DASS-21 Reliability and Factor Structure in Pakistani University Students

The DASS-21 showed good internal consistency and an adequately fitting factor structure among 602 Pakistani university students, supporting its usefulness for measuring depression, anxiety, and stress symptoms in this population. However, the very strong correlations among its latent depression, anxiety, and stress factors suggest an important qualification: the three subscales may be statistically distinguishable while still capturing a substantial amount of shared psychological distress.

These findings come from our 2026 BMC Psychology study of medical and non-medical undergraduate students in Lahore, Pakistan. The study went beyond reporting symptom prevalence by examining the psychometric performance of the Depression Anxiety Stress Scales-21 alongside the PHQ-9. That distinction matters because a questionnaire should not be assumed to work equally well in every cultural, educational, or linguistic setting simply because it is widely used internationally.

What Is the DASS-21?

The Depression Anxiety Stress Scales-21, usually abbreviated as DASS-21, is a 21-item self-report questionnaire designed to measure three related dimensions of negative emotional experience: depression, anxiety, and stress. It is the shorter version of the original 42-item DASS developed by Peter F. Lovibond and Sydney H. Lovibond.

The instrument contains seven items for each of its three domains. Respondents indicate how much each statement applied to them during the recent period specified by the questionnaire. The three resulting scores are intended to characterize symptom severity rather than establish psychiatric diagnoses.

According to the University of New South Wales DASS resource, the scales were developed to measure related but theoretically distinguishable negative emotional states. Depression emphasizes features such as low positive affect and hopelessness, anxiety emphasizes physiological arousal and anxious experience, and stress reflects persistent tension, irritability, and difficulty relaxing.

This structure makes the DASS-21 attractive for university mental-health research. A single brief questionnaire can provide information about several dimensions of psychological distress without requiring three separate instruments. But convenience does not automatically establish measurement quality. Researchers still need evidence that the items behave coherently and that the proposed dimensions can be recovered statistically in the population being studied.

Why DASS-21 Reliability in Pakistani Students Needs to Be Tested

Psychological questionnaires are often developed and initially validated in particular countries, languages, and demographic groups. Their later use in other populations introduces questions about cultural interpretation, educational context, response patterns, and the way different emotional concepts overlap.

For Pakistani university students, these questions are especially relevant because the DASS-21 is widely practical for student research, yet evidence from one setting cannot simply be transferred to another. An item that functions well in an Australian validation sample, for example, may not have precisely the same relationship with its intended psychological construct among students in Lahore.

Psychometric evaluation therefore asks questions such as:

• Do items intended to measure the same domain produce internally consistent responses?
• Does the proposed depression-anxiety-stress structure fit the observed student data?
• Are the three dimensions sufficiently distinct to justify interpreting them separately?
• Does the scale operate comparably across relevant groups, such as medical and non-medical students?

Our study addressed several of these issues in a sample specifically drawn from Pakistani universities.

The Pakistani University Student Study

The study included 602 undergraduate students from universities in Lahore, Pakistan. Of these, 424 were medical students and 178 were studying non-medical disciplines. Participants completed the DASS-21 and the Patient Health Questionnaire-9, or PHQ-9.

The analyses were broader than a conventional prevalence survey. They included descriptive comparisons, multivariable regression, reliability assessment, confirmatory factor analysis, measurement-invariance testing, item response theory, and symptom-network analysis. The complete published study, Psychometric properties and symptom profiles of the PHQ-9 and DASS-21 among medical and non-medical university students, provides the full methodological and statistical context.

For understanding DASS-21 measurement quality, three findings are particularly important: internal consistency was good, confirmatory factor analysis produced acceptable-to-good fit indices, and the latent factors were extremely strongly correlated.

DASS-21 Reliability Was Good

Reliability describes how consistently a set of items measures a construct. One common indicator is Cronbach’s alpha, which evaluates the degree to which items belonging to a scale tend to vary together.

