Technology Acceptance Model (TAM) – Davis (1989)

Why was the model made?

Fred Davis developed the Technology Acceptance Model to address a critical gap in information systems research and practice. The fundamental problem motivating TAM was straightforward but significant: Information Systems practitioners and organizations needed predictive tools for identifying which technologies would achieve user acceptance and which would face resistance. Yet existing information systems research lacked adequate theoretical frameworks for predicting and explaining technology acceptance in organizational contexts. The motivation for TAM emerged from recognizing that costly technology implementations frequently failed due to poor user acceptance despite technical adequacy. Organizations invested heavily in developing or acquiring information systems that proved technically sound and functionally capable, yet failed due to user rejection, low adoption, or underutilization. Understanding why users accepted some technologies while rejecting others became critical practical concern for IS organizations.

Prior information systems research had identified numerous factors potentially influencing technology acceptance including user characteristics, system characteristics, organizational factors, and environmental factors. However, this pluralistic factor approach lacked theoretical integration and predictive power. Different studies emphasized different factors, producing inconsistent understanding of technology acceptance. The field needed a more parsimonious, theoretically grounded model identifying the most critical factors determining user acceptance while explaining why these factors mattered. Davis explicitly grounded TAM in the Theory of Reasoned Action (TRA), recognizing that technology acceptance represents a behavioral attitude problem. Users develop attitudes toward technologies based on their beliefs about those technologies, and these attitudes influence user intentions to adopt technologies, which in turn influence actual usage. However, Davis recognized that general behavioral models like TRA required technology- specific operationalization.

Generic attitude measures and outcome beliefs might not capture the specific beliefs and attitudes shaping technology acceptance. Davis proposed that two beliefs held particular importance for technology acceptance: perceived usefulness and perceived ease of use. These beliefs are technology-specific (they vary based on particular technology characteristics and user perceptions of those characteristics), readily measurable, and theoretically derived from general behavioral attitude models. By identifying these specific belief categories as particularly influential for technology acceptance, TAM provided a parsimonious, focused model explaining technology acceptance while maintaining grounding in established behavioral theory.

How was the model’s internal validity tested?

Davis established the Technology Acceptance Model’s internal validity through multiple validation approaches: Theoretical derivation from established theory: TAM built directly on the Theory of Reasoned Action, maintaining TRA’s core structure (attitudes determining intentions determining behavior) while specializing it to technology acceptance. By deriving TAM from TRA, Davis imported the theoretical validity of the broader behavioral framework. However, Davis specialized TRA by identifying perceived usefulness and perceived ease of use as the specific beliefs most important for technology acceptance, requiring empirical validation that these specific constructs adequately explain technology acceptance attitudes.

• Operationalization of key constructs: Davis provided explicit measurement scales for perceived usefulness, perceived ease of use, attitudes toward technology use, behavioral intention to use, and actual system usage. The perceived usefulness scale assessed users’ beliefs that using a system would enhance work performance. The perceived ease of use scale assessed users’ beliefs that using a system would be effortless. Attitude items assessed overall favorable/unfavorable evaluations of technology use. Behavioral intention items assessed plans to use technology. Usage was measured through actual system usage logs when available. This explicit operationalization enabled precise measurement and replication
• Validation with multiple information systems: Davis validated TAM using two different information systems: email and file managers within a single organization. By demonstrating that perceived usefulness and perceived ease of use predicted user acceptance across different systems, Davis provided evidence that TAM captured fundamental aspects of technology acceptance rather than system-specific phenomena. The consistency of findings across different systems suggested robust internal validity
• Structural equation modeling: Davis employed structural equation modeling to test whether the hypothesized TAM causal structure fit data adequately. SEM enabled testing whether the proposed relationships (perceived usefulness and ease of use influencing attitudes, attitudes influencing intentions, intentions influencing usage) fit the actual data patterns. Good model fit across systems suggested the theoretical structure captured actual relationships in technology acceptance
• Measurement validity assessment: Davis assessed both reliability and validity of measures. Internal consistency analysis evaluated whether multi- item measures for each construct consistently measured single underlying constructs. Convergent validity analysis examined whether different measures of the same construct correlated strongly. Discriminant validity analysis examined whether measures of different constructs remained appropriately distinct. These validity assessments provided evidence that measures validly operationalized the theoretical constructs

How was the model’s external validity tested?

