The reported internal consistency figures merit careful interpretation. A Cronbach’s alpha of 0.857 for the full scale falls within the range commonly regarded as good for research instruments, and subscale coefficients in the vicinity of 0.81 to 0.82 indicate that each dimension, despite containing relatively few items after refinement, still produces coherent responses. In classical test theory, alpha depends on both the interrelatedness of items and the number of items, so achieving such values with shortened subscales is a favorable sign. It should be noted, however, that alpha estimates assume tau-equivalence and can be sensitive to sample characteristics; complementary reliability indices such as McDonald’s omega are often reported in contemporary validation studies to triangulate these estimates. The authors’ examination of composite reliability, average variance extracted, maximum shared variance, and average shared variance addresses convergent and discriminant validity at the construct level. When average variance extracted exceeds the squared inter-construct correlations—as the comparison of AVE with MSV and ASV is intended to demonstrate—each latent factor can be said to capture variance that its indicators share more with it than with neighboring constructs.

The two-stage factor-analytic sequence used in the study follows a logic that has become conventional in instrument development. Exploratory factor analysis serves as a hypothesis-generating step in which the underlying dimensionality of the item pool is discovered rather than imposed, while confirmatory factor analysis functions as a hypothesis-testing step in which the structure identified in one sample is evaluated against independent data. Demonstrating good fit in the confirmatory stage—typically assessed through indices such as the comparative fit index, the Tucker-Lewis index, the root mean square error of approximation, and the standardized root mean residual—provides stronger evidence than fit statistics from the exploratory stage alone, because the model was specified a priori and tested on data that did not shape its form. For readers evaluating the instrument’s readiness for use, the key question is whether the confirmatory evidence was obtained from a sample independent of the one used for item reduction; cross-validation of this kind materially strengthens claims about structural stability.

The developmental rationale for a child-specific instrument deserves emphasis. Children in elementary and middle school are at a stage in which concrete operational thinking predominates and transitions toward formal reasoning are uneven across individuals. Assessment items that rely on abstract vocabulary—terms such as algorithm, training data, or model bias—may measure reading comprehension or exposure to technical language rather than genuine conceptual understanding. Conversely, items anchored in familiar experiences, such as voice assistants, recommendation feeds, or image-recognition features, can tap into authentic mental models that children have formed through daily interaction. The literature on children’s conceptions of technology has repeatedly shown that young learners attribute intentionality, emotion, and even moral agency to intelligent systems, and that these anthropomorphic beliefs shape how they trust, question, or comply with AI outputs. An instrument calibrated to this developmental profile can therefore reveal misconceptions that adult-oriented tools would miss entirely.

The sampling frame of 870 students across grades four through eight, drawn from both elementary and middle schools, provides a reasonably broad developmental span for calibration. Including 424 boys and 346 girls, the sample covers the age range during which attitudes toward technology and self-efficacy beliefs about computing often begin to diverge along gendered lines, making the instrument potentially useful for monitoring such trends. A sample of this size is generally adequate for factor-analytic work, particularly when the number of items is moderate and communalities are acceptable. Nevertheless, the general survey model on which the design rests assumes that the participating schools and students represent the wider population of interest; if the sample was drawn from a limited geographic area or from schools with above-average technology access, generalization to other regions or under-resourced settings should be made cautiously. Measurement invariance testing—examining whether the factor structure, loadings, and intercepts hold across gender, grade level, and socioeconomic subgroups—would be a natural next step to confirm that scores mean the same thing for all students and can be compared fairly across groups.

The positioning of AI literacy alongside digital, data, and computational literacy situates the construct within a family of emerging literacies that education systems worldwide are beginning to institutionalize. Curricular frameworks in several countries now include AI concepts at the primary level, and international organizations have advocated for early and equitable exposure to AI education. Within this policy landscape, valid measurement tools function as the connective tissue between curriculum standards and classroom practice: without them, educators cannot determine whether instructional interventions produce gains in understanding, nor can policymakers evaluate the reach of national AI-literacy initiatives. The availability of a scale designed specifically for younger students addresses a documented gap, since the majority of prior instruments targeted adolescents, undergraduates, or adults whose cognitive profiles and technological experiences differ substantially from those of children in the middle years of schooling.

