Development and psychometric validation of the Thai AI Literacy Scale for Nursing Students in Thailand: a methodological study

Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

The conceptual framework draws on internationally established AI literacy dimensions, suggesting potential transferability to other Southeast Asian nursing curricula where similar professional standards and AI adoption trajectories apply.

“The conceptual framework draws on internationally established AI literacy dimensions [1,2], suggesting potential transferability to other Southeast Asian nursing curricula where similar professional standards and AI adoption trajectories apply.”

• cites [1] sourceNot checked — no full text of the source
• cites [2] sourcePartly supported

  The Abstract identifies four AI literacy dimensions drawn from a review of 30 articles, supporting the claim that the citing framework draws on AI literacy dimensions. But it describes AI literacy as newly emerging and under-explored, not the dimensions as internationally established. The suggested transferability to Southeast Asian nursing curricula is the citing authors’ inference, not a finding attributed to this source. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

It may state two separate claims as one.

Existing scales provide robust measures of general AI literacy but lack domain-specific items relevant to nursing practice and were validated in Western educational contexts.

“Existing scales, such as those developed by Laupichler et al. [3,4], provide robust measures of general AI literacy, but they lack domain-specific items relevant to nursing practice and were validated in Western educational contexts.”

• cites [3] sourceNot checked — no full text of the source
• cites [4] sourcePartly supported

  The Background describes SNAIL as a validated measure of broad AI literacy, and the Discussion reports good model fit and excellent internal consistency in this study. The Methods list general AI-literacy topics and identify the study population as German medical students, but the paper does not establish that the scales lack nursing-specific items or that all their validation occurred in Western educational contexts. It also identifies SNAIL as a scale developed in an earlier Laupichler et al. work, rather than in this study. (LLM, conf=0.83)

  Experimental — accuracy not yet measured

It may state two separate claims as one.

The HTMT value between positive attitude and readiness was 0.969, exceeding the 0.90 threshold and suggesting potential construct overlap.

“however, the HTMT value between positive attitude and readiness was 0.969, exceeding the 0.90 threshold and suggesting potential construct overlap.”

• Also stated in: § Interpretation · hedged

  “However, the HTMT ratio between positive attitude and readiness (0.969) exceeded the 0.90 threshold, suggesting partial construct overlap.”

• cites [10] sourceSupported

It may state two separate claims as one.

Most HTMT ratios were below 0.90, supporting discriminant validity.

“Most heterotrait-monotrait ratios were below 0.90, supporting discriminant validity.”

• Also stated in: § Validity · hedged

  “Most HTMT values were below 0.90, supporting discriminant validity [10]; however, the HTMT value between positive attitude and readiness was 0.969, exceeding the 0.90 threshold and suggesting potential construct overlap.”

• cites [10] sourceSupported

It may state two separate claims as one.

Items d3 and d5 loaded exclusively on the AI anxiety factor, supporting the conceptual separation of this domain from positive attitude.

“Items d3 and d5 loaded exclusively on the AI anxiety factor, supporting the conceptual separation of this domain from positive attitude (Supplement 5).”

It may state two separate claims as one.

The TAILS-NS was developed through a 3-phase process resulting in a 40-item instrument comprising a 15-item knowledge test and 25 Likert-scale items across 6 domains.

“The TAILS-NS was developed through item generation, expert content-validity assessment using the item-objective congruence index, and a 2-phase pilot study, resulting in a 40-item instrument comprising a 15-item knowledge test and 25 Likert-scale items across 6 domains.”

• Also stated in: § Data sources and measurement · plain

  “The instrument comprised a 15-item knowledge test, scored dichotomously as 1=correct and 0=incorrect, and 25 Likertscale items, scored from 1=strongly disagree to 5=strongly agree, across 6 conceptual domains.”

It may state two separate claims as one.

The 2-item AI anxiety specification was successfully estimated within the 6-factor model in the development sample (n=410) but did not achieve stable convergence in the smaller independent validation sample (n=157).

