Quantitative Psychology Dissertation Topics: 110 Titles Built Around Real Models and Datasets

Are you trying to work out whether your quantitative psychology dissertation should focus on measurement, modelling or replication? The topics below are written as near-ready proposal titles, so you can see the dataset, the model and the question in one line. They are free to use, adapt or combine.
These are starting ideas, not finished studies. Your supervisor and your university’s ethics rules have the final say, so check both before you commit to one. If you want to compare with other branches of the subject first, the general psychology dissertation topics page is a good place to browse, and the guide to choosing a research topic helps you narrow down.
Psychometrics and Scale Validation Topics
Psychometric projects suit students who like clean, defensible numbers. Most need a questionnaire dataset of a few hundred responses and a confirmatory factor analysis, which you can run in R with lavaan or in JASP.
1. Does the PHQ-9 measure one depression factor or two in UK university students?
Compare a one-factor, two-factor and bifactor model on a student sample, and report which fits best and why.
2. Rasch analysis of the short Warwick-Edinburgh Mental Wellbeing Scale in postgraduate students
Check item fit, person separation and whether the response categories work as intended.
3. Is Cronbach’s alpha hiding problems in the Perceived Stress Scale? A comparison of alpha, omega total and omega hierarchical
4. How much information is lost when the GAD-7 is cut down to the GAD-2?
An item response theory comparison of test information curves across the anxiety range.
5. Developing and validating a short scale of academic procrastination for UK undergraduates
Generate items, run an exploratory factor analysis on one half of the sample, then confirm the structure on the other half.
6. Do negatively worded items create a method factor in the Rosenberg Self-Esteem Scale?
7. Exploratory graph analysis or parallel analysis: which recovers the intended factor structure better in an open Big Five dataset?
8. Statistics anxiety measurement: a two-parameter logistic IRT analysis of the Statistical Anxiety Rating Scale
Useful if you can recruit from first-year research methods modules, because the sample is easy to reach.
9. How reliable is a single-item life satisfaction measure compared with a multi-item scale over a two-week interval?
10. Item-level performance of the Bergen Social Media Addiction Scale: a graded response model
Measurement Invariance Dissertation Topics
Measurement invariance asks whether a scale means the same thing for different groups or at different times. Putnick and Bornstein’s review of measurement invariance conventions is a solid starting point for the logic of configural, metric and scalar steps.
11. Is the Big Five Inventory-2 invariant across gender in a UK student sample?
12. Does the WEMWBS mean the same thing for home and international students?
Test configural, metric and scalar invariance, then compare latent means only if scalar or partial scalar invariance holds.
13. Does the PHQ-9 function equivalently across ethnic groups in an openly available dataset?
14. How many freed intercepts are too many? A simulation of partial invariance and latent mean bias
15. Alignment optimisation versus multi-group CFA for human values across countries in the European Social Survey
16. Longitudinal invariance of the three-item UCLA loneliness scale in older adults
The English Longitudinal Study of Ageing includes loneliness items across several waves, so confirm exactly which waves in the data dictionary before you plan the analysis.
17. Do fixed ΔCFI cutoffs agree with simulation-based cutoffs when testing invariance?
18. Differential item functioning in a numeracy test by first language, using the mirt package
19. Do paper and online versions of the same wellbeing questionnaire produce equivalent measurement?
20. Age-group invariance of a life satisfaction measure across adulthood in an open panel dataset
Structural Equation Modelling Topics in Psychology
Structural equation modelling lets you test a whole theory at once, with measurement error handled properly. It works best when you can justify every path in advance, so draw the model before you collect anything.
21. Does sleep quality mediate the link between evening screen time and next-day mood in students?
Use a latent variable mediation model with bootstrapped confidence intervals rather than separate regressions.
22. Testing the Theory of Planned Behaviour as a latent SEM rather than a regression: does it change the conclusions?
23. Bifactor versus higher-order models of general psychopathology in adolescents using the Strengths and Difficulties Questionnaire
The Millennium Cohort Study includes the SDQ at several ages, so you can check which sweeps suit your design.
24. Network models or latent variable models for depression and anxiety symptoms: which explains the data better?
25. A MIMIC model of differences in stress by course type, controlling for measurement bias
26. Does model modification by modification indices overfit? A split-sample demonstration
27. Exploratory SEM versus CFA for the Big Five: is the cross-loading penalty justified?
28. Latent interaction models: does social support buffer the effect of workload on burnout?
Estimated with the latent moderated structural equations approach, which is available in Mplus.