Practical applications of the instrument extend across several domains. In formative assessment, teachers could administer the scale before and after an AI-themed unit to gauge growth in specific subscales, identifying, for example, whether lessons improved students’ ability to distinguish AI-driven features from ordinary software. In research, the scale enables correlational and longitudinal studies linking AI literacy to outcomes such as critical thinking about online information, attitudes toward technology careers, or susceptibility to misinformation generated by AI systems. In program evaluation, schools deploying AI curricula, robotics clubs, or digital citizenship programs can use the instrument to document impact with a common metric. The six-dimension structure is particularly valuable here, because interventions often target narrow competencies—such as recognizing AI applications—while leaving evaluative and ethical dimensions untouched; a multidimensional measure can expose such uneven development.

Several limitations and future directions follow naturally from the study’s design. First, self-report scales capture students’ perceptions of their own competence, which may diverge from performance-based measures of actual skill; pairing the instrument with scenario-based or problem-solving tasks would strengthen the evidence base. Second, the rapid evolution of generative AI means that item content anchored in current technologies may require periodic revision as familiar applications change; instruments in this field face unusual shelf-life pressures compared with traditional psychological measures. Third, the study’s cross-sectional design cannot speak to how AI literacy develops over time or how instruction influences its trajectory; longitudinal administration of the scale across grade levels would illuminate developmental progressions. Fourth, validation in additional linguistic and cultural contexts is needed before the instrument can support international comparative research, since translation and adaptation can alter item difficulty and factor structure.

The ethical dimension of AI literacy among children adds urgency to this line of work. Young users routinely interact with recommendation algorithms, conversational agents, and personalized content streams without the scaffolding needed to recognize persuasion, data collection, or fabricated outputs. Research on children’s information evaluation has documented pronounced difficulty in distinguishing synthetic media from authentic content, and early adolescence is a period in which habits of verification either take root or fail to form. A validated measure of the evaluative and ethical facets of AI literacy gives educators a diagnostic entry point for these concerns, allowing them to identify classrooms or cohorts where critical scrutiny of AI outputs is weak and to design interventions accordingly. In this respect, the instrument contributes not only to measurement science but to the broader societal project of preparing minors to navigate an information environment increasingly co-produced by intelligent systems.

Finally, the open-access publication of the study facilitates reuse and scrutiny, conditions that historically accelerate the maturation of measurement tools. As other research groups administer the scale in diverse samples, accumulating evidence about its reliability, invariance, and predictive validity will either consolidate or qualify the initial findings reported here. The trajectory of similar instruments in adjacent fields suggests that early versions are frequently refined through item revision, subscale consolidation, or the addition of performance components; the present scale should be understood as a foundational step in an iterative process rather than a finished artifact. For the community of researchers and educators concerned with preparing young learners for an AI-saturated world, the study offers a concrete, developmentally attuned tool whose continued refinement will depend on the breadth and rigor of its future use.

Subject of Research: Development and validation of an artificial intelligence literacy scale for elementary and middle school students

Article Title: Development and validation of an artificial intelligence literacy scale for students

Article References: Demirtaş, E. N., & İşçi, T. G. (2026). Development and validation of an artificial intelligence literacy scale for students. Discover Education, 5(1), Article 1176. https://doi.org/10.1007/s44217-026-02048-x

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02048-x

Keywords: artificial intelligence literacy, scale development, psychometrics, elementary education, middle school students, exploratory factor analysis, confirmatory factor analysis, educational measurement, AI ethics, Cronbach's alpha, Development, validation

The conceptual heart of IDBEL is a decades-old idea from educational psychology: the affective domain. Drawing on the classic taxonomy of Krathwohl, Bloom, and Masia, the researchers treat digital citizenship attitudes as a process of progressive value internalisation with five levels. Receiving captures a student’s initial awareness and acceptance of digital norms. Responding reflects active engagement with those norms. Valuing indicates genuine appreciation of responsible digital behaviour. Organization describes the integration of digital values into a coherent personal value system, and characterization represents consistent, value-driven conduct online. Each of the five dimensions was translated into five Likert-scale statements, producing a 25-item questionnaire delivered entirely through a digital application, so that the mode of assessment mirrors the digital construct being measured.