“Although this specification was successfully estimated within the 6-factor model in the development sample (n=410) (Fig. 3), it did not achieve stable convergence in the smaller independent validation sample (n=157).”

It may state two separate claims as one.

No significant mean difference was observed between Time 1 and Time 2 for the total score (paired t(64)=-0.863, P=0.391), indicating the absence of a systematic response shift.

“No significant mean difference was observed between Time 1 and Time 2 for the total score (paired t(64)=-0.863,P=0.391), indicating the absence of a systematic response shift.”

It may state two separate claims as one.

Both positive attitude and readiness factors demonstrated adequate AVE, satisfied the Fornell-Larcker criterion, and retained unique item content.

“Importantly, both factors demonstrated adequate AVE, satisfied the Fornell-Larcker criterion, and retained unique item content.”

It may state two separate claims as one.

In the independent validation sample, RMSEA was 0.138 (90% CI, 0.128-0.147), higher than in the development sample (0.090), likely reflecting the smaller sample size and RMSEA's sensitivity to sample size.

“RMSEA was 0.138 (90% confidence interval [CI], 0.128-0.147),which was higher than that in the development sample (0.090), likely reflecting the smaller sample size and RMSEA's sensitivity to sample size [11].”

• cites [11] sourceNot checked — no full text of the source

It may state two separate claims as one.

In the independent validation sample, all standardized factor loadings were statistically significant (P<0.001) and ranged from 0.676 to 0.966.

“All standardized factor loadings were statistically significant (P<0.001) and ranged from 0.676 to 0.966 (Supplement 7).”

It may state two separate claims as one.

The TAILS-NS addresses nursing-specific competencies, including patient data privacy, professional accountability, and the limitations of AI in psychosocial care.

“The TAILS-NS addresses nursing-specific competencies, including patient data privacy, professional accountability, and the limitations of AI in psychosocial care.”

It may state two separate claims as one.

Average variance extracted exceeded 0.50 for all factors, supporting convergent validity.

“Average variance extracted exceeded 0.50 for all factors, supporting convergent validity.”

• Also stated in: § Validity · plain

  “Convergent validity was supported, as AVE exceeded 0.50 for all factors (range, 0.743-0.834)[9].”

• cites [9] sourceNot checked — no full text of the source

It may state two separate claims as one.

All standardized factor loadings in the 6-factor model were statistically significant (P<0.001) and ranged from 0.765 to 0.978.

“All standardized factor loadings were statistically significant (P<0.001) and ranged from 0.765 to 0.978 (Table 5, Fig. 3).”

It may state two separate claims as one.

Domain-level ICCs ranged from 0.4331 for AI anxiety to 0.5355 for positive attitude, with values of 0.4899 for AI awareness, 0.4438 for AI skills, 0.4590 for ethics and professionalism, and 0.4416 for readiness.

“Domain-level ICCs ranged from 0.4331 for AI anxiety to 0.5355 for positive attitude, with values of 0.4899 for AI awareness, 0.4438 for AI skills, 0.4590 for ethics and professionalism, and 0.4416 for readiness.”

It may state two separate claims as one. The quote only partly states this claim.

Confirmatory factor analysis showed acceptable model fit for the proposed 6-factor model, although RMSEA indicated marginal fit (CFI=0.919, TLI=0.906, RMSEA=0.090).

“The proposed 6-factor model demonstrated acceptable fit (Table 4), representing a meaningful improvement over the 5-factor solution (ΔCFI=0.022).”

Probably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.

No incentives were offered, minimizing potential response bias.

“No incentives were offered, minimizing potential response bias.”

Probably not a substantial claim. The paper’s conclusions may not depend on it.

The study was reported in accordance with STROBE guidelines.

“The study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (Supplement 1).”

It may state two separate claims as one.

The inclusion of a knowledge test alongside attitudinal measures distinguishes the TAILS-NS from purely self-report instruments, providing a more comprehensive assessment of AI literacy as both a cognitive and dispositional construct.

“The inclusion of a knowledge test alongside attitudinal measures distinguishes the TAILS-NS from purely self-report instruments, providing a more comprehensive assessment of AI literacy as both a cognitive and dispositional construct.”