29. Can Bayesian SEM with weakly informative priors rescue a model at n = 100?
30. Is student engagement reflective or formative? Comparing the two specifications on the same data
Longitudinal and Within-Person Modelling Topics
Longitudinal work is where a quantitative psychology dissertation can stand out. The key idea is separating differences between people from changes within a person, which Hamaker, Kuiper and Grasman introduced to cross-lagged panel models through the random intercept version.
31. Does the link between screen time and wellbeing survive once within-person and between-person effects are separated?
Compare a standard cross-lagged panel model with a random intercept cross-lagged panel model on three or more waves.
32. A latent growth curve of anxiety across the first year of university
33. Latent change score models of self-esteem and academic performance
34. Growth mixture modelling of life satisfaction trajectories in Understanding Society
35. Is the within-person reliability of a three-item mood measure good enough for diary research?
Use multilevel reliability estimates and compare them with the usual between-person alpha.
36. Dynamic structural equation modelling of daily stress and sleep duration
37. Does attrition bias growth estimates in a cohort study? A sensitivity analysis
38. Cross-lagged relationships between loneliness and depressive symptoms in older adults
39. How many prompts per day are enough? Prompt frequency and estimated mood variability in experience sampling
40. Time-varying effects of exam stress on sleep: a piecewise growth model
Multilevel and Mixed-Effects Modelling Topics
Multilevel models are a natural fit when participants sit inside classes, clinics or repeated trials. They also suit students who want a technical project without needing large clinical samples.
41. How much of the variance in maths anxiety sits between classrooms rather than between pupils?
42. Do random slopes change your conclusions? Maximal versus parsimonious mixed models in reaction time data
43. Mixed-effects models or repeated measures ANOVA for a within-subject experiment: which is more informative?
44. Crossed random effects for participants and stimuli in a word recognition task
45. Within-person centring or grand-mean centring: what changes in a diary study?
46. Multilevel CFA of a classroom climate questionnaire
47. Cross-level interactions between school climate and pupil motivation in an open school dataset
48. Generalised linear mixed models for binary accuracy data in a memory experiment
49. How few clusters is too few? A simulation of level-2 sample size in psychology designs
50. Location-scale models: predicting mood variability, not just mood level
Bayesian Methods Dissertation Topics
Bayesian projects need clear reasoning about priors, so they reward students who enjoy explaining their choices. The brms and JASP packages make these analyses accessible without a statistics degree.
51. Do Bayes factors and p-values agree? A re-analysis of ten published psychology experiments with open data
52. How sensitive are posterior estimates to the choice of prior in a small-sample intervention study?
53. Bayesian estimation of an item response model in brms
54. Bayesian meta-analysis with informative priors on between-study heterogeneity
55. Can equivalence testing and the region of practical equivalence support a claim of no effect?
56. Modelling reaction times with an ex-Gaussian distribution in a Bayesian mixed model
57. Posterior predictive checks: what do they reveal that fit indices miss?
58. Bayesian multilevel models with few clusters: when do priors help?
59. Updating beliefs across replications: sequential Bayesian analysis of a classic effect
60. Do psychology students interpret credible intervals more accurately than confidence intervals?
A short online survey experiment with a between-subjects design and a clear scoring key.
Meta-Analysis and Evidence Synthesis Topics
A meta-analysis needs no participants, which makes it attractive when ethics approval or time is tight. It does need a registered protocol, a reproducible search and careful coding, and the metafor package in R handles most models.
61. A three-level meta-analysis of mindfulness-based interventions for stress in students
62. Which publication bias method performs best? PET-PEESE, selection models and p-curve on the same set of studies
63. A reliability generalisation meta-analysis of a widely used depression scale
64. What does the published literature say about measurement invariance of one scale across cultures?
65. How much do corrections for range restriction change a meta-analytic correlation?
66. Meta-regression of sample type: do student samples inflate effect sizes?
67. A meta-analysis of gender differences in mathematics anxiety
68. Aggregate or individual participant data: does the choice change the answer?
69. A multiverse of meta-analytic decisions: how stable is one pooled estimate?
70. An umbrella review of meta-analyses on screen time and adolescent wellbeing
Replication, Preregistration and Open Science Topics
The replication debate shaped modern psychology. The Open Science Collaboration’s project estimating the reproducibility of psychological science and the many-analysts study by Silberzahn and colleagues are both worth reading before you pick a topic here.