The technical groundwork began before any student saw the instrument. Five experts in educational assessment, language, civic education, and information technology rated the relevance of each of the 25 initial items on a five-point scale, and the researchers quantified agreement using Aiken’s V, a content-validity coefficient. The scores ranged from 0.70 to 0.95, with a mean of 0.83, and the team applied an operational threshold of 0.80. Twenty items cleared the bar; five, specifically Items 3, 5, 10, 13, and 20, fell short and were revised in response to expert feedback. Notably, the researchers acknowledge that the revised versions were never re-rated, so the content-validity evidence describes the initial drafts rather than a guarantee that every final item meets the criterion. That kind of candour is rare in instrument-development papers and sets the tone for the entire study.

Data collection ran from October 2025 to February 2026 through the IDBEL application, drawing Grade X and XI students from seven public senior secondary schools in Banjarmasin, selected purposively to span urban and peri-urban contexts. Grade XII students were excluded because of final examinations. The final analytical matrix contained 661 students, with school contributions ranging from 34 to 142 participants. The descriptive results told a familiar story with an uncomfortable twist: students overwhelmingly reported positive digital citizenship attitudes, with combined high and very high responses ranging from 66.0 percent to 97.4 percent across items. One item, Item 12, drew a maximum-category endorsement from 61.4 percent of respondents, while another, Item 23, drew only 14.5 percent. That spread matters, because when responses pile up at the top of a scale, the instrument loses its ability to distinguish between students who are good and students who are exceptional.

Before any structural modelling, the team tested whether the data were even suitable for factor analysis, and the answer was emphatically yes. The Kaiser-Meyer-Olkin measure of sampling adequacy reached 0.963, a value considered excellent, and item-level adequacy statistics ranged from 0.930 to 0.980. Bartlett’s test of sphericity was overwhelmingly significant, χ²(300) = 9,751.236, p < 0.001, confirming that the correlation matrix contained enough shared variance to justify extracting factors. In plain terms, the items were related to one another in exactly the way a coherent attitude construct should be. The question then became whether the relationships followed the five-dimensional architecture the theorists had prescribed.

The answer, delivered through a prespecified second-order confirmatory factor analysis, was a qualified no. Standardized item loadings were respectable, ranging from 0.567 to 0.776, and the second-order loadings linking the five dimensions to the overarching attitude construct were extremely high, between 0.929 and 0.971. But the overall model fit fell short of accepted standards: the chi-square ratio of 5.567 was high, RMSEA came in at 0.083, just above the conventional 0.08 guideline, and the incremental fit indices CFI, TLI, and IFI all landed below 0.90, at 0.871, 0.857, and 0.872 respectively. Only SRMR, at 0.054, was comfortably within target. The researchers also flagged a methodological caveat: they used normal-theory maximum likelihood on Pearson correlations, which is suboptimal for ordinal items concentrated in upper response categories, and they call for an ordinal CFA using polychoric correlations and a weighted estimator in future work.

Perhaps the most intriguing technical finding concerns what those very high second-order loadings imply. Together with a McDonald’s omega hierarchical of 0.940, they suggest that a single dominant general factor accounts for most of the reliable variance in scores. In other words, the five affective dimensions may be so conceptually intertwined, progressive stages of the same internalisation process, that they cannot yet be interpreted as distinct subscales. Modification diagnostics reinforced this concern, identifying strong local dependencies between Items 1 and 2 and between Items 23 and 24, with additional cross-dimensional overlap for pairs such as Items 4 and 8 and Items 15 and 25. Crucially, the team refused the temptation to patch the model post hoc by correlating error terms or deleting items, arguing that such changes require content justification and independent-sample cross-validation.

Reliability, by contrast, was stellar. Cronbach’s alpha reached 0.953 for the total scale, McDonald’s omega total was 0.958, and corrected item-total correlations ranged from 0.489 to 0.727, with no single item’s removal materially improving consistency. Domain-level alpha coefficients ranged from 0.792 to 0.854 and composite reliability from 0.806 to 0.854, all comfortably above the 0.70 threshold. Yet the convergent validity picture was murkier. Average variance extracted, a measure of how much variance a latent dimension explains in its own indicators, ranged from 0.456 to 0.540 and crossed the conventional 0.50 criterion only for the Organization dimension, with Characterization sitting essentially at the threshold. The lesson the authors draw is an important one for anyone who reads assessment research: very high internal consistency does not by itself establish that an instrument measures what it claims, and it may partly reflect item similarity and response concentration rather than construct strength.