Probably not a substantial claim. It may state two separate claims as one.

Items with IOC values ≥0.50 were retained, whereas items with values of 0-0.49 were revised or deleted.

“Items with IOC values ≥0.50 were retained, whereas items with values of 0-0.49 were revised or deleted [6].”

• cites [6] Rovinelli, 1976Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Probably not a substantial claim. The paper’s conclusions may not depend on it.

All participants provided informed consent before completing the online survey.

“All participants provided informed consent before completing the online survey.”

It may state two separate claims as one.

The obtained sample of 410 participants exceeded the minimum sample size threshold and satisfied the recommended minimum of 300 participants for stable factor solutions.

“The obtained sample of 410 participants exceeded this threshold and satisfied the recommended minimum of 300 participants for stable factor solutions.”

Probably not a substantial claim. The paper’s conclusions may not depend on it.

The study was approved by the Suan Sunandha Rajabhat University Ethics Committee (certificate number: COE.2-109/2026).

“This study was approved by the Suan Sunandha Rajabhat University Ethics Committee (certificate number: COE.2-109/2026).”

Probably not a substantial claim. The paper’s conclusions may not depend on it.

In the independent validation sample, 157 respondents were included in the final analysis after excluding incomplete responses (n=4) and duplicate anonymous identifiers (n=4).

“Of 165 respondents, 157 were included in the final analysis after incomplete responses (n=4) and duplicate anonymous identifiers (n=4) were excluded.”

It may state two separate claims as one.

The sampling strategy relied on non-probability purposive institutional selection with voluntary responses, which may introduce selection bias toward students with greater digital access or motivation to participate.

“Finally, the sampling strategy relied on non-probability purposive institutional selection with voluntary responses, which may introduce selection bias toward students with greater digital access or motivation to participate.”

Probably not a substantial claim. The paper’s conclusions may not depend on it.

A try-out study was conducted with 30 nursing students not included in the main sample to assess item clarity and preliminary reliability.

“A try-out study was subsequently conducted with 30 nursing students who were not included in the main sample to assess item clarity and preliminary reliability (Supplement 4).”

It may state two separate claims as one.

A 2-item factor has zero degrees of freedom and is not identified without external constraints.

“A 2-item factor has zero degrees of freedom and is not identified without external constraints, such as correlations with other latent factors in the model [13].”

• cites [13] sourceNot checked yet

It may state two separate claims as one.

The TAILS-NS was developed and validated in a multi-regional Thai sample spanning diverse institutional types, supporting its applicability across nursing education contexts in Thailand.

“The TAILS-NS was developed and validated in a multi-regional Thai sample spanning diverse institutional types, supporting its applicability across nursing education contexts in Thailand.”

• AssumptionRests on an assumption: The multi-regional Thai sample is representative of nursing education contexts in Thailand.

Probably not a substantial claim. The paper’s conclusions may not depend on it.

Purposive institutional sampling was used to select institutions that would provide geographic representation across Thailand's major regions.

“Purposive institutional sampling was used to select institutions that would provide geographic representation across Thailand's major regions.”

Probably not a substantial claim.

A subset of 65 participants from the validation sample completed the TAILS-NS a second time approximately 2 weeks after initial administration.

“A subset of 65 participants from the validation sample completed the TAILS-NS a second time approximately 2 weeks after initial administration.”

Probably not a substantial claim. It may be too vague to check against evidence.

Conflating AI anxiety and positive attitude dimensions would obscure important nuances relevant to curriculum design.

“and conflating these dimensions would obscure important nuances relevant to curriculum design.”

Probably not a substantial claim.

Students who hold favorable attitudes toward AI may simultaneously perceive themselves as more prepared to adopt it, reflecting a motivational pathway rather than a measurement artifact.

“This finding is interpretable in context: students who hold favorable attitudes toward AI may simultaneously perceive themselves as more prepared to adopt it, reflecting a motivational pathway rather than a measurement artifact [4].”

• cites [4] sourceNot checked yet