The design of the test rests on a developmental vocabulary list rather than the frequency-based word lists that dominate second-language testing. This choice matters because Chinese literacy acquisition does not unfold the way alphabetic literacy does. Chinese children must learn thousands of distinct characters, each composed of strokes and radicals, and words are typically written with combinations of those characters. A word list built specifically around what children are expected to learn at each stage of compulsory education gives the test a developmental logic: the words sampled for a Grade 1 subtest are those a first grader should plausibly be mastering, while the words in the Grade 9 subtest reflect the far larger lexicon expected of a student about to enter senior secondary school.

Structurally, the C-VLT consists of four subtests, each matched to a different band of grade levels. This staged architecture allows teachers and researchers to place a student’s receptive vocabulary knowledge, meaning the words a child can recognize and understand when encountered in print, along a developmental continuum. Receptive knowledge is deliberately the target here. Distinguishing what learners can recognize from what they can actively produce has long been a central theme in vocabulary research, and the authors designed the C-VLT to interpret scores specifically as indicators of recognition-based receptive knowledge and stage-based vocabulary mastery, not as measures of writing or speaking ability.

The empirical scale of the validation study is one of its most striking features. Using probability proportional to size sampling, a survey method that gives larger schools a proportionally greater chance of selection so the sample mirrors the population, the team recruited 16 schools across four counties, spanning both urban and rural settings. In total, 10,460 students sat the test. Samples of this magnitude are rare in vocabulary assessment research, where validation studies often rely on a few hundred participants, and they give the psychometric findings unusual stability.

The analysis combined two complementary statistical traditions. Classical test theory, the older framework, evaluates tests through indices such as reliability coefficients and item difficulty proportions computed directly on observed scores. Item response theory, the more modern approach, models the probability that a test-taker of a given ability answers a given item correctly, producing estimates of each item’s difficulty and discrimination along a common ability scale. When the researchers applied both frameworks to the C-VLT data, the results pointed in the same direction: the items showed reasonable difficulty and discrimination, and all four subtests demonstrated high reliability in the classical sense.

A particularly important finding concerns fairness. The team tested every item for differential item functioning, a statistical procedure that checks whether an item behaves differently for test-takers of equal ability who belong to different groups, for example students in different regions or school types. An item with differential item functioning is biased: it advantages one group over another for reasons unrelated to the construct being measured. In the C-VLT, no items showed such effects, which the authors interpret as evidence of high item quality and satisfactory test fairness across the diverse sample of schools.

Measurement precision was evaluated through item response theory information curves and their associated standard errors. In item response theory, the information function describes how accurately a test measures ability at different points on the ability scale, and the standard error of measurement is inversely related to that information. For the C-VLT, the information curves and standard errors indicated good measurement precision for most students, meaning the test distinguishes ability differences reliably across the broad middle of the distribution rather than only at its extremes. That property is essential for a test intended to be used diagnostically with ordinary classrooms rather than only with exceptional or struggling readers.

Validity evidence came from three converging sources. Evidence based on test content showed that the items genuinely sampled the vocabulary children are expected to learn at each educational stage, drawing on the developmental word list that anchors the instrument. Evidence based on internal structure confirmed that the relationships among items and subtests matched the intended design. Most compellingly, the test showed theoretically expected relations to educational stage and word frequency: item difficulty generally increased as word frequency decreased and as educational stage increased. In other words, rarer words were harder, and words targeted at later grades were harder than words targeted at earlier grades, exactly the pattern a valid developmental vocabulary measure should display.

The significance of this work extends beyond Chinese classrooms. Vocabulary size is one of the strongest known predictors of reading comprehension, listening comprehension, and academic achievement, and meta-analytic work on reading in Chinese has confirmed that vocabulary plays a central role alongside decoding in that writing system as well. Yet until now, researchers studying native Chinese-speaking children have lacked a levels-based instrument comparable to the Vocabulary Levels Test tradition in English, which partitions vocabulary knowledge into frequency or stage bands and estimates mastery within each band. Existing Chinese tools, such as quick character-based proficiency checks or tests built for second-language learners of Chinese, were not designed to capture the stage-by-stage growth of a child’s first-language lexicon.