So where does IDBEL stand? The researchers position it deliberately as initial psychometric evidence, not a finished, validated scale. The study’s limitations are laid out with unusual precision: the sample covers one Indonesian city and excludes vocational schools and Grade XII students; the data are entirely self-reported and vulnerable to social desirability bias, since digital citizenship items describe morally desirable behaviour; and the validity argument remains incomplete, lacking evidence on relations to external variables, measurement invariance, test-retest reliability, and the consequences of test use. Future refinements they propose include scenario-based items presenting realistic digital dilemmas, carefully constructed reverse-coded items to counter acquiescence, and Item Response Theory or Rasch analysis to map item difficulty and discrimination, particularly at the upper end of the attitude continuum where current items blur together.

For educators, the practical message is twofold. First, digital citizenship can indeed be measured as an affective construct, not merely as technical skill, and a digital platform is a natural home for such assessment in civic education classrooms. Second, caution is warranted: IDBEL is suitable as a formative, low-stakes tool and a foundation for further development, but it should be paired with teacher observations, student reflections, and scenario-based tasks rather than used for high-stakes judgements. In an era when schools everywhere are being asked to certify not just what students know but who they are online, this study offers both a promising template and a sobering reminder that building a trustworthy measure of character is harder, and more honest, work than it looks.

Subject of Research: Psychometric validation of a digital instrument assessing students' digital citizenship attitudes

Article Title: Initial psychometric evidence of an innovative digital-based learning evaluation instrument for assessing students’ digital citizenship attitudes

Article References: Adawiah, R., Ruchliyadi, D. A., & Nugroho, D. A. (2026). Initial psychometric evidence of an innovative digital-based learning evaluation instrument for assessing students’ digital citizenship attitudes. Discover Education, 5(1), Article 997. https://doi.org/10.1007/s44217-026-02201-6

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02201-6

Keywords: digital citizenship, psychometrics, affective assessment, instrument validation, confirmatory factor analysis, digital assessment, civic education, reliability, convergent validity, Likert scale, Indonesia, educational measurement

The study, published in the journal Large-scale Assessments in Education, drew on the German National Educational Panel Study, a long-running effort that follows children’s competence development across the life span. The researchers, led by Jana Welling of the Leibniz Institute for Educational Trajectories, analyzed reading and mathematics assessments administered in Grades 2 and 4, with 1,512 children participating in the second-grade assessments and 1,470 in the fourth-grade versions. Because the tests were administered on tablets or computers, every response came with a precise timestamp, allowing the team to reconstruct exactly how long each child spent on each item.

Rapid guessing matters because large-scale assessments are typically low-stakes: nothing happens to a child’s grades, placement, or future based on the outcome. Expectancy-value theory suggests that when a task carries no personal consequence, motivation to invest effort can collapse. The demands-capacity model adds a second dimension, proposing that effort depends on the match between what an item demands and the cognitive resources a test-taker is willing and able to deploy. When either willingness or capacity falls short, disengaged responding emerges—and in its most extreme form, responses arrive at implausibly high speed with accuracy near chance level.

Among adolescents and adults, this is no small problem. A 2022 meta-analysis found that roughly 7 percent of responses in low-stakes assessments are rapid guesses, and about 28 percent of examinees rapid-guess at least once. Worse, uncorrected proficiency estimates are biased downward by an average of 0.13 standard deviations. But children, the researchers reasoned, may be a different story. Young children tend to be overconfident in their abilities, which could boost their expectancy of success and keep them engaged. At the same time, their shorter attention spans could work in the opposite direction, making long tests feel more demanding and pushing them toward disengagement.

To detect rapid guessing, the team used response-time thresholds: any answer delivered faster than a cutoff was flagged as a guess and scored as missing. Because no single threshold method works for every test design, the researchers compared several approaches. For the reading tests and the fourth-grade mathematics test, they applied normative threshold methods, defining the cutoff as 10 or 20 percent of the average response time for each item. For the second-grade mathematics test—where items were read aloud via pre-recorded audio and children could not answer before the audio finished—they used visual inspection methods based on the shape of the response-time distribution and on how accuracy changes with speed. Thresholds were then validated by checking whether flagged responses showed chance-level accuracy and near-zero correlation with overall test scores.