The Technology Acceptance Model’s external validity was established through application across diverse technological systems and contexts: Multiple information systems: Davis validated TAM across email systems and file managers, demonstrating applicability across different information technologies. Email represented communication and productivity tools, while file managers represented basic system utilities. Showing consistent relationships across different technological domains suggested the model’s generalizability beyond any single system type.

• Different user populations: Davis examined TAM with different organizational populations—demonstrating that the fundamental relationships held across different user groups, suggesting broad applicability rather than population-specific effects
• Organizational context: Validation in actual organizational settings (rather than laboratory contexts) using real systems and authentic work tasks provided evidence for external validity in ecologically valid contexts. Results from field studies in actual organizations carry greater confidence regarding real-world applicability than laboratory studies
• Usage measurement: Davis demonstrated relationships between measured intentions and actual system usage, providing evidence that attitudinal measures successfully predicted meaningful behavioral outcomes. Some attitude studies show weak relationships between attitudes and behavior; Davis’s demonstration that perceived usefulness, ease of use, and attitudes significantly predicted actual usage provided external validity evidence for behavioral prediction
• Longitudinal validation: Davis validated TAM using longitudinal data collection, assessing the same users at multiple time points. This temporal validation demonstrated that prior beliefs about usefulness and ease of use predicted subsequent usage, strengthening causal inference and external validity evidence beyond cross-sectional correlation

How is the model intended to be used in practice?

Davis designed the Technology Acceptance Model as both research framework and practical tool for predicting and improving technology acceptance: User acceptance prediction: Information systems organizations can use TAM to predict which newly developed or acquired information technologies will achieve user acceptance. By measuring potential users’ perceived usefulness and perceived ease of use before or shortly after technology introduction, organizations can forecast acceptance likelihood. High perceived usefulness and ease of use predict strong acceptance; low values predict acceptance problems. This predictive capability enables organizations to anticipate acceptance issues early, before costly implementation failures.

• Acceptance barrier diagnosis: TAM enables organizations to diagnose which factors create acceptance barriers. If acceptance problems emerge despite adequate functionality, TAM specifies that the barriers reflect either low perceived usefulness (users question whether technology enhances performance) or low perceived ease of use (users perceive technology as difficult or effortful). This diagnostic distinction guides intervention design. If usefulness perceptions are low, interventions should focus on demonstrating performance benefits. If ease of use perceptions are low, interventions should focus on simplifying interfaces, providing training, or reducing required effort
• Design guidance for system developers: TAM provides systems developers with practical design guidance emphasizing that technical functionality alone proves insufficient for user acceptance. Even highly functional systems fail if users perceive them as difficult to use or unable to enhance their performance. Developers should prioritize perceived ease of use by designing intuitive interfaces, minimizing required keystrokes, providing helpful feedback, and ensuring minimal training requirements. Developers should also ensure functionality directly addresses user needs, enhancing performance on important work dimensions rather than offering capabilities users don’t value
• Implementation strategy development: Organizations can use TAM to design technology implementation strategies maximizing acceptance. By first assessing baseline perceived usefulness and ease of use before implementation, organizations identify which barrier types require addressing. Organizations can develop targeted strategies increasing usefulness perceptions through demonstrations, training showing performance benefits, or performance metrics documenting improvements. Organizations can increase ease of use perceptions through improved interfaces, comprehensive training, help desk support, or system customization reducing learning requirements
• Intervention evaluation: Organizations can measure perceived usefulness and ease of use before interventions, during implementation, and after adoption to assess whether specific improvements successfully enhanced these perceptions. Measuring usage changes alongside perception changes enables evaluation of whether perception improvements translated into actual acceptance and usage increases
• Comparative technology assessment: Organizations evaluating competing technology options can assess which options score highest on perceived usefulness and ease of use among target user populations. This assessment informs technology selection, potentially identifying user- acceptable options that might otherwise be overlooked if decisions rested only on technical specifications. Conversely, technologies scoring low on user perceptions despite high technical capability may need redesign or implementation strategies addressing perception gaps before deployment

What does the model measure?