The C-VLT fills that gap with unusual transparency. The test data, materials, and all source code have been deposited in an openly accessible repository on the Open Science Framework, allowing other researchers to scrutinize the psychometric analyses, adapt the instrument, or build new versions. The study was approved by the institutional review board of the Collaborative Innovation Center of Assessment for Basic Education Quality at Beijing Normal University, and all participants gave oral informed consent. Funded through China’s STI 2030 Major Projects program and university research grants, the project demonstrates how large-scale sampling, dual psychometric frameworks, and open materials can be combined to produce an assessment whose scores are not only reliable but genuinely interpretable as a map of how vocabulary knowledge grows, year by year, across the nine years of compulsory schooling.

Subject of Research: Development and validation of a Chinese Vocabulary Levels Test for assessing receptive vocabulary knowledge in primary and secondary school learners

Article Title: C-VLT: A reliable Chinese Vocabulary Levels Test for primary and secondary school learners

Article References: Zeng, B., Jiang, Y., & Wen, H. (2026). C-VLT: A reliable Chinese Vocabulary Levels Test for primary and secondary school learners. Behavior Research Methods, 58(11), Article 312. https://doi.org/10.3758/s13428-026-03166-y

Image Credits: AI Generated

DOI: 10.3758/s13428-026-03166-y

Keywords: vocabulary assessment, Chinese language, psychometrics, item response theory, classical test theory, receptive vocabulary, test validity, language education, literacy development, differential item functioning, Behavior Research Methods, schoolchildren

The research team, led by Alexandra Salinas-Palma of the Universidad Estatal de Milagro and Rodrigo Moreta-Herrera of the Pontificia Universidad Católica del Ecuador, together with colleagues from several Ecuadorian and Peruvian institutions, recruited 485 adults ranging in age from 18 to 53 years. The average age of participants was about 24 years, with a standard deviation of roughly 7.6, indicating a sample skewed toward young adults. Women made up 55.1 percent of the sample and men 44.9 percent, a split close enough to balanced to allow meaningful comparisons between the two groups. The study received ethical approval from the Institutional Review Committee of the Pontificia Universidad Católica del Ecuador Sede Ambato, and participants provided written informed consent before taking part.

What makes the study methodologically interesting is that the researchers did not rely on a single statistical approach. Instead, they combined two long-standing traditions in psychometrics, the science of psychological measurement. The first is Classical Test Theory, the older framework that treats an observed score as the sum of a true score and random error, and that underpins familiar concepts such as reliability coefficients and factor analysis. The second is Item Response Theory, a more modern framework that models the probability of a person responding in a particular way to an individual item as a function of that person’s underlying level of the trait being measured. By applying both frameworks to the same data, the team could check whether conclusions about the scale held up regardless of the analytical lens used.

Under Classical Test Theory, the first task was to confirm the scale’s internal structure. The General Self-Efficacy Scale is designed to tap a single underlying dimension, and confirmatory factor analysis supported this unifactorial structure in the Ecuadorian sample. That finding matters because a scale can only be meaningfully compared across groups if it measures one coherent construct in the same way in each group. The researchers then moved to the heart of the study: testing measurement equivalence, also known as measurement invariance, across gender. This is a hierarchical procedure in which increasingly strict constraints are placed on the model. First, researchers check whether the same factor structure holds in both groups, known as configural invariance. Next, they test whether the items load equally strongly on the factor in both groups, known as metric invariance. Finally, they test whether the thresholds, the points on the latent continuum at which respondents shift from one response category to the next, are equal across groups, known as scalar invariance.

The results showed that the scale achieved measurement invariance at the threshold, or scalar, level across gender. In practical terms, this is the level of invariance that psychologists generally consider the minimum requirement for comparing group means. When scalar invariance holds, a man and a woman with the same underlying level of self-efficacy are expected to produce the same observed score on the scale, so any difference in average scores between the groups can be interpreted as a real difference in the construct rather than a quirk of the instrument. Consistent with this, the researchers found no statistically significant differences in the latent means of the male and female groups, with p values above the conventional .05 threshold. In other words, Ecuadorian men and women in this sample did not differ in their underlying general self-efficacy as measured by the scale.