The headline result: rapid guessing was scarce. In second-grade reading, 3.3 percent of responses were classified as rapid guesses; in second-grade mathematics, just 0.8 percent. By fourth grade, the rates fell even lower, to 0.5 percent in reading and 0.4 percent in mathematics. Contrary to the researchers’ hypothesis, disengagement did not increase between Grades 2 and 4—in the reading domain, it actually declined. The authors suggest that the fourth-grade reading test may simply have been too easy for most children, since easier items tend to elicit less rapid guessing, and that the real rise in disengagement documented in earlier studies typically emerges around puberty rather than in early elementary school.

Who rapid-guesses? Across all four assessments, only one characteristic reliably predicted the behavior: proficiency. Less proficient children were consistently more likely to fire off quick answers than their higher-performing peers, echoing patterns well established in older samples. A one-unit increase in estimated ability raised the probability of engaged responding by between 0.52 and 1.96 percentage points depending on the assessment. Notably, gender and parental socioeconomic status—both of which have been linked to test-taking effort in adolescent and adult studies—showed no significant effects. Descriptively, boys and children without a university-educated parent did guess somewhat more often, but these differences were too small and inconsistent to reach statistical significance, suggesting that the social gradients in test engagement familiar from older populations may simply not have taken hold yet in eight- and ten-year-olds.

The most consequential finding concerns what rapid guessing does to the numbers that policymakers actually use. The researchers compared estimates of ability differences between girls and boys, and between children of low and high parental socioeconomic status, computed once from the raw data and once from data in which rapid guesses had been scored as missing. The differences were essentially zero. Whether the team examined the modest gender gap in fourth-grade reading, the male advantage in mathematics, or the substantial socioeconomic gaps present in every domain, accounting for rapid guessing shifted the estimates by at most 0.01 standard deviations, well within the bootstrap confidence intervals. In other words, even the small amount of guessing observed did not distort subgroup comparisons in any meaningful way.

Why doesn’t a behavior that biases average scores elsewhere leave group comparisons untouched? The authors point to two mechanisms. First, when most test-takers are engaged, the handful of rapid guesses carries little statistical weight. Second, rapid guessing is strongly tied to low proficiency—and because less proficient respondents have a low probability of answering correctly even when trying, a random guess changes little. For a bias in group comparisons to emerge, the groups would need to differ substantially in their guessing rates, which appears to be rare in practice, particularly among children.

The study carries practical weight for the assessment community. It suggests that the alarming disengagement rates documented in adolescent samples cannot simply be projected downward onto younger children, and that large-scale assessments involving elementary schoolers may be more robust than critics fear. But the authors caution against complacency. Prevalence varied considerably between assessments even within the same children, pointing to the strong influence of test design, time limits, item presentation, and proctoring conditions—and a lack of rapid guessing does not guarantee full engagement, since children may express disengagement in other ways, such as omissions or slow, effortful guessing. The team also notes that no single threshold method fits every test, and that the rarity of the behavior itself makes its predictors statistically hard to pin down. Future work, they argue, should examine broader forms of disengagement, test model-based identification approaches in child samples, and experimentally vary assessment settings. For now, though, the message is clear: the youngest test-takers, by and large, are still trying.

Subject of Research: Rapid guessing behavior and test-taking engagement among elementary school children in large-scale educational assessments

Article Title: Prevalence, predictors, and consequences of rapid guessing among children in elementary school

Article References: Welling, J., Fischer, L., Thums, K., & Gnambs, T. (2026). Prevalence, predictors, and consequences of rapid guessing among children in elementary school. Large-scale Assessments in Education, 14(1), Article 51. https://doi.org/10.1186/s40536-026-00319-5

Image Credits: AI Generated

DOI: 10.1186/s40536-026-00319-5

Keywords: rapid guessing, test-taking engagement, elementary school, large-scale assessment, response times, psychometrics, educational measurement, reading assessment, mathematics assessment, socioeconomic status, test validity, National Educational Panel Study

The research team, led by Leticia Sam of the University of Cape Coast in Ghana, analysed a total of 837 examination items: 550 multiple-choice questions and 287 essay or sub-question items drawn from officially published past papers of the West African Examinations Council. Their approach was deliberately quantitative and replicable. Two independent raters with expertise in financial accounting and prior familiarity with WAEC examination processes coded every item twice over, once for structural quality against established item-writing principles and once for cognitive demand using the Revised Bloom’s Taxonomy, the widely used hierarchy that ranks thinking skills from remembering and understanding up through analysing, evaluating and creating.