71. Can a classic effect be replicated with a preregistered Bayesian analysis and open materials?
Pick an effect whose original materials are public, contact the original authors, and register your analysis plan before data collection.
72. A specification curve analysis of one published finding
73. A multiverse analysis of data exclusion rules in an online experiment
74. Do registered reports report effect sizes differently from standard articles?
75. How accurate are reported statistics? A statcheck audit of articles in one psychology subfield
76. A computational reproducibility audit of articles that share data and code
77. How often do published studies deviate from their preregistrations?
78. What is the median statistical power in a chosen subfield of psychology?
79. Questionable measurement practices: how often do studies report the reliability and validity of their scales?
80. Are Cohen’s effect size benchmarks realistic for individual differences research?
UK Secondary Data Analysis Topics
Secondary data gives you a large sample and lets you start analysing sooner. Understanding Society, the Millennium Cohort Study, the English Longitudinal Study of Ageing and Next Steps are all available through the UK Data Service after registration, but variable availability differs by wave, so read the data dictionary first. Some resources, such as UK Biobank and ALSPAC, need a formal application, so allow extra time.
81. The factor structure of the GHQ-12 across waves of Understanding Society
82. Trajectories of emotional and behavioural difficulties from childhood to adolescence in the Millennium Cohort Study
83. School engagement and later outcomes in Next Steps: a latent variable approach
84. Loneliness and cognitive function in the English Longitudinal Study of Ageing
85. Does the neuroticism measure in UK Biobank predict later mental health outcomes?
86. Childhood cognitive ability and adult mental health in the 1970 British Cohort Study
87. Personality stability and change across adulthood using the Big Five items collected in selected waves of Understanding Society
88. Does missing data in a cohort study differ by socioeconomic position, and does it matter for your model?
89. Regional variation in recovery rates in published NHS Talking Therapies statistics
This uses aggregate public statistics rather than individual records, so ethics requirements are usually simpler. Confirm that with your ethics committee.
90. Personal wellbeing measures from the Office for National Statistics: an analysis of group differences and measurement quality
Missing Data, Power and Simulation Topics
Simulation studies let you answer a methods question with code rather than participants. They are demanding, but a supervisor with a quantitative background will usually welcome one.
91. Multiple imputation or full information maximum likelihood in SEM with 30 percent missing data?
92. Sensitivity analyses for data missing not at random in a wellbeing survey
93. Monte Carlo power analysis for a random intercept cross-lagged panel model
94. How biased is Cronbach’s alpha when its assumptions are violated? A simulation against omega
95. Does the number of Likert response categories change CFA results?
96. WLSMV or maximum likelihood for ordinal questionnaire data: when does it matter?
97. How do different outlier rules change the conclusions of a typical psychology experiment?
98. Statistical power for interaction effects: how large a sample do you really need?
99. Sequential testing and stopping rules in online experiments
100. Planned missing data designs: can a shorter questionnaire battery recover the same estimates?
Machine Learning, Text Data and Online Data Quality Topics
Online samples are now routine in psychology, which raises measurement and data quality questions that older textbooks do not cover. These topics suit students with some coding experience.
101. Which careless responding indices work best on a crowdsourced sample? Long-string, Mahalanobis distance and instructed-response items compared
102. Can large language models code open-text responses as reliably as trained human coders?
Measure agreement with Cohen’s kappa or Krippendorff’s alpha against two human coders.
103. Do text embeddings add predictive value beyond scale scores for self-reported depression?
104. Does machine learning beat logistic regression when predicting student wellbeing?
Use cross-validation and report out-of-sample performance, not just training fit.
105. Response times as a data quality indicator in online questionnaires
106. How well can simple checks detect automated responses in an online survey?
107. Regularised regression for variable selection: does it improve out-of-sample prediction in psychological data?
108. Sentiment analysis tools against validated mood scales: how strong is the relationship?
109. Automated short-form construction: can an algorithm pick items as well as an expert?
110. Do attention checks improve data quality or change who stays in the sample?
Three Worked Examples of Quantitative Psychology Dissertation Topics
A topic becomes a proposal when it has an aim, a few objectives and a clear analysis. Here are three of the ideas above turned into that shape. The research aims and objectives generator can do the same for your own idea.
Example A: PHQ-9 factor structure (topic 1)
Aim: To establish whether the PHQ-9 is best represented as a one-factor, two-factor or bifactor structure in UK university students.
- Collect PHQ-9 responses from a student sample large enough for a confirmatory factor analysis.
