Expectation-Confirmation Model (ECM) - Bhattacherjee (2001)

Theory Publication Information

Citation Information

Why Was the Model Created?

Bhattacherjee developed the Expectation-Confirmation Model to address a critical gap in information systems research that had focused almost exclusively on adoption decisions while neglecting post-adoption continuance behavior. The author recognized that technology adoption and technology continuance are conceptually distinct phenomena requiring different theoretical frameworks. Research had established that many users adopt new information systems but discontinue or minimize usage shortly after implementation, suggesting that adoption factors do not automatically predict sustained use. This discontinuance problem threatens the return on investment from expensive technology implementations and undermines organizational efforts to achieve technology value realization.

The original Technology Acceptance Model and related adoption theories predict whether users will initially accept and use new technologies, but they provide insufficient theoretical explanation of what determines whether users will continue using technologies after initial adoption. Bhattacherjee recognized that post-adoption continuance involves different psychological mechanisms than initial adoption decisions. The author proposed adapting expectation-confirmation theory from consumer behavior research, where confirmation of expectations about product performance and resulting satisfaction have long been established as determinants of consumer repurchase and loyalty decisions.

To establish the ECM’s validity in information systems contexts, Bhattacherjee conducted an empirical study of online banking users, a population representing post-adoption users who had chosen to use internet banking systems and were making ongoing decisions about continued usage. The study proposed that IS continuance intention is determined by user satisfaction with prior IS use and perceived usefulness, where satisfaction is determined by confirmation of expectations versus actual performance experience. This model explicitly separates adoption antecedents from continuance antecedents, establishing the conceptual distinction between initial adoption and sustained usage decisions.

Core Concepts and Definitions

The Expectation-Confirmation Model draws on six conceptual constructs from consumer behavior and expectation-confirmation theory. Bhattacherjee’s research model (Figure 2, p. 356) measures four of these directly (Perceived Usefulness, Confirmation, Satisfaction, Continuance Intention); the other two (Expectations and Perceived Performance) are subsumed into the Confirmation construct, which captures users’ cognitive comparison between the two:

• Expectations: Pre-use beliefs about how well an information system will perform and what benefits it will deliver. Expectations are formed before or during initial adoption based on marketing claims, peer recommendations, or prior experience with similar systems.
• Perceived Performance: Post-use beliefs about how well the information system actually performed in delivering expected benefits and meeting user needs. Perceived performance reflects actual user experience with the system compared to pre-use understanding.
• Confirmation: The degree to which perceived performance matches or exceeds prior expectations. Confirmation is a cognitive evaluation of the gap between expected and actual performance, where positive confirmation occurs when actual performance meets or exceeds expectations, and negative confirmation occurs when performance falls short.
• Satisfaction:A user’s emotional and evaluative response to prior information system use, determined by the degree of confirmation. Satisfaction reflects emotional reactions to whether the system lived up to expectations, encompassing both affective and evaluative components.
• Perceived Usefulness (Post-Adoption): The degree to which a user believes the information system will enhance job performance or create value. Post-adoption usefulness reflects updated beliefs based on actual usage experience, distinguishing post-adoption usefulness from pre-adoption expectations.
• Continuance Intention:A user’s intention to continue using an information system in the future. Continuance intention is the key outcome variable, reflecting decisions about sustained usage rather than initial adoption.

Preceding Models or Theories

The Expectation-Confirmation Model built upon established theories from consumer behavior and information systems:

• Expectation-Confirmation Theory (Oliver, 1980): Original consumer behavior theory establishing that product satisfaction is determined by confirmation of expectations, defined as the difference between expected and perceived performance. ECM applies this foundational theory to information systems contexts.
• Technology Acceptance Model (Davis, 1989): Established perceived usefulness as primary determinant of technology adoption. ECM extends TAM by incorporating satisfaction and confirmation as post-adoption continuance mechanisms.
• Expectancy Disconfirmation Paradigm (Oliver, 1977): Proposed that satisfaction results from perceived performance relative to expectations. ECM operationalizes disconfirmation paradigm specifically for post-adoption IS usage contexts.
• Consumer Behavior and Loyalty Research (Cronin & Taylor, 1992): Demonstrated that satisfaction is primary determinant of service repurchase and loyalty, providing behavioral intention frameworks applicable to IS continuance.
• IS Adoption Research: Accumulated evidence showing adoption and continuance are distinct phenomena with different drivers, motivating development of post-adoption specific theoretical frameworks.