The reliability of that coding was statistically verified. The two raters worked separately without consultation, and their agreement was assessed with Cohen’s Kappa, a coefficient that measures consensus beyond chance. The resulting value of 0.70 exceeds the commonly recommended threshold of 0.60, indicating substantial agreement. Disagreements were then resolved through deliberative consensus, and the final dataset was analysed in SPSS using frequencies and percentages to map the distribution of flaws and cognitive levels across the eleven examination cycles.

On the structural side, the results were sobering. Of the 550 multiple-choice questions examined, 309 items, or 56.18 percent, violated at least one recognised item-writing principle, while only 241 items, or 43.82 percent, fully complied. The single most common defect was the unfocused or ambiguous stem, the sentence that frames the question itself, which accounted for 175 cases, or 56.63 percent of all identified flaws. In 2020 the problem peaked dramatically, with 19 of that year’s 21 flawed items, some 90.48 percent, marred by unclear or poorly structured stems. Ambiguous wording forces candidates to spend their mental effort interpreting the question rather than demonstrating their grasp of accounting concepts, which threatens the validity of the entire assessment.

The second most frequent flaw, accounting for 88 cases or 28.48 percent, was the failure to arrange answer options alphabetically or chronologically. Though it may sound trivial, inconsistent ordering can create unintended patterns that savvy test-takers exploit, allowing them to identify correct answers through superficial cues rather than genuine knowledge. Unequal option lengths, the classic giveaway in which the correct answer is noticeably longer than its distractors, accounted for 46 cases or 14.89 percent. This flaw was concentrated in the early years of the study, particularly 2014 and 2015, and vanished entirely in 2018 and 2020, suggesting that examiners gradually learned to balance their alternatives even as other problems persisted.

The year-by-year trajectory tells a story of intermittent progress rather than sustained reform. In 2014, a dismal 72 percent of multiple-choice items were flawed, the worst performance in the entire period. Compliance improved unevenly through the decade, reaching its high point in 2023, when 60 percent of items met acceptable standards. Yet the gains proved fragile: in 2024 the proportion of standard items collapsed back to just 38 percent. The researchers interpret this reversal as evidence that improvements in examination construction occurred sporadically rather than as an institutionalised practice, pointing to weaknesses in examiner training, item review procedures and moderation before administration.

The cognitive analysis was even more striking. Of the 837 items classified against the Revised Bloom’s Taxonomy, 756, or 90.3 percent, targeted lower-order cognitive skills: remembering, understanding and applying. Only 81 items, or 9.7 percent, demanded higher-order thinking. Breaking the numbers down further, 39.2 percent of all questions tested pure recall of facts such as the definitions of depreciation or ledger entries, 15.2 percent measured comprehension, and 36.0 percent assessed procedural application, such as preparing accounts or calculating profit. At the top of the hierarchy the picture thins out almost to nothing: just 5.7 percent of items required analysis, 2.4 percent demanded evaluative judgment, and a mere 1.6 percent, thirteen questions in eleven years, asked students to create or synthesise anything original.

This imbalance matters because the WASSCE does more than grade students; it shapes what teachers teach. High-stakes examinations exert a powerful backwash effect on classroom instruction, and when questions consistently reward memorisation, teachers rationally respond with rote pedagogy. The researchers argue that the pattern risks producing graduates who can execute routine calculations but struggle to interpret complex financial scenarios, evaluate alternative accounting methods, or exercise the professional judgment that tertiary study and the accounting profession demand. The finding also echoes a broader international literature: comparable studies have found that 89 percent of WASSCE Business Management items tested lower-order skills, that only 38.4 percent of veterinary examination questions in the United States required higher-order thinking, and that Finnish teacher-education entrance exams similarly leaned on recall and comprehension.