The Technology Acceptance Model operationalizes several key measurement constructs: Perceived usefulness: Measured through multi-item scales assessing users’ beliefs that using a system enhances work performance. Items capture perceptions that system use improves job performance, increases work productivity, enhances work effectiveness, and makes work more efficient. Perceived usefulness represents outcome expectations—beliefs that technology adoption produces valued consequences (improved performance, greater productivity, enhanced effectiveness).

• Perceived ease of use: Measured through multi-item scales assessing users’ beliefs that using a system requires minimal cognitive and physical effort. Items assess perceptions that learning a system is easy, that interaction with a system is clear and understandable, that the system is easy to operate, and that using the system flexibly becomes easy. Perceived ease of use represents effort expectations—beliefs that technology use requires moderate rather than excessive learning or effort
• Attitude toward using the system: Measured through multi-item evaluative scales assessing overall favorable or unfavorable evaluations of system use. Attitude captures holistic evaluations of whether using technology is good/bad, harmful/beneficial, or unpleasant/pleasant. Attitudes represent overall affective and evaluative responses to technology use
• Behavioral intention to use: Measured through items assessing users’ plans, likelihood, or expectations to use systems. Behavioral intention captures the readiness to use technology, the stated likelihood of using technology, or the plan to use technology in the future
• Actual system usage: Measured through system usage logs (amount of time using system, frequency of use, breadth of features used) or user self- report of usage frequency. Actual usage represents meaningful behavioral outcomes—whether favorable beliefs and intentions translated into actual technology utilization

What are the main strengths of the model?

The Technology Acceptance Model possesses several considerable strengths: Theoretical parsimony: TAM provides parsimonious explanation of technology acceptance through two primary belief categories (perceived usefulness and ease of use). This economy of constructs makes the model relatively simple to understand, measure, and apply compared to models incorporating numerous factors. The parsimony does not sacrifice explanatory power—these two constructs explain substantial variance in technology acceptance. This combination of simplicity and explanatory power makes TAM remarkably practical and implementable.

• Technology-specific operationalization: Unlike generic behavioral models treating technology adoption identically to other behaviors, TAM identifies specific beliefs central to technology acceptance. By focusing on usefulness and ease of use rather than generic outcome beliefs, TAM captures what actually matters for technology decisions. Users evaluating technologies ask specifically about performance impacts and effort requirements. TAM directly operationalizes these technology-specific concerns
• Grounding in established behavioral theory: TAM builds on the Theory of Reasoned Action, inheriting the theoretical rigor and empirical support of TRA while specializing it to technology contexts. This grounding in established behavioral theory provides theoretical credibility and connection to the broader behavioral science literature
• Empirical validation: Davis provided solid empirical evidence supporting TAM. Validation across multiple systems, demonstration of relationships between attitudes and usage, and longitudinal prediction of behavior provided confidence in TAM’s validity. The solid empirical foundation distinguished TAM from purely theoretical models lacking validation
• Practical applicability: TAM provides clear guidance for predicting and improving technology acceptance. Organizations can measure perceived usefulness and ease of use, diagnose barriers, and design interventions. The practical utility made TAM immediately valuable for IS professionals, enabling evidence-based technology acceptance management
• Successful prediction of behavior: Unlike many attitude studies showing weak attitude-behavior relationships, TAM demonstrated that measured attitudes successfully predicted actual technology usage. This behavioral prediction validity distinguishes TAM and strengthens confidence in its practical utility
• Clear conceptual distinction: The clear distinction between perceived ease of use (effort required) and perceived usefulness (performance benefits) represents conceptual clarity. These distinct constructs address different decision dimensions—people care about both whether technology works and whether it requires excessive effort. Separating these concerns enables nuanced understanding of technology acceptance drivers
• Applicability across diverse technologies: TAM successfully predicted acceptance for both communication systems (email) and productivity tools (file managers), suggesting generalizability across technology types. Subsequent research extended TAM to spreadsheets, word processors, and numerous other technologies, confirming broad applicability

What are the main weaknesses of the model?

Despite considerable strengths, the Technology Acceptance Model has notable limitations: Limited construct breadth: TAM focuses narrowly on perceived usefulness and ease of use as technology attitude determinants. However, other factors influence technology acceptance: social influence and normative pressures, trust in technology and system providers, compatibility with existing work processes, organizational support and implementation quality, individual differences in technology anxiety or experience, and external constraints. By excluding these constructs, TAM provides incomplete explanation of technology acceptance variance. Studies incorporating additional constructs often find they explain variance beyond TAM’s core constructs.