Describe The Model

The Expectation-Confirmation Model proposes that post-adoption continuance intention is determined by two primary factors: user satisfaction with prior IS use and perceived usefulness of the system. User satisfaction is determined by confirmation of prior expectations through actual use experience. The model incorporates a temporal dynamic: users form expectations before or during initial adoption, then accumulate actual usage experience that either confirms or disconfirms those expectations. The degree of confirmation shapes both perceived usefulness and user satisfaction, while perceived usefulness and satisfaction together influence continuance intention. Confirmed expectations result in high satisfaction and sustained beliefs in usefulness, encouraging continued usage. Disconfirmed expectations result in low satisfaction and erosion of usefulness beliefs, increasing discontinuance likelihood.

ECM Causal Mechanisms

• Expectations-to-Confirmation (theoretical): At the ECT theoretical level, confirmation is the cognitive result of comparing pre-use expectations to perceived performance. In the ECM research model (Figure 2, p. 356), confirmation is measured directly as a single construct rather than decomposed into separate expectations and performance measures.
• H1. Satisfaction -> Continuance Intention:Satisfaction is the strongest predictor of continuance intention (standardized path coefficient 0.567, p < .001 per Figure 3, p. 363). Reported as beta = 0.57 in prose (p. 364), accounting for 32 percent of continuance intention variance.
• H2. Confirmation -> Satisfaction:Confirmation is the dominant antecedent of satisfaction (standardized path coefficient 0.451, p < .001 per Figure 3; beta = 0.53 in prose, p. 364), accounting for 28 percent of satisfaction variance.
• H3. Perceived Usefulness -> Satisfaction:Post-adoption perceived usefulness positively influences satisfaction (standardized path coefficient 0.227, p < .01 per Figure 3; beta = 0.23 in prose, p. 364), accounting for 5 percent of satisfaction variance. In ECM, perceived usefulness influences satisfaction, not the reverse.
• H4. Perceived Usefulness -> Continuance Intention:Post-adoption perceived usefulness directly predicts continuance intention (standardized path coefficient 0.294, p < .001 per Figure 3; beta = 0.29 in prose, p. 364), accounting for 9 percent of continuance intention variance.
• H5. Confirmation -> Perceived Usefulness:Confirmation shapes post-adoption perceived usefulness (standardized path coefficient 0.525, p < .001 per Figure 3; reported as beta = 0.45 in the prose on p. 364), accounting for 20 percent of perceived usefulness variance.
• Temporal Dynamics: Model captures time progression from pre-adoption expectations through post-adoption confirmation and satisfaction to continuance decisions, reflecting real user experience trajectories.

Main Strengths

• Addresses post-adoption research gap: First major information systems model explicitly focusing on continuance rather than adoption, establishing continuance as distinct research stream.
• Solid theoretical foundation: Grounded in well-established expectation-confirmation theory from consumer behavior, which has been widely applied to repurchase and loyalty decisions.
• Explained continuance variance:Original study explained 41 percent of continuance intention variance (R² = 0.41), 33 percent of satisfaction variance (R² = 0.33), and 20 percent of perceived usefulness variance (R² = 0.20), demonstrating substantial predictive power for post-adoption behaviors.
• Temporal validity: Model captures dynamic user experience progression from expectations through performance evaluation to continuance decisions, reflecting actual user decision processes.
• Practical relevance: Identifies actionable mechanisms for improving user continuance through expectation management and performance delivery aligned with user expectations.
• Parsimony with explanatory power: Simple, elegant model with few constructs and relationships explains complex post-adoption continuance phenomena effectively.
• Empirical validation: Tested in real online banking context with 122 users engaged in post-adoption usage decisions, establishing behavioral relevance.