The study’s authors frame the problem as systemic rather than incidental. Standardised, easily markable items are attractive precisely because they allow consistent scoring across multiple West African countries, but that convenience comes at the cost of cognitive rigor. Their recommendations are concrete: WAEC should prioritise examiner training and structured item-writing workshops, implement rigorous moderation and quality assurance protocols that verify the balance of cognitive levels on every paper, and deliberately design questions that require students to analyse complex accounting scenarios, evaluate competing solutions and synthesise information. They further call for assessment literacy to be embedded in teacher education programmes, including training in item construction and the application of Bloom’s taxonomy.

The implications extend well beyond the examination hall. Because WASSCE results gate access to tertiary education and professional training, technical flaws and shallow cognitive demand in the instrument itself mean that certification may not faithfully represent what graduates actually know and can do, with consequences for employers, universities and the wider economy. The researchers acknowledge limitations: their analysis covered only Financial Accounting, relied on expert judgment in coding, and did not link item characteristics to actual student performance. But the core message stands. An examination that devotes barely one question in ten to higher-order thinking, and in which a majority of multiple-choice items carry technical defects, cannot fully certify the analytical competence that modern accounting demands. Whether WAEC can institutionalise the improvements that have so far appeared only in fits and starts will determine if the credential keeps pace with the profession it is meant to open.

Subject of Research: Quality and cognitive complexity of financial accounting examination questions in the West African Senior School Certificate Examination

Article Title: Assessment of financial accounting questions in the May/June West African Senior School Certificate Examination from 2014 to 2024

Article References: Sam, L., Agormedah, G. T., Kwarteng, S., Osei, M. A., Aidoo, H., Doe, R., & Sah, S. K. (2026). Assessment of financial accounting questions in the May/June West African Senior School Certificate Examination from 2014 to 2024. Discover Education, 5(1), Article 1081. https://doi.org/10.1007/s44217-026-02237-8

Image Credits: AI Generated

DOI: 10.1007/s44217-026-02237-8

Keywords: WASSCE, financial accounting, assessment, item-writing flaws, Bloom's Taxonomy, cognitive levels, Ghana, examination validity, multiple-choice questions, educational measurement, WAEC, higher-order thinking

The study, published in the journal Large-scale Assessments in Education, was conducted by Markéta Žáková and Tomáš Lintner of Masaryk University in the Czech Republic. Drawing on PISA 2018 data from more than 600,000 fifteen-year-old students across 75 education systems, the pair set out to answer a deceptively simple question: does it matter which operational definition of academic resilience a researcher uses? The answer, they found, is a resounding yes, with consequences that ripple through prevalence estimates, country rankings, and conclusions about which factors help disadvantaged students beat the odds.

Academic resilience combines two ingredients: adversity, usually measured by socioeconomic status, and positive adaptation, usually measured by achievement. But researchers disagree about how to stitch those ingredients together. One common approach simply identifies students who land in the bottom slice of the socioeconomic distribution and the top slice of achievement, for example the poorest quarter who score in the top quarter. A second approach instead computes what each student’s achievement should be, statistically speaking, given their family background, and flags students who substantially outperform that expectation, an approach known as the residual method. A third treats resilience as a process rather than an outcome, predicting achievement as a continuous variable within the disadvantaged group and never labeling anyone resilient at all.

Žáková and Lintner compared these approaches systematically. They estimated a top-achiever definition, a residual definition computed within each country, and a residual definition benchmarked against an international standard, each at three different threshold levels, 20, 25, and 33 percent, spanning the range most commonly used in the published literature. All told, this produced nine binary operationalizations plus the continuous-outcome model, yielding nearly a thousand country-level analyses. Each estimate was pooled across ten plausible values for achievement and twenty multiply imputed datasets, with standard errors that fully accounted for PISA’s complex two-stage sampling design.

The headline finding is stark. The cross-country mean prevalence of academically resilient students ranged from 7.5 percent under the strictest definition to 35.3 percent under the most inclusive residual benchmark, a nearly fivefold difference computed from identical data. Part of this spread is arithmetic, since looser thresholds mechanically admit more students, but the deeper problem emerges when countries are ranked. Rankings were reasonably stable across thresholds within a given definition, yet they diverged sharply across definitions. The correlation between rankings produced by the top-achiever approach and the residual approaches fell as low as 0.51, and between the two residual variants, which differ only in whether the statistical expectation is local or global, it dropped to between 0.32 and 0.47. A country celebrated as an equity champion under one definition can rank unremarkably under another, and the discrepancy is itself informative about what its disadvantaged students actually do well.