• Underspecified causal mechanisms: TAM specifies that perceived ease of use influences perceived usefulness (easier systems seem more useful) and both influence attitudes, but provides limited theoretical specification of these relationships. Why exactly should ease of use influence usefulness perceptions? The model suggests that easier systems enable better performance, but this relationship may not always hold. Complex technologies sometimes require effort but produce dramatic performance improvements. The mechanism linking ease of use to usefulness deserves deeper theoretical explication
• Limited attention to usage intention stability: TAM treats behavioral intention as stable predictor of usage. However, intentions may shift between measurement and behavior due to intervening experiences. Users may intend to use systems but abandon usage due to actual problems encountered, changed work circumstances, or competing demands. Users with initial resistance may increase usage as skills develop. TAM provides limited specification of how intentions change over implementation periods or how to maintain intended usage through implementation challenges
• Insufficient attention to organizational context: While TAM addresses individual user beliefs, it provides limited specification of organizational factors affecting technology acceptance. Organizational implementation quality, support systems, change management, reward structures, and management advocacy substantially affect technology acceptance but receive limited attention in TAM. User beliefs about usefulness and ease may prove optimistic or pessimistic depending on organizational support and implementation quality. TAM assumes that individual beliefs determine acceptance but underemphasizes organizational context
• Actual usage measurement challenges: The model predicts usage from intentions and beliefs, but actual usage measurement proves problematic. System log data captures quantitative usage but not quality of usage or integration with work processes. Users might use systems minimally (checking email once weekly) while still being measured as users. Some usage (mandatory compliance) differs fundamentally from intentional usage. TAM provides limited attention to meaningful usage versus superficial compliance
• Individual differences undertheorized: TAM treats perceived usefulness and ease of use as sufficient explanations but provides limited attention to individual differences in technology experience, technological anxiety, cognitive styles, or learning preferences. Individuals with identical usefulness and ease of use perceptions may show different acceptance based on personal technology history or confidence. Similarly, system complexity appropriate for experienced users may exceed capacity for inexperienced users despite identical objective difficulty. Individual moderators of TAM relationships deserve greater theoretical attention
• Professional context limitations: TAM was developed and validated in professional information systems contexts. Applicability to consumer technologies, home information systems, or personal technologies remains less established. Consumer technology decisions involve different considerations than workplace technology adoption. TAM’s applicability generalizing beyond professional contexts requires additional validation
• Limited attention to implementation factors: TAM focuses on belief formation but provides limited specification of how successful implementation translates beliefs into actual integrated work practices. A user might believe a system is useful and easy yet fail to implement it into work processes due to resistance from others, workflow incompatibilities, or competing priorities. The gap between intention and integrated usage deserves greater attention

How does this model differ from older models?

The Technology Acceptance Model represented significant advancement from prior information systems research: Technology-specific focus: Earlier IS research examined user attitudes toward information systems but treated them as generic organizational innovations without technology-specific consideration. TAM recognized that technology acceptance involves specific beliefs about technology performance and effort requirements. Rather than applying generic innovation adoption frameworks, TAM identified technology-specific constructs central to technology decisions.

• Specialization of general behavioral theory: Earlier IS research either developed purely descriptive accounts of technology acceptance factors or attempted applying general behavior models without specialization. Davis’s fundamental innovation was taking the Theory of Reasoned Action—a general behavioral framework—and specializing it to technology contexts by identifying perceived usefulness and ease of use as the primary beliefs shaping technology attitudes. This specialization combined theoretical rigor with practical relevance
• Quantitative attitude measurement: Earlier IS studies often assessed acceptance qualitatively through interviews or observations. Davis developed reliable quantitative measurement scales enabling precise measurement and statistical testing. This methodological advancement enabled more rigorous model testing and replication
• Integration of effort and performance dimensions: Earlier technology adoption research often emphasized either performance benefits or ease of use but rarely integrated both. TAM recognized these as distinct but interrelated dimensions both influencing acceptance. The integration acknowledged that technology acceptance depends on both achieving valued outcomes and requiring reasonable effort
• Focus on user beliefs rather than features: Earlier technology adoption work sometimes focused on technological features or capabilities, assuming that advanced features would drive acceptance. TAM shifted focus to user perceptions of technology, recognizing that actual technology characteristics matter only insofar as they shape user beliefs. Two systems with identical technical capabilities might produce different acceptance based on how users perceive them. This shift toward perceptions rather than objective features represented important conceptual advance
• Explicit attention to attitudes as acceptance mediators: Earlier models sometimes treated technology characteristics as directly determining usage. TAM specified the mechanism through which technology characteristics affect usage: through shaping user attitudes, which influence intentions, which drive usage. This explicit attention to psychological mediation represented conceptual advancement over purely structural or characteristic-based models
• Longitudinal behavioral prediction: Earlier attitude studies typically examined cross-sectional relationships. Davis demonstrated longitudinal prediction—measured attitudes predicting subsequent actual usage months later. This temporal validation distinguished TAM from studies showing attitudes and behavior correlate but unable to establish temporal precedence