Main Weaknesses

• Limited sampling diversity: Original validation used online banking users only. Technology types vary substantially in characteristics (social systems, productivity tools, infrastructure), potentially moderating model relationships.
• Cross-cultural generalizability unclear: Developed with U.S. sample; expectation formation, confirmation interpretation, and satisfaction drivers may differ in cultures with different communication norms and service expectations.
• Expectations measurement challenges: Operationalizing pre-use expectations and measuring confirmation retrospectively introduces recall bias and interpretation challenges. Users may reframe initial expectations to match experienced outcomes.
• Single time-point post-adoption measurement: While model is temporal in theory, original study measured confirmation and satisfaction at single post-adoption occasion. Multiple measurement points would strengthen evidence for temporal dynamics.
• Discontinuance vs. reduced usage: Model measures continuance intention but does not distinguish between complete discontinuance and reduced-frequency usage, which may have different psychological drivers.
• Organizational mandates not incorporated: Model focuses on voluntary continued usage decisions. Organizational mandates requiring continued system use despite low satisfaction create boundary conditions for model applicability.
• Switching cost and habit effects: Model does not incorporate switching costs or habitual usage patterns that may maintain continued usage despite low confirmation or satisfaction.
• Limited investigation of expectation sources: Model does not examine how initial expectations form or vary by source (marketing claims, peer recommendations, prior experience), which may moderate confirmation effects.

Key Contributions

• Continuance as distinct research domain: Established that post-adoption continuance requires different theoretical frameworks than initial adoption, creating entire new research stream examining sustained technology usage rather than adoption decisions.
• Expectation-confirmation application to IS: Successfully adapted well-established consumer behavior theory to information systems context, demonstrating that consumer satisfaction theory applies to post-adoption technology usage.
• Satisfaction as continuance mechanism: Identified user satisfaction as primary determinant of post-adoption continuance, providing more nuanced understanding of post-adoption behavior than adoption models.
• Temporal dynamics integration: Explicitly modeled time progression from expectations through confirmation and satisfaction to continuance decisions, capturing realistic user experience trajectories over time.
• Expectations-performance gap operationalization: Introduced confirmation as specific mechanism through which expectation-performance alignment influences post-adoption continuance, operationalizing abstract disconfirmation concept.
• Distinction from adoption factors: Demonstrated that continuance drivers (satisfaction, confirmation) differ from adoption drivers (ease of use, usefulness), challenging assumptions that single model applies across adoption-continuance lifecycle.
• Acceptance-discontinuance anomaly:Bhattacherjee frames ECM as an explanation for the “acceptance-discontinuance anomaly” (abstract and p. 352), the empirical pattern in which users accept a system at adoption but later discontinue. TAM cannot explain this, because the same pre-acceptance variables should predict both behaviors; ECM explains it through disconfirmation and dissatisfaction occurring after use.
• Empirical validation of continuance model: Provided empirical evidence that expectation-confirmation model predicts post-adoption continuance intentions, establishing behavioral validity for proposed mechanisms.

Internal Validity

The Expectation-Confirmation Model employed sound methodology to establish internal validity:

• Validated theoretical basis: Grounded in expectation-confirmation theory with extensive prior empirical validation in consumer behavior contexts, providing strong theoretical foundation.
• Appropriate sample selection: Studied online banking users, a population representing actual post-adoption users making real continuance decisions, ensuring ecological validity.
• Sample size and sampling frame: 1,000 online banking customers were sampled from a mid-sized U.S. bank customer base; 122 usable responses (approximately 12 percent response rate) were analyzed, providing reasonable statistical power for the four-construct covariance-based structural equation model.
• Measurement validation: Reported composite reliabilities (Table 3, p. 362) of 0.83 (Continuance Intention), 0.87 (Satisfaction), 0.88 (Perceived Usefulness), and 0.82 (Confirmation), all above the 0.70 threshold.
• Convergent and discriminant validity:Average variance extracted (AVE) values ranged from 0.60 to 0.65 (Table 3, p. 362), all above the 0.50 threshold. Chi-square difference tests (Table 4, p. 363) between constrained and unconstrained measurement models were significant (p < .001) for all construct pairs, supporting discriminant validity.
• Structural equation modeling with EQS: Confirmatory factor analysis and structural model estimated with the EQS program (Bentler, 1989) using maximum likelihood estimation on the covariance matrix, testing the hypothesized causal relationships directly.
• Model fit assessment:Structural model fit (Figure 3, p. 363): chi-square = 116.76, df = 68, p < .001; chi-square/df = 1.717; NFI = 0.883, NNFI = 0.928, CFI = 0.946. All five hypothesized paths were significant (p < .01). The model explained 41 percent of continuance intention variance, 33 percent of satisfaction variance, and 20 percent of perceived usefulness variance.

External Validity

External validity considerations require careful interpretation of generalizability:

• Technology context limitation: Original validation focused specifically on online banking, a relatively mature technology with clear performance outcomes. Generalization to emerging technologies, complex enterprise systems, or systems with ambiguous performance requires investigation.
• Single technology study: While expectation-confirmation theory generalizes across consumer products, information systems vary dramatically in complexity, interface design, and performance tangibility. Different IS types may show different confirmation-satisfaction-continuance relationships.
• Voluntary usage context: Online banking represents primarily voluntary technology use. Applicability to mandatory enterprise system implementations where organizational requirements override personal continuance decisions requires clarification.
• U.S. sample limitations: Single U.S. study limits generalization to international contexts where expectation formation, service quality expectations, and satisfaction drivers vary culturally.
• Cross-domain generalization: Model validated in consumer financial services context. Generalization to business-to-business systems, social media platforms, or professional tools requires empirical confirmation.
• Temporal boundary: Study conducted in early 2000s when online banking was novel. Contemporary technology landscape with ubiquitous digital services and high user expectations may show different confirmation and satisfaction dynamics.
• Experience level variation: Model not tested across user experience levels. Confirmation and satisfaction effects may differ between novice users forming initial expectations and experienced users with established system familiarity.

Relevance to Technology Adoption

The Expectation-Confirmation Model directly addresses a critical gap in understanding technology adoption barriers by explaining why initial adoption often fails to translate into sustained technology use. Organizations frequently encounter situations where technology implementations appear successful in adoption metrics yet show declining usage rates within months. ECM reveals that post-adoption continuance involves different psychological mechanisms than initial adoption, requiring distinct management strategies. By identifying confirmation of expectations as a continuance driver, the model reveals that technology discontinuance often stems not from technology deficiencies but from unmet expectations created during adoption phases.

Barriers to Sustained Technology Use Identified

• Expectation-performance gaps: When actual system performance falls short of pre-adoption expectations, users experience negative confirmation, reducing satisfaction and continuance likelihood.
• Oversold adoption claims: Marketing or training communications promising performance improvements that systems fail to deliver create negative confirmation and user dissatisfaction.
• Low perceived post-adoption usefulness: Even if expectations were moderate, if users do not perceive ongoing value from continued use, continuance intention remains low.
• Hidden implementation costs: When actual usage time requirements, learning demands, or integration challenges exceed expectations, users experience dissatisfaction and discontinuance intentions.
• System reliability issues: Performance problems, downtime, or data accuracy issues create negative confirmation, undermining belief that system delivers expected benefits.
• Inadequate user support post-adoption: When implementation support diminishes after initial adoption, users struggle to realize expected benefits, reducing satisfaction.
• Misaligned expectations across user populations: Different user groups with divergent expectations experience different confirmation patterns, with some discontinuing despite successful implementation for others.