In the Pakistani study, Cronbach’s alpha coefficients across the evaluated scales ranged from 0.82 to 0.88. These values indicate good internal consistency in the sample. In practical terms, respondents who endorsed one symptom within a domain tended to respond coherently to other items intended to measure related symptoms.

This is reassuring for researchers considering the DASS-21 for Pakistani university samples. Very weak internal consistency would suggest that the items were not functioning together as expected. That was not the pattern observed here.

At the same time, reliability should not be mistaken for validity. A collection of items can be highly correlated without necessarily measuring several clearly distinct concepts. Cronbach’s alpha answers a relatively narrow question about consistency. It does not, by itself, prove that depression, anxiety, and stress form three separable psychological dimensions.

DASS-21 Factor Structure: What the CFA Showed

Confirmatory factor analysis, or CFA, is particularly useful when a questionnaire has a predefined theoretical structure. Rather than exploring an unlimited number of possible patterns, CFA evaluates how well a hypothesized measurement model reproduces relationships observed among questionnaire items.

For the DASS-21, the central question is whether responses are compatible with the proposed organization of items into depression, anxiety, and stress dimensions.

In our Pakistani university sample, the reported CFA fit statistics were:

• Comparative Fit Index (CFI): 0.954
• Tucker-Lewis Index (TLI): 0.948
• Root Mean Square Error of Approximation (RMSEA): 0.052

Taken together, these values indicate that the tested DASS-21 measurement model represented the observed response patterns reasonably well. The results therefore provide support for the instrument’s intended factor structure in this sample.

It is better, however, to interpret model-fit indices collectively rather than treating a single numerical threshold as a pass-or-fail test. CFA fit depends on the model, estimator, data characteristics, sample size, and several other assumptions. The important conclusion here is not that the questionnaire achieved a theoretically perfect structure, but that the proposed model showed adequate empirical compatibility with the responses of the participating students.

The Most Important Finding: The Factors Were Highly Correlated

The more challenging psychometric result emerged when the relationships among the latent DASS-21 factors were examined.

The estimated correlation between the Depression and Stress factors was 0.939, while the correlation between Anxiety and Stress reached 0.949. Correlations this high mean that although the model can represent depression, anxiety, and stress as separate factors, those factors shared a very large amount of variance in this sample.

This creates an important distinction between factorial fit and discriminant validity. A three-factor model can fit data adequately while its factors remain so highly correlated that their practical separation deserves scrutiny.

Consider two different questions. First: can the item responses be represented using the expected DASS-21 factor arrangement? The CFA results support that interpretation. Second: do the resulting depression, anxiety, and stress scores capture strongly independent psychological constructs in these students? The very high latent correlations make that conclusion less certain.

This does not mean the DASS-21 failed. Depression, anxiety, and stress are genuinely related experiences, and substantial overlap is expected. University students experiencing marked distress may report low mood, anxiety symptoms, tension, irritability, sleep difficulties, and impaired concentration simultaneously. Statistical overlap can therefore reflect both measurement characteristics and real clinical or psychological comorbidity.

Nevertheless, near-unity factor correlations should discourage overly rigid interpretations in which modest differences among DASS-21 subscale scores are treated as evidence for completely separate underlying conditions.

Reliability and Factor Distinctiveness Are Different Questions

One of the most useful lessons from these findings is that a scale can be reliable without showing strong separation among every proposed dimension.

Good Cronbach’s alpha values tell us that groups of DASS-21 items behave consistently. The CFA fit indices tell us that the proposed measurement structure offers a reasonable description of the data. The factor correlations add another layer by asking whether the latent constructs are empirically distinct from one another.

Researchers should therefore avoid summarizing a psychometric evaluation with a statement such as “the DASS-21 is valid because alpha was above 0.80.” Measurement quality is multidimensional. Internal consistency, structural validity, convergent validity, discriminant validity, measurement invariance, responsiveness, and other properties answer different questions.