What Barriers to Technology Adoption does the model identify?

The Technology Acceptance Model identifies barriers to technology adoption organized around two primary psychosocial dimensions: Perceived usefulness barriers: The model identifies that users fail to adopt technologies when they perceive low usefulness—when they question whether technology adoption will enhance work performance or productivity. Perceived usefulness barriers include beliefs that technology adoption does not address important work needs. Users adopting technology that solves problems they don’t experience or adds capabilities they don’t require develop low usefulness perceptions. For example, workers in jobs not involving data analysis perceive spreadsheet technology as low usefulness. Technologies addressing specific job types but not others create use barriers within organizations. Usefulness barriers additionally emerge from beliefs that technology adoption offers minimal performance benefits relative to current practices.

Users whose existing processes prove reasonably efficient perceive new technology as offering marginal improvement. Technologies requiring substantial learning and transition to achieve modest performance gains encounter usefulness barriers. Usefulness barriers include poor integration with work requirements and workflows. Technology designed for different work contexts, industries, or user needs creates usefulness barriers. Systems designed for office professionals prove low-usefulness for field workers or creative professionals with different work processes. Usefulness barriers further include beliefs that technology adoption introduces unintended negative consequences reducing overall usefulness. Technology adoption might improve some work dimensions while creating problems in others. Systems increasing productivity but reducing work enjoyment, increasing monitoring, or fragmenting social collaboration create usefulness barriers when negative consequences outweigh positive benefits. Usefulness barriers additionally reflect insufficient performance evidence.

Users lacking clear evidence that technology improves performance perceive low usefulness. Without demonstrations, testimonials from satisfied users, or performance metrics showing improvements, users remain skeptical about usefulness claims.

• Perceived ease of use barriers: The model identifies that users resist technology adoption when they perceive high effort or difficulty requirements. Ease of use barriers include beliefs that technology requires excessive learning or training. Complex systems with unintuitive interfaces, non-obvious functionality, or steep learning curves create ease barriers. Users perceiving months of training requirements before proficiency develop negative ease of use perceptions. Systems requiring background knowledge (programming, system administration, or specialized terminology) create ease barriers for users lacking this knowledge. Ease of use barriers additionally include concerns that interaction with technology proves cognitively or physically demanding. Systems requiring sustained attention, complex decision sequences, or high precision input create barriers for users with limited cognitive resources or physical capabilities. Workers with learning differences, older workers with declining cognitive capacity, or individuals with physical disabilities perceive ease barriers others might not encounter. Ease of use barriers include concerns that technology requires excessive effort to master complex features. Even relatively simple systems prove difficult if documentation proves inadequate, interface design proves unintuitive, or help systems prove hard to access. Ease of use barriers additionally reflect concerns that technology adoption introduces disruption and temporary productivity loss during transition. Users accepting that technology will eventually ease work but fearing transition difficulties develop ease barriers. The prospect of months without proficiency, during which work productivity drops, creates barriers despite expected eventual benefits. Ease of use barriers further include concerns about maintaining learned skills. Technologies requiring periodic use might create barriers if users fear forgetting how to use systems between usage periods, requiring relearning
• Interaction of usefulness and ease of use barriers: The Technology Acceptance Model recognizes that these barrier types interact. Technologies requiring high effort but producing high usefulness overcome ease barriers through clear benefit demonstration. Users accepting that learning requires effort may persevere if persuaded that benefits justify effort. Conversely, technologies requiring low effort but providing minimal usefulness fail despite ease of use. Users adopting easy-to-use systems prove unwilling to use them if they question usefulness. The combinations of high-effort/high-usefulness, low-effort/high-usefulness, and low-effort/low- usefulness create different adoption profiles. High-effort/low-usefulness technologies face insurmountable barriers
• Secondary barriers reflecting usefulness and ease concerns: The Technology Acceptance Model implies that perceived usefulness and ease of use represent fundamental barriers that manifest through multiple specific concerns. Concerns about job security, professional deskilling, or work satisfaction represent manifestations of usefulness barriers when users believe technology will make work less meaningful. Concerns about technology malfunction, data loss, or security breaches represent ease of use barriers when users worry about effort required to manage these risks. Concerns about supervisor support or peer adoption represent usefulness and ease barriers through social demonstration: when important others haven’t adopted, users doubt usefulness or ease