Leadership Actions the Model Prescribes

• Manage expectations carefully during adoption: Provide realistic, evidence-based information about what systems will and will not accomplish rather than overselling promised benefits.
• Set achievable performance targets: Define and communicate specific, measurable benefits users can expect. Deliver results that meet or exceed these defined targets.
• Measure actual performance systematically: Conduct post-implementation performance assessments comparing actual to expected outcomes. Share results demonstrating that systems deliver expected benefits.
• Maintain post-adoption support: Continue training, support, and optimization efforts beyond initial implementation. This enables users to achieve full expected benefits, increasing confirmation and satisfaction.
• Create feedback loops: Regularly solicit user feedback on confirmation of expectations, identifying gaps early before users make discontinuance decisions.
• Address system reliability and performance: Ensure systems perform reliably and deliver promised functionality. Technical problems create powerful negative confirmation effects.
• Segment user populations by expectations: Different users may have different expectations. Tailor support and messaging to ensure each segment achieves relevant confirmation.
• Distinguish adoption success from continuance success: Recognize that initial adoption metrics do not guarantee sustained usage. Implement post-adoption measurement and management programs addressing continuance specifically.

Following Models or Theories

The Expectation-Confirmation Model established an influential post-adoption research stream:

• ECM extensions and modifications: Numerous researchers have extended ECM by incorporating additional post-adoption factors including habit formation, switching costs, social influences, and system updates.
• IS continuance research stream: ECM spawned entire research domain examining post-adoption phenomena including extended ECM models, habit-based continuance, and discontinuance drivers.
• Mobile and consumer technology continuance: Researchers applied ECM to smartphone applications, social media, cloud services, and consumer technology to understand what drives continued adoption.
• Online service continuance: ECM became foundational model for understanding continuance of online services including e-commerce, streaming, social networks, and digital platforms.
• Enterprise system continuance: Organizational researchers adapted ECM to examine post-implementation continuance in mandatory organizational technology contexts including ERP systems.
• Habit and automaticity in continuance: Later research integrated ECM with habit formation theory, demonstrating that habit becomes stronger continuance driver than satisfaction as experience accumulates.

References

1. Bhattacherjee, A. (2001). Understanding information systems continuance: An expectation-confirmation model. MIS Quarterly, 25(3), 351-370. https://doi.org/10.2307/3250921
2. Oliver, R. L. (1980). A cognitive model of the antecedents and consequences of satisfaction decisions. Journal of Marketing Research, 17(4), 460-469.↩ https://doi.org/10.1177/002224378001700405
3. Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13(3), 319-340.↩ https://doi.org/10.2307/249008
4. Oliver, R. L. (1977). Effect of expectation and disconfirmation on post-exposure product evaluations: An alternative interpretation. Journal of Applied Psychology, 62(4), 480-486.↩
5. Cronin, J. J., & Taylor, S. A. (1992). Measuring service quality: A reexamination and extension. Journal of Marketing, 56(3), 55-68.↩

Further Reading

1. Venkatesh, V., & Davis, F. D. (2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science, 46(2), 186-204.
2. Bhattacherjee, A., & Premkumar, G. (2004). Understanding changes in belief and attitude toward information technology usage: A study of switch users.MIS Quarterly, 28(4), 625-641.
3. Venkatesh, V., Morris, M. G., Davis, G. B., & Davis, F. D. (2003). User acceptance of information technology: Toward a unified view. MIS Quarterly, 27(3), 425-478. https://doi.org/10.2307/30036540
4. DeLone, W. H., & McLean, E. R. (2003). The DeLone and McLean model of information systems success: A ten-year update. Journal of Management Information Systems, 19(4), 9-30. https://doi.org/10.1080/07421222.2003.11045748
5. Limayem, M., Hirt, S. G., & Cheung, C. M. (2007). How habit limits the prediction of usage: The case of MS Word. ACM SIGMIS Database: the DATABASE for Advances in Information Systems, 38(4), 29-46. https://doi.org/10.2307/25148817

Series Navigation