The Item Response Theory analysis added a second, finer-grained layer of evidence. Using an approach consistent with the graded response model, a model developed for items with ordered response categories, the team estimated two key parameters for each item. The first is the discrimination parameter, usually denoted a, which describes how well an item distinguishes between people with slightly different levels of self-efficacy. Items with high discrimination are sharply informative: they separate high scorers from low scorers with precision. The second is the difficulty or location parameter, denoted b, which indicates where along the latent continuum an item provides the most information, that is, how much self-efficacy a person needs to endorse the item’s higher response categories. The researchers reported that both the a and b parameters were adequate across the items, meaning the items functioned as intended, discriminating well and covering a useful range of the trait.

The most direct test of gender bias, however, came from the analysis of differential item functioning, commonly abbreviated as DIF. Differential item functioning occurs when people from different groups but with exactly the same level of the underlying trait have different probabilities of responding in a particular way to a specific item. An item showing DIF against women, for example, would be one that women find harder to endorse than men with identical self-efficacy, perhaps because of how the item is worded or because of culturally gendered interpretations of its content. DIF is a serious threat to fair measurement because it can inflate or deflate scores for one group without any real difference in the construct. Detection methods range from logistic regression procedures, in which the grouping variable is tested as a predictor of item responses after controlling for the trait level, to more complex model-based approaches. In this study, the analysis found no evidence of gender-based DIF in any of the scale’s items, providing item-level confirmation that the questionnaire treats men and women equitably.

The findings carry practical weight well beyond Ecuador. The General Self-Efficacy Scale is rooted in Albert Bandura’s theory of self-efficacy, the belief in one’s own capacity to organize and execute the actions needed to manage situations, which has been linked to outcomes ranging from academic achievement and coping with illness to workplace performance and psychological well-being. Researchers frequently compare self-efficacy between men and women, for instance in studies of the so-called confidence gap in academic and professional settings, or in investigations of how self-efficacy relates to anxiety, loneliness, and coping. If the instrument itself were biased, those comparisons would be compromised. The Ecuadorian results suggest that, at least in this population, such comparisons can be made with confidence: the scale is reliable, its single-factor structure is confirmed, and gender does not significantly affect its measurement properties.

The study also contributes to a broader conversation about measurement fairness in psychology, a field that has faced increasing scrutiny over whether instruments developed in one cultural context transfer cleanly to others. Previous validations of the General Self-Efficacy Scale in Spanish-speaking Latin American samples, including work in Chile, Peru, and earlier Ecuadorian studies by some of the same authors, had supported its psychometric soundness, but the question of gender-based measurement variability had not been examined in this population with the combined CTT and IRT toolkit. The instrumental design of the new study, which focuses on evaluating the properties of the measurement instrument itself rather than testing a substantive psychological hypothesis, reflects a growing emphasis in Latin American psychology on establishing rigorous local evidence before instruments are used in research and applied settings.

Some caveats are worth noting. The sample of 485 adults was relatively young, with a mean age of about 24, so the findings may not generalize to older adults, and the data were drawn from a single country. The authors note that the data are available to interested parties for scientific and non-commercial purposes upon request. Still, the convergence of evidence is striking: a confirmed unifactorial structure, scalar measurement invariance across gender, equivalent latent means, adequate item discrimination and difficulty parameters, and no differential item functioning. For researchers in Ecuador and comparable Spanish-speaking contexts, the message is clear. The General Self-Efficacy Scale can be used to compare men and women without worrying that the yardstick itself is bent, and the combined Classical Test Theory and Item Response Theory approach demonstrated here offers a template for how other widely used psychological instruments can be audited for fairness in new populations.

Subject of Research: Psychometric evaluation of gender-based measurement invariance in the General Self-Efficacy Scale among Ecuadorian adults

Article Title: Is Gender a Source of Measurement Variability in the General Self-Efficacy Scale? Psychometric Analysis in Ecuadorian Adults

Article References: Salinas-Palma, A., Moreta-Herrera, R., Lascano-Arias, G., Mena-Freire, M., Mascialino, G., Almeida-Márquez, L., & Caycho-Rodríguez, T. (2025). Is Gender a Source of Measurement Variability in the General Self-Efficacy Scale? Psychometric Analysis in Ecuadorian Adults. Trends in Psychology, 34(1), 375-393. https://doi.org/10.1007/s43076-025-00495-3

Image Credits: AI Generated

DOI: 10.1007/s43076-025-00495-3

Keywords: self-efficacy, psychometrics, measurement invariance, differential item functioning, item response theory, classical test theory, gender differences, General Self-Efficacy Scale, Ecuador, graded response model, factor analysis, psychological assessment