- Fit and compare the three competing models using fit indices and information criteria.
- Report omega coefficients for the best model alongside alpha.
Example B: Within-person effects of screen time (topic 31)
Aim: To test whether the association between screen time and wellbeing reflects stable differences between people or changes within people over time.
- Identify a longitudinal dataset with at least three waves of both variables.
- Estimate a standard cross-lagged panel model and a random intercept version.
- Compare the cross-lagged paths and explain what any differences imply.
Example C: Careless responding indices (topic 101)
Aim: To compare how well common careless responding indices identify low-quality responses in a crowdsourced sample.
- Embed instructed-response items in a questionnaire as a benchmark.
- Compute long-string and Mahalanobis distance indices for every participant.
- Test how exclusion rules from each index change the factor structure and reliability.
How to Choose a Quantitative Psychology Dissertation Topic for Your Level
Your level changes what is realistic, so match the topic to the time and skills you actually have.
- Undergraduate: choose one scale and one question, such as a reliability check, a simple invariance test or a mediation model. Topics 1, 3, 9 and 21 fit well.
- Masters: add a second layer, such as comparing two models or analysing a cohort dataset. Topics 12, 31, 34 and 61 work at this level.
- PhD or research proposals: methods questions and simulations, such as topics 14, 49, 62 and 93, give you room for an original contribution.
Once you have a shortlist, test whether you can turn it into a clear question with the guide to writing a research question, and then build testable predictions using the hypothesis development guide. The dissertation topic finder is useful if several ideas still look equal.
Methods and Software for a Quantitative Psychology Dissertation
Choose software your department supports, because that is where you will get help. The usual options are listed below.
- R: lavaan and semTools for SEM, mirt for item response theory, lme4 for mixed models, brms for Bayesian models, metafor for meta-analysis and psych for descriptive psychometrics.
- JASP: a free point-and-click option that covers CFA, Bayesian tests and meta-analysis and shows the R code behind many analyses.
- Mplus, SPSS and Stata: common in departments with existing licences, and Mplus is widely used for latent variable and growth mixture models.
Before you collect data, check your sample size with the dissertation sample size calculator and your test choice with the statistical test selector. If the methodology chapter is where you feel stuck, the dissertation methodology help page explains the support available, and the psychology dissertation writing service covers wider projects.
Ethics and Data Rules for Quantitative Psychology Projects
Most UK psychology departments base their ethics review on the British Psychological Society Code of Human Research Ethics. Projects that recruit through the NHS usually need separate Health Research Authority approval, which takes far longer than a university review. Secondary data projects still need you to follow the licence and use conditions of the data provider, and any personal data must be handled in line with UK GDPR. Rules vary by university, so check your own department’s guidance before you start.
Common Mistakes in Quantitative Psychology Dissertations
- Running a factor analysis on a sample too small for the number of items.
- Comparing group means before checking that the scale measures the same thing in each group.
- Adjusting a model until it fits, then presenting it as if it was planned.
- Reporting alpha alone and ignoring the assumptions behind it.
- Treating a cross-sectional mediation model as proof of causal order.
- Leaving the analysis code unsaved, which makes results impossible to reproduce.
When your results are ready, the dissertation results help page and the literature review guide cover the writing side. If you are still choosing between psychology branches, compare this page with cognitive psychology research topics, clinical psychology dissertation topics and data science dissertation topics.
Quantitative Psychology Dissertation Topics FAQs
Do I need advanced maths for a quantitative psychology dissertation?
No. Most projects need a solid grasp of regression and a willingness to learn one new method, and software does the heavy calculation. Choose a topic where you can explain every step in plain words.
Can I do a quantitative psychology dissertation without collecting my own data?
Often yes. Secondary data analysis, meta-analysis, simulation and reproducibility audits all avoid recruitment, but your department must allow them, so confirm that early.
What is the difference between psychometrics and quantitative psychology?
Psychometrics is the part concerned with measuring psychological constructs, including reliability, validity and scale development. Quantitative psychology is wider and also covers statistical modelling, research design and methodology.
Which software should I use for my quantitative psychology dissertation?
R is the most flexible and free, JASP is easiest to start with, and SPSS or Mplus make sense if your department teaches them. Pick the one your supervisor can support.
How many participants do I need for a CFA or SEM dissertation?
It depends on model size, factor loadings and estimator, so there is no single number. Run a power analysis or simulation for your own model and discuss the choice with your supervisor.