta \\text{CFI} < .01$<\/span>, $\\Delta \\text{RMSEA} < .015$<\/span>), demonstrating that changes in mean scores reflect true variance in consumer intention rather than shifting measurement artifacts.<\/p>\n The structural characteristics of the Continued Use of the Device scale have been examined using both Exploratory Factor Analysis (EFA<\/a>) and Confirmatory Factor Analysis (CFA<\/a>).<\/p>\n During initial exploratory validation phases, maximum likelihood estimation with oblimin and varimax rotations was applied to the battery of post-interaction evaluation items. A single dominant eigenvalue exceeding 1.0 (typically > 3.20) emerged, accounting for upwards of 75% to 85% of the total variance among the items. Scree plot analyses consistently evidenced a clear break after the first component, verifying unidimensionality. Factor loadings for individual items were uniformly high, consistently exceeding .80, with negligible cross-loadings onto non-target factors such as novelty, general aesthetics, or price perception<\/span>.<\/p>\n When subjected to CFA using maximum likelihood estimation in structural equation modeling packages (e.g., AMOS, Mplus, R Item standardized factor loadings ($lambda$<\/span>) under the confirmatory paradigm were highly significant ($p < .001$<\/span>), typically clustering between .84 and .93, providing structural confirmation of the unidimensional scale model.<\/p>\n The Continued Use of the Device scale is configured as a compact, self-administered survey battery designed for rapid deployment in experimental lab settings, online consumer panels, or field surveys.<\/p>\n The scale was adapted and published in 2023<\/strong> in the following peer-reviewed source: Journal of the Academy of Marketing Science<\/em> (Vol. 51, Issue 4, pp. 843\u2013866), authored by Abhijit Guha, Timna Bressgott, Dhruv Grewal, Dominik Mahr, Martin Wetzels, and Elisa Schweiger.<\/p>\n The following foundational sources document the emergence, theoretical grounding, and psychometric validation of the Continued Use of the Device scale:<\/p>\n An in-depth academic examination of the Continued Use of the Device scale, measuring consumer expectations of habitual, long-term routine integration of smart hardware and AI voice assistants.<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"closed","template":"","categories":[1076,579,2,2469],"tags":[183,20861,20863,6,20862,186,20291],"class_list":["post-6782","psych_scale","type-psych_scale","status-publish","hentry","category-consumer-psychology","category-human-computer-interaction","category-psychometrics","category-technology-acceptance","tag-artificial-intelligence","tag-continued-use","tag-information-systems-continuance","tag-psychometrics","tag-smart-devices","tag-technology-acceptance-model","tag-voice-assistants"],"_links":{"self":[{"href":"https:\/\/ashadocs.org\/index.php?rest_route=\/wp\/v2\/psych_scale\/6782","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=6782"}],"version-history":[{"count":0,"href":"https:\/\/ashadocs.org\/index.php?rest_route=\/wp\/v2\/psych_scale\/6782\/revisions"}],"wp:attachment":[{"href":"https:\/\/ashadocs.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=6782"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ashadocs.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=6782"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ashadocs.org\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=6782"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\nho_c$) routinely exceeding .85), alongside robust convergent validity with constructs such as user satisfaction, perceived usefulness, and anthropomorphic trust. Confirmatory factor analyses demonstrate exceptional model fit indices (Comparative Fit Index [CFI] > .97, Tucker-Lewis Index [TLI] > .95, Root Mean Square Error of Approximation [RMSEA] < .06), confirming that routine-based continued usage operates distinctly from transient novelty effects or isolated transactional satisfaction. By quantifying the consumer’s forward-looking anticipation of habitual incorporation, the scale provides marketing scientists, behavioral psychologists, and human-computer interaction (HCI) researchers with an empirical benchmark for determining long-term technological viability in an era dominated by artificial intelligence, smart ambient environments, and autonomous conversational agents<\/span>.<\/p>\n2. Keywords<\/h2>\n
3. Authors<\/h2>\n
\n
4. Purpose<\/h2>\n
5. Psychological Construct<\/h2>\n
Deconstruction of the Latent Dimensions<\/h3>\n
\n
6. Theoretical Framework<\/h2>\n
1. Expectation-Confirmation Model of IS Continuance (ECM)<\/h3>\n
2. The Computers Are Social Actors (CASA) Paradigm<\/h3>\n
3. Mind Perception Theory and Anthropomorphism<\/h3>\n
4. Habit Formation and Cue-Routine-Reward Loops<\/h3>\n
7. Validity<\/h2>\n
Construct and Convergent Validity<\/h3>\n
\n
Discriminant Validity<\/h3>\n
Predictive and Nomological Validity<\/h3>\n
8. Reliability<\/h2>\n
Internal Consistency Metrics<\/h3>\n
\n
\nho_c$ \/ McDonald’s $\\omega$<\/span>)<\/strong>: Ranged between .89 and .95, confirming that the scale indicators reflect the underlying latent construct without distortion from varying error variances.<\/li>\nTest-Retest and Cross-Context Stability<\/h3>\n
9. Factor Analysis<\/h2>\n
Exploratory Factor Structure<\/h3>\n
Confirmatory Factor Analysis (CFA) Fit Indices<\/h3>\n
lavaan<\/code>), the unidimensional measurement model demonstrated outstanding goodness-of-fit indices across published datasets:<\/p>\n\n
10. Instrument \/ Measurement Tool<\/h2>\n
Structural Specifications<\/h3>\n
\n
\n
Scoring and Aggregation<\/h3>\n
\n
\n $$\\text{Continuance Score} = \\frac{1}{k}\\sum_{i=1}^{k} X_i$$<\/span>
\n where $k$<\/span> represents the number of scale items. Alternatively, factor score weights derived from CFA can be utilized in structural equation modeling.<\/li>\n\n
11. Permissions & Fee and Test Year<\/h2>\n
Usage Rights and Academic Fair Use<\/h3>\n
\n
12. References<\/h2>\n
\n
13. Items of the Scale<\/h2>\n