BB