What the scale is designed to measure
Researchers in Taiwan have developed a 21-item questionnaire intended to measure how college students understand, judge and seek information about human papillomavirus (HPV) prevention. In separate analyses that explored and then checked the questionnaire's structure, the scale showed high internal consistency and was associated with another questionnaire, the MMHLQ. The findings support the tool as a measurement instrument in this sample, but the study assessed scale properties rather than whether using it improves vaccination, screening, behavior or HPV-related health outcomes.
That distinction is important. This was a cross-sectional study, so it captured responses at one point rather than following students over time. No intervention was assigned and there was no control group. The evidence therefore concerns the questionnaire's construction and measurement properties, not the causal effectiveness of HPV-prevention education or clinical interventions.
The tool was organized around two related abilities: comprehension and evaluation of general information, and seeking information from others. In plain terms, its questions were meant to cover both making sense of HPV-prevention information and looking for help or advice.
From interviews to factor testing
The questions were built in stages. Qualitative interviews with 26 college students produced an initial pool of 49 items. Five experts then assessed whether the questions were relevant, accurate and clearly worded.
That review removed five items, leaving 44. The researchers used item- and scale-level content-validity checks, with reported minimum thresholds of .78 for the item index and .90 for the average scale index; the remaining items ranged from .80 to 1.00 on the item index, and the scale average was .90.
A pilot group of 30 students completed the 44-item version on a four-point scale running from 1, very difficult, to 4, very easy. Item analysis removed seven questions. Corrected item-total correlations ranged from .54 to .83, while reported factor loadings ranged from .22 to 1.08; tests comparing high- and low-scoring respondents were reported as p>.05.
The first factor analysis used 37 items and the 150-person EFA sample. The data met the study's reported checks for factor analysis, including a KMO value of .93 and Bartlett's chi-square of 5,608.231 with p<.001. Two factors were selected using eigenvalues above 1 and the scree plot.
Two dimensions, then a shorter final scale
The first factor, comprehension and evaluation of general information, contained 19 items and accounted for 51.58% of the variance. The second, seeking information from others, contained nine items and accounted for 8.72%. Together, the factors explained 60.30% of the total variance, with item loadings ranging from .459 to .996.
Researchers then tested the proposed structure in the separate CFA sample of 222 students. The confirmatory analysis began with 28 EFA-derived items and used a second-order model, placing the two first-order dimensions under one overall HPV-prevention health-literacy factor.
The reported fit measures were broadly favorable: the chi-square-to-degrees-of-freedom ratio was 2.21, below the reported comparison value of 3; CFI was .92, TLI .91, RMSEA .07 and SRMR .02. GFI was .81. The overall chi-square was 606.190 with p<.001, so the researchers noted a significant result on that test alongside the other fit measures.
Other checks supported the distinction between the two dimensions. Composite-reliability values, a measure of how consistently items represent each dimension, were .96 and .93. Average-variance-extracted values, which indicate how much of the item information each dimension captures, were .56 and .64; they exceeded the squared-correlation criterion used to assess whether the dimensions remained distinguishable. The factor correlation was .66 with p<.001.
After item deletion among the CFA participants, the final questionnaire contained 21 items. Overall Cronbach's alpha, a measure of internal consistency, was .959; the two subscale values were .955 and .939. In practical terms, the items tended to produce coherent scores in this sample. The study did not assess test-retest reliability, so it cannot show whether the same people would receive stable scores at a later time.
The questionnaire also showed concurrent validity with the MMHLQ: scores were correlated at .68, with p<.001. That result indicates that the new scale moved in step with the comparison measure in this sample. It does not establish that the scale predicts or causes a change in HPV-prevention behavior.
Useful evidence, narrow reach
The sample sets a clear boundary around the finding. Participants came from two universities in northern Taiwan and one in central Taiwan. Recruitment used convenience sampling supplemented by snowball sampling, and the study applied a specific eligibility rule concerning age and prior HPV-information exposure. Those choices leave the scale's performance in other universities, regions and student groups uncertain.
The strongest next tests would be practical ones: repeat the scale with the same participants to assess test-retest reliability, validate it in broader random samples, examine it with student groups not represented by the reported eligibility criteria, and follow scores against educational or behavioral outcomes over time. Until then, the HPV-HLS is best understood as a 21-item measurement tool with evidence of internal coherence and relationships with other measures, not as evidence that a questionnaire itself improves prevention.
Paper data and sources
Original title: Development and Psychometric Validation of the Human Papillomavirus Prevention Health Literacy Scale for College Students in Taiwan.
Authors: Ching-Yi Lai, Shu-Chen Yang, Chieh-Yu Liu et al.
Journal/Repository: The journal of nursing research : JNR
Status: Peer-reviewed
First online: 2026-10-01
DOI: 10.1097/jnr.0000000000000767
Original paper · Full text