Danish Stress Questionnaire Passes Its Teenage Test, Study Finds

When researchers want to know how much stress a person is under, one of the most popular tools in the world is a short, ten-item questionnaire called the Perceived Stress Scale, or PSS-10. Developed decades ago and translated into dozens of languages, it asks respondents how often in the past month they felt things like uncontrollable, overwhelmed, or on top of things. But a questionnaire is only useful if it has been rigorously tested in the population where it will be applied. In a study published in BMC Psychiatry, a team of Danish researchers led by Pernille Bach Steen of Gødstrup Hospital and Aarhus University has now provided the most comprehensive evaluation to date of the Danish version of the PSS-10 in adolescents, and their results suggest the instrument stands up well to scrutiny in this age group.

The motivation behind the study is straightforward. Adolescence and early adulthood are periods of heightened vulnerability to mental health problems, and perceived stress is both a symptom and a risk factor for conditions such as depression and anxiety. In Denmark, the PSS-10 has been used in adult populations, but its measurement properties had never been formally assessed in teenagers aged 15 to 20. Without such validation, researchers and clinicians cannot be certain whether the Danish scale actually measures the same construct in young people, whether its scores are stable over time, or whether a given score means the same thing in a 16-year-old as it does in a 40-year-old. The new study set out to answer these questions using a battery of classical psychometric techniques.

The design was ambitious in scale but careful in structure. The team drew a random sample of 5,699 adolescents aged 15 to 20 years and invited them to participate. Of these, 667 completed the PSS-10 in the first round of data collection, forming the test sample. Roughly one week later, 182 of these participants completed the questionnaire a second time, allowing the researchers to assess test-retest reliability, the degree to which scores remain stable when nothing has changed. To examine whether the questionnaire truly relates to clinical reality, a subsample of 38 participants underwent a Schedules for Clinical Assessment in Neuropsychiatry, or SCAN, interview, a gold-standard diagnostic instrument administered by trained psychologists that establishes the presence or absence of psychiatric diagnoses. Data collection was registered and approved by the regional register of research projects on 3 January 2023, and all participants provided informed consent.

The first question the researchers tackled was structural validity: does the ten-item scale measure one underlying construct, or two? This matters because different users score the PSS-10 in different ways. Some researchers compute a single total score, while others, following the original authors, split the items into two subscales, one capturing perceived helplessness and the other perceived self-efficacy, that is, the reverse-worded items asking how often respondents felt in control and on top of things. Using confirmatory factor analysis, a statistical technique that tests how well observed data fit a hypothesized factor structure, the team compared one-factor and two-factor models. The two-factor model showed the best statistical fit, which mirrors findings from many other countries and languages. However, the researchers concluded that the overall pattern of findings supported the use of the total PSS-10 score, a pragmatic conclusion that simplifies interpretation and comparison across studies.

Reliability, the question of how free a measurement is from random error, came out impressively well on both fronts. Internal consistency, which reflects how strongly the ten items hang together, was evaluated using Cronbach’s alpha and reached 0.90, a value conventionally regarded as excellent and one that indicates the items are tapping a coherent construct without being so redundant as to waste respondents’ time. Test-retest reliability over the one-week interval was assessed with the intraclass correlation coefficient, yielding an ICC(2,1) of 0.89, which is high and suggests that, in the absence of real change, an adolescent’s score is remarkably stable across repeat administration. For a self-report scale filled out by teenagers, whose moods and circumstances can shift quickly, this stability is a meaningful endorsement of the instrument.

Perhaps the most practically important number in the paper is the standard error of measurement, which the researchers calculated at 2.65 points. From this they derived the minimal detectable change, the smallest difference between two measurements that can be considered real rather than noise, at 7.34 points. To put that in perspective, the PSS-10 ranges from 0 to 40, so a change of 7.34 corresponds to roughly an 18 percent shift in score. This figure matters enormously for anyone planning to use the scale in intervention studies or clinical monitoring: if a school program or a therapy is evaluated by before-and-after PSS-10 scores, a change smaller than about seven points cannot be confidently interpreted as genuine improvement or deterioration. Providing this measurement-precision benchmark for Danish adolescents gives future studies a principled yardstick rather than a guess.

The study then moved from measurement properties to meaning. Construct validity asks whether scores on the scale correlate with other measures in ways that theory predicts. The PSS-10 scores showed strong correlations with psychiatric symptom scores, supporting the idea that the Danish scale is indeed capturing the psychological distress it is supposed to measure rather than something unrelated. Criterion validity goes one step further and asks whether the scale can stand in for an external standard. Using the subsample of 38 adolescents who completed the SCAN interview, the researchers employed receiver operating characteristic analysis, a technique that plots sensitivity against specificity across all possible cutoff points to summarize discriminative ability in a single number, the area under the curve. The result was an AUC of 0.88, which falls in the good range and indicates that the self-administered questionnaire discriminates between adolescents with and without clinically assessed psychiatric diagnoses considerably better than chance.

That ROC analysis also allowed the team to explore a potential clinical threshold. In the exploratory analyses, a score of 24 or higher emerged as the cutoff that optimally separated adolescents with a psychiatric diagnosis from those without one. If confirmed, such a threshold could eventually help schools, general practitioners, and mental health services identify young people whose self-reported stress warrants a closer clinical look, using a tool that takes only a few minutes to complete. The authors are appropriately cautious on this point, however. The clinical subsample was small, and they explicitly state that larger validation studies are needed before the threshold of 24 can be recommended for practical use. A cutoff derived from 38 participants is a hypothesis to be tested, not a rule to be applied.

The broader significance of the study lies in what it makes possible. Denmark has rich registry data, longitudinal cohort studies, and a strong tradition of adolescent mental health research, but until now the Danish PSS-10 in teenagers rested on an assumption of validity rather than evidence. That assumption can now be replaced with documented estimates of structural validity, internal consistency, test-retest reliability, measurement precision, and construct and criterion validity, all favorable. Researchers designing school-based stress interventions, national surveys, or studies linking perceived stress to later psychiatric outcomes can proceed with confidence in the scale’s total score. The two-factor structure that fit best statistically, combined with the recommendation to use the total score, also reflects a now-common finding in PSS research: the split between positively and negatively worded items often produces a method-driven second factor, but the full scale remains the most robust unit of analysis. For clinicians and researchers elsewhere in Scandinavia and beyond, the study also offers a template, showing how a random population sample, a retest arm, and a clinical interview subsample can together produce a complete psychometric picture. The work was funded by Sygeforsikringen Danmark, which had no role in study design, analysis, or interpretation, and it is published open access. As adolescent stress continues to rise across much of the world, having a validated, translated, and quantitatively characterized tool is a small but consequential step toward measuring, understanding, and ultimately reducing the burden on young minds.

Subject of Research: Psychometric validation of the Danish PSS-10 questionnaire for measuring perceived stress in adolescents

Article Title: Psychometric properties of the Danish PSS-10 in adolescents

Article References: Psychometric properties of the Danish PSS-10 in adolescents. (n.d.). https://doi.org/10.1186/s12888-026-08601-0

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08601-0

Keywords: PSS-10, perceived stress, adolescents, psychometrics, validation, reliability, Cronbach's alpha, test-retest reliability, confirmatory factor analysis, ROC analysis, BMC Psychiatry, Denmark