What does the model instruct leaders to do in order to reduce these barriers?

The Technology Acceptance Model provides explicit guidance for leaders designing interventions to reduce technology adoption barriers: Establish perceived usefulness through benefit demonstration: Leaders instructed by TAM should demonstrate clear performance benefits that technology adoption will provide. Rather than assuming users will recognize benefits, leaders should proactively communicate and demonstrate specific improvements in productivity, work efficiency, error reduction, quality improvements, or reduced time requirements. Demonstration methods should directly show benefits relevant to specific user populations. Leaders should provide access to demonstrations enabling potential users to observe technology in operation. Leaders should share testimonials and case studies from comparable users in similar work contexts showing performance improvements.

• Leaders should quantify benefits using performance metrics: “Users of this system reduce data entry time by 30%” or “Organizations implementing this system report 15% productivity improvements.” Quantified benefits prove more persuasive than general claims. Leaders should tailor benefit demonstrations to specific user needs and job requirements. Salespeople care about benefits enhancing customer relationships; accountants care about benefits improving accuracy; managers care about benefits providing visibility. Matching benefit demonstrations to specific role concerns increases persuasiveness
• Address specific usefulness concerns: Leaders should conduct formative research identifying whether usefulness barriers reflect concerns about performance, work integration, negative consequences, or insufficient evidence. For technologies addressing non-core job functions, leaders should help users understand how the technology integrates with their actual work and produces benefits in their specific context. For technologies potentially introducing negative consequences (monitoring, reduced autonomy, job changes), leaders should address these concerns directly. Demonstrating that the organization implements technology in ways preserving autonomy, limiting surveillance, or supporting rather than replacing workers can alleviate concerns. For technologies where evidence of usefulness remains limited, leaders should invest in performance measurement systems documenting improvements. Collecting usage and performance data for early adopters enables generation of evidence convincing skeptical non-adopters
• Reduce perceived complexity through design and training: Leaders should ensure technologies provide perceived ease of use through both system design and support structures. From a design perspective, user interface design should emphasize simplicity and intuitiveness. Interfaces should require minimal training, use familiar language and terminology, organize functions logically, and provide clear feedback to user actions. Systems should support multiple entry methods accommodating different user preferences and expertise levels. Help functions should be accessible, context-sensitive, and comprehensible to non-expert users. Leaders should avoid assuming users possess technical background knowledge; systems should accommodate users lacking technology experience
• Provide comprehensive training and support: Leaders should provide training extensive enough that users achieve proficiency before adoption demands accelerate. Training should accommodate different learning styles and paces, with extended practice opportunities for slower learners. Training should emphasize practical skills (how to accomplish specific work tasks) rather than technical features. Training should build confidence through graduated complexity, starting with simple functions and advancing to complex features as skills develop. Training should establish support systems (help desks, peer experts, documentation) accessible to users encountering difficulties post-training. Ready access to support reduces perceived effort by enabling problem resolution when issues arise
• Manage implementation transition strategically: Leaders should recognize that even easy-to-use systems create transition barriers when implementation disrupts work. Leaders should manage transitions through phased implementation enabling users to learn during low-pressure periods, pilot programs enabling limited experimentation before full implementation, or temporary productivity reductions expected and accepted by management. During transitions, leaders should maintain reduced expectations for productivity, acknowledging that workers acquiring new skills will have reduced capacity for original work temporarily. Communicating these expectations reduces user anxiety about transition performance. Leaders should provide intensive support during transition