The authors illustrate why with a thought experiment grounded in their results. A lower-performing system with a steep socioeconomic gradient may rank poorly on the top-achiever definition, because few of its disadvantaged students reach absolute excellence, yet rank highly on the within-country residual definition, because modest local expectations are easy to exceed. That same system may then sink again on the internationally benchmarked residual definition, since beating a weak local bar is not the same as meeting a global standard. The researchers argue that resilience rankings should therefore be reported under multiple definitions side by side, as complementary views of equity rather than competing estimates of a single quantity, a practice the OECD itself briefly adopted nearly a decade ago.

What about the protective factors that resilience research is meant to uncover? Here the news is more reassuring at the global level and more troubling at the level of individual countries. When the authors pooled effects meta-analytically across all 75 systems, three motivational constructs drawn from expectancy-value theory, students’ self-perceived reading competence, their enjoyment of reading, and their attitude toward learning, were consistently and positively associated with resilience under every operationalization and threshold, and girls consistently outperformed boys. Read that result alone, and the choice of definition seems inconsequential.

The country-by-country picture tells a different story. When each education system was analyzed separately, as is standard in PISA-based research, findings were consistent across all nine binary specifications in only 20 percent of systems for gender, 40 percent for attitude toward learning, 69 percent for reading enjoyment, and 72 percent for self-perceived competence. Just two of the 75 systems produced fully consistent results for all four predictors. Threshold level alone flipped conclusions in somewhere between 3 and 37 percent of countries depending on the factor. Much of this inconsistency reflects statistical power, since stricter thresholds shrink samples and smaller-effect predictors, notably gender and attitude toward learning, are the least stable. But the practical consequence does not depend on the cause: a researcher studying one country under one definition could legitimately conclude that a factor matters there when a defensible alternative would say it does not.

The most conceptually striking result came from a complementary interaction analysis that requires no thresholds at all. By modeling whether the socioeconomic gradient in achievement flattens at higher levels of each factor, the authors tested what the term protective factor actually implies. The findings complicate comfortable assumptions: the gradient was flatter, not steeper, among students with higher self-perceived reading competence and stronger attitudes toward learning, but it was significantly steeper, not flatter, among students who enjoyed reading more. In other words, reading enjoyment, though positively associated with achievement on average, was linked to wider rather than narrower socioeconomic gaps, a pattern that no threshold-based definition could ever detect. These moderation effects also varied in direction across countries, appearing in only a minority of systems individually.

The authors close with four practical recommendations: match the operationalization to the research question, report sensitivity analyses at a minimum of two thresholds, present cross-country rankings under multiple definitions, and document or pre-register operationalization decisions so findings can be interpreted in light of the choices that produced them. They caution that their analysis is cross-sectional and cannot establish causation, that their predictors were individual-level only, and that PISA’s socioeconomic index has itself attracted methodological criticism. Still, the broader message is hard to escape. Academic resilience, as measured in large-scale assessments, is not a fixed quantity waiting to be counted but a construct whose observed properties depend partly on how it is defined, and both researchers and policymakers ignore that dependence at their peril.

Subject of Research: How operationalization and threshold choices affect estimates of academic resilience and its protective factors across 75 education systems using PISA 2018 data.

Article Title: Academic resilience in 75 education systems: how operationalization and threshold choices shape findings on protective factors

Article References: Žáková, M., & Lintner, T. (2026). Academic resilience in 75 education systems: how operationalization and threshold choices shape findings on protective factors. Large-scale Assessments in Education, 14(1), Article 44. https://doi.org/10.1186/s40536-026-00318-6

Image Credits: AI Generated

DOI: 10.1186/s40536-026-00318-6

Keywords: academic resilience, PISA 2018, socioeconomic status, educational equity, large-scale assessment, protective factors, operationalization, thresholds, student motivation, expectancy-value theory, educational measurement, cross-national comparison