Measurement Invariance Across Medical and Non-Medical Students

Another useful result was support for measurement invariance across academic discipline. Measurement invariance evaluates whether a questionnaire measures constructs comparably across groups.

This issue is crucial when researchers want to compare medical and non-medical students. Suppose one group interprets particular DASS-21 items differently from another. An observed difference in total or subscale scores could then reflect differences in measurement rather than genuine differences in symptom burden.

Support for invariance strengthens the case for making comparisons between the medical and non-medical groups in this dataset because the scale did not show evidence, within the reported analysis, that its measurement properties fundamentally changed according to discipline.

This is particularly relevant because the study found higher adjusted depression and anxiety scores among non-medical students compared with medical students, while the adjusted stress difference did not reach statistical significance. Measurement-invariance evidence makes those group comparisons more interpretable, although it cannot eliminate every possible source of bias in an observational study.

What Do the Results Mean for Pakistani Researchers?

For researchers studying university mental health in Pakistan, the findings provide several practical lessons.

1. The DASS-21 appears suitable for research use in similar student populations

The combination of good internal consistency and adequate CFA fit supports continued use of the instrument for studying negative emotional symptoms among Pakistani university students, particularly in settings reasonably comparable with the Lahore sample.

2. Subscale scores should still be interpreted carefully

Researchers can report depression, anxiety, and stress scores separately, but they should recognize their substantial overlap. Analyses that describe them as completely independent outcomes may exaggerate distinctions that were not strongly evident at the latent level.

3. A general-distress component may deserve attention

The high factor correlations are compatible with the possibility that a strong common negative-affect or general-distress component contributes to responses across the DASS-21. That interpretation should be tested formally rather than assumed, for example through alternative factor models in future studies.

4. Local psychometric evidence matters

Researchers should cite evidence relevant to the population they are studying instead of relying exclusively on the original development literature. The Lahore data provide a useful Pakistani reference point, but replication across provinces, universities, languages, and age groups is still valuable.

What the Findings Do Not Prove

The study should not be interpreted as establishing universal DASS-21 validity for every Pakistani population. The participants were undergraduate university students recruited in Lahore, and psychometric performance can vary according to age, language, clinical status, education, location, and administration conditions.

The cross-sectional design also means that the study did not establish properties requiring repeated measurements, such as test-retest reliability or responsiveness to changes over time.

Likewise, good internal consistency does not establish diagnostic accuracy. The DASS-21 measures symptom dimensions; it is not a substitute for a structured psychiatric assessment, and elevated scores should not automatically be translated into diagnoses of major depressive disorder, an anxiety disorder, or another psychiatric condition.

Why High Depression-Anxiety-Stress Correlations Are Plausible

Depressive, anxious, and stress-related symptoms frequently occur together. Conceptually, the DASS was designed precisely because these negative emotional states are related but not identical. In real student populations, however, common pressures can produce symptoms spanning several dimensions at the same time.

A student experiencing prolonged academic, financial, family, or social strain might simultaneously report difficulty relaxing, increased nervous arousal, reduced enjoyment, low motivation, and feelings of hopelessness. A questionnaire can assign those experiences to different subscales while their underlying causes and subjective experiences remain strongly connected.

High latent correlations are therefore not automatically evidence of a poorly designed questionnaire. They do, however, tell researchers something important about the structure of the data: distinctions among the three dimensions may be less pronounced than the subscale labels alone imply.

Directions for Future DASS-21 Research in Pakistan

The current findings create several opportunities for further psychometric research.

First, independent studies could replicate the factor analysis across universities in different Pakistani provinces. A broader geographic sample would help determine whether the observed structure generalizes beyond Lahore.

Second, studies could evaluate Urdu and other regional-language versions when applicable. Translation involves more than replacing words. Researchers should assess semantic equivalence, cultural interpretation, item functioning, and factor structure after translation.

