Research & Evidence

Search for a method

Search titles, questions, territories and MSC identifiers.

    ← All methods
    METHOD DOSSIERMSC-P-027Measurement scienceVerified scientific dossier

    TAM, UTAUT or UTAUT2: which framework should be used to study technology acceptance?

    UTAUT2 adds hedonic motivation, price value and habit to UTAUT; TAM stays with perceived usefulness and ease of use. Construct diagram and measurement items. The choice depends on the population, context and behavior under study, not on the number of available variables.

    Scientific editorial team: Marketing Science Center

    Direct answer

    Choose a framework, define its constructs, and test a measurement model followed by a structural model in a declared population.

    UTAUT2 adds hedonic motivation, price value and habit to UTAUT; TAM stays with perceived usefulness and ease of use. Construct diagram and measurement items. The choice depends on the population, context and behavior under study, not on the number of available variables.

    Davis, 1989, Perceived Usefulness, Perceived Ease of Use, and User Acceptance of Information TechnologyVenkatesh et al., 2003, User Acceptance of Information Technology: Toward a Unified View

    01

    The answer in 30 seconds

    TAM focuses on perceived usefulness and ease of use. UTAUT adds organisational context, facilitating conditions and moderators. UTAUT2 extends the framework to consumers with hedonic motivation, price value and habit. Select the framework before analysis from the population and decision, never from the most favourable result.

    02

    Three reading levels

    • Decision maker: which framework fits the decision?
    • Practitioner: which items, groups and occasions should be collected?
    • Analyst: measurement, structure, uncertainty and robustness.

    03

    Concrete marketing situation

    A team is preparing a digital service launch. It must distinguish employees’ initial acceptance, consumers’ voluntary use and repeated use. These populations do not imply the same constructs or decisions.

    04

    Scientific question

    Within a declared population, context and horizon, which constructs are associated with intention and self-reported use, and with what uncertainty?

    05

    Why an automatic choice fails

    Adding every construct increases respondent burden, overlap risk and analytical degrees of freedom. Choosing the framework with the highest R² after estimation turns comparison into opportunistic selection.

    06

    Intuition of the three frameworks

    FrameworkDominant questionContext
    TAMPerceived usefulness and effortParsimonious acceptance
    UTAUTPerformance, effort, influence, conditionsOrganisation and deployment
    UTAUT2UTAUT + hedonic motivation, price value, habitConsumer use
    The UTAUT2 model and its constructsPerformance expectancy≈ TAMEffort expectancy≈ TAMSocial influenceFacilitating conditionsHedonic motivation+ UTAUT2Price value+ UTAUT2Habit+ UTAUT2Behavioral intentionUseModerators: age, gender, experienceon most of the links
    Theoretical model of UTAUT2 (Venkatesh, Thong and Xu, 2012): seven constructs are posited as determinants of behavioral intention; intention, facilitating conditions and habit as determinants of use. UTAUT2 adds hedonic motivation, price value and habit to UTAUT (marked “+ UTAUT2”), as well as the link from facilitating conditions to intention (thick line). By contrast, UTAUT2 drops voluntariness of use, a UTAUT moderator. Age, gender and experience moderate most of the links, not always all three. Performance expectancy subsumes TAM’s perceived usefulness, and effort expectancy its perceived ease of use (marked “≈ TAM”).

    07

    Required data

    • Population, technology, voluntary or mandatory setting, date and horizon.
    • Items from the selected scale, documented translation and identical response options.
    • Intention measured before self-reported use when use is the target.

    UTAUT2 and TAM measurement items

    Measurement instrument of UTAUT2 and the TAM: constructs and items
    ConstructCodeWhat the items measureNumber of itemsOriginal scale
    Performance expectancyPEUsefulness of the technology in daily life, speed and productivity4 (3 kept after measurement validation)Venkatesh et al. (2003)
    Effort expectancyEEEase of learning and use, clarity of the interaction4Venkatesh et al. (2003)
    Social influenceSIWhat important others and influential people think of the use3Venkatesh et al. (2003)
    Facilitating conditionsFCAvailable resources, knowledge, compatibility and help4Venkatesh et al. (2003)
    Hedonic motivationHMFun and entertainment derived from use3Kim et al. (2005)
    Price valuePVPrice judged reasonable for what the technology delivers3Dodds et al. (1991)
    HabitHTUse that has become a habit, up to dependence4 (3 kept after measurement validation)Limayem & Hirt (2003)
    Behavioral intentionBIIntention to keep using the technology (existing users)3Venkatesh et al. (2003)
    UseUSEVariety and frequency of use of six applications, from “never” to “many times per day” (formative index)6Venkatesh, Thong & Xu (2012)
    Perceived usefulness (TAM)PUExpected performance gain from the technology6Davis (1989)
    Perceived ease of use (TAM)PEOUEffort required to use the technology6Davis (1989)