Third, alternative structural models could be compared directly. For example, researchers may examine whether models incorporating a broader general-distress factor better explain the strong correlations among depression, anxiety, and stress dimensions. Such work should be theoretically motivated and preferably tested in independent datasets rather than selected solely because it produces better fit statistics.

Fourth, measurement invariance can be expanded beyond academic discipline. Gender, language, institution type, socioeconomic background, and clinical versus non-clinical status may all be relevant grouping variables when sample sizes are adequate.

Finally, longitudinal research could investigate test-retest reliability, sensitivity to genuine changes in psychological state, and whether DASS-21 scores predict meaningful future outcomes. These questions cannot be answered by a single cross-sectional psychometric analysis.

How Should Universities Use the DASS-21?

Universities may find the DASS-21 useful for research, needs assessment, or structured monitoring when appropriately implemented. Its brevity makes it feasible for large student surveys, and the Pakistani findings support its internal consistency and structural performance in a university context.

However, questionnaire administration should be accompanied by a clear plan for interpretation. A high score represents elevated self-reported symptoms, not a definitive psychiatric diagnosis. Institutional screening programs also need appropriate pathways for students who report substantial distress or request help.

The instrument should therefore be viewed as one source of information within a broader mental-health framework rather than a stand-alone decision-making system.

The Bottom Line

The DASS-21 performed well overall among 602 Pakistani university students. Internal consistency was good, with Cronbach’s alpha coefficients within the reported scales ranging from 0.82 to 0.88. Confirmatory factor analysis produced encouraging fit statistics, including a CFI of 0.954, TLI of 0.948, and RMSEA of 0.052. Measurement invariance across medical and non-medical disciplines was also supported.

The main psychometric caution is equally important: Depression-Stress and Anxiety-Stress latent correlations were extremely high at 0.939 and 0.949, respectively. The findings therefore support the DASS-21 as a useful research instrument while suggesting that researchers should avoid assuming that its three subscales represent sharply independent dimensions in Pakistani university students.

For future work, replication across more diverse Pakistani populations and direct testing of alternative structural models could help clarify whether the substantial overlap is best understood as expected comorbidity, shared negative affect, a broader general-distress dimension, or some combination of these explanations.

Frequently Asked Questions

Is the DASS-21 reliable in Pakistani university students?

In this study of 602 university students in Lahore, the evaluated scales showed good internal consistency, with Cronbach’s alpha coefficients ranging from 0.82 to 0.88. These findings support reliability in this particular student sample, although they should not automatically be generalized to every population in Pakistan.

Did the DASS-21 factor structure fit the Pakistani data?

Yes. Confirmatory factor analysis showed adequate model fit, with CFI = 0.954, TLI = 0.948, and RMSEA = 0.052. These values support the tested factor structure while not implying that the model is perfect.

Are DASS-21 depression, anxiety, and stress completely separate constructs?

Not in this dataset. The latent dimensions were very strongly related. Depression and Stress correlated at 0.939, while Anxiety and Stress correlated at 0.949. Researchers can still examine the subscales separately, but their substantial overlap should be acknowledged.

Can DASS-21 scores diagnose depression or anxiety disorders?

No. The DASS-21 is a self-report symptom measure rather than a diagnostic interview. Elevated scores can indicate greater symptom severity and may support further assessment, but psychiatric diagnoses require appropriate clinical evaluation.

Can medical and non-medical Pakistani students be compared using the DASS-21?

The study reported support for measurement invariance across academic discipline, strengthening the interpretability of comparisons between the medical and non-medical groups studied. Replication in other institutions would provide additional confidence.

Medical disclaimer: This article is intended for educational and research information only. The DASS-21 is not a substitute for diagnosis or individualized assessment by a qualified healthcare or mental-health professional. Anyone experiencing significant, persistent, or worsening psychological symptoms should seek appropriate professional evaluation.