    In Venkatesh, Thong and Xu, the construct items concern mobile Internet in Hong Kong and are rated on a 7-point Likert scale from “strongly disagree” to “strongly agree”; use is rated as a frequency; age is measured in years and experience in months. One performance expectancy item and one habit item were dropped after estimation, for low loadings in that sample. The items, written in English, are in the article’s appendix and are not reproduced here; the questionnaire was administered in Chinese after translation and back-translation. For another technology or language, adapt the wording, document the translation, then re-establish reliability and validity before any estimation. The formative use index is not adapted by rewording: its list of applications must be rebuilt for the technology under study, and it is judged by its weights, not by internal consistency.

    08

    Formal model and symbols

    η = Bη + Γξ + ζ

    η denotes endogenous constructs, ξ exogenous constructs, B relations among endogenous constructs, Γ effects of exogenous constructs on η, and ζ the residual. Moderation enters only through prespecified interaction terms; the measurement model links latent constructs to their items.

    09

    Declared calculation

    1. Declare the framework and exclusions.
    2. Test the measurement model before the structural model.
    3. Estimate relations, intervals and diagnostics on the declared specification.
    4. Compare only prespecified models on the same observations.

    10

    Published numerical example

    In the Hong Kong UTAUT2 study of experienced mobile Internet users, 4,127 valid first-wave respondents become 2,220 at wave two and 1,512 in the final sample. Mean age is about 31 and only one technology is studied. Reported R² rises from 0.56 to 0.74 for intention and from 0.40 to 0.52 for self-reported use. The 0.18 and 0.12 differences have no published uncertainty interval; they describe that sample and specification, not universal superiority.

    11

    Validity assumptions

    • Items represent constructs in the studied population.
    • Compared groups share sufficiently invariant measurement.
    • A cross-sectional association proves neither temporality nor causality.

    12

    Diagnostics and uncertainty

    Document missing data, reliability, convergent and discriminant validity, invariance, intervals, holdout or later-wave stability, and sensitivity to removed items. Unavailable remains unavailable, never zero.

    13

    Interpretation

    A coefficient or R² describes the studied specification. Read it with uncertainty, measurement quality and scope. It does not automatically rank product initiatives.

    14

    Supported and forbidden conclusions

    Supported

    In this population, these constructs are associated with intention or use under the declared model.

    Forbidden

    Claim that changing a score will cause adoption, that the framework is universal, or that R² validates a decision.

    15

    Possible marketing decision

    The team may select TAM for parsimonious acceptance diagnosis, UTAUT for organisational deployment, or UTAUT2 for consumer use where pleasure, price and habit are substantively required. The final choice remains human and documented.

    16

    When to use each framework

    Do not use these frameworks alone to estimate causal effects, forecast aggregate demand, value a technology financially or replace usage research. Alternatives include experiments, longitudinal panels, choice models, qualitative research or diffusion models depending on the decision.

    17

    Reproducible implementations

    The reference file recalculates only the two reported R² differences, without refitting the model or fabricating individual data.

    CSV · Python 3.13.7

    # R 4.5.1: transform(read.csv("msc-p027-tam-utaut.csv"), delta=r2_utaut2-r2_utaut)
    * SPSS 31: COMPUTE delta = r2_utaut2 - r2_utaut.
    * SAS 9.4: data result; set input; delta=r2_utaut2-r2_utaut; run;

    18

    Final deliverable

    • Decision, population, technology, horizon and target behaviour.
    • Selected framework, included and excluded constructs, item provenance and analysis plan.
    • Results with uncertainty, diagnostics, limitations and reassessment rule.

    19

    Sources and evidence level

    E3 evidence: three foundational full texts verified at claim level. The comparison remains a methodological synthesis; it proves neither commercial performance nor causality.

    Read next

    MSC-P-005How do you measure a marketing construct that is not directly observable?→MSC-P-006How do you design and validate a measurement scale?→MSC-P-008How do you validate a marketing measurement scale?→MSC-P-009PCA, EFA or CFA: which method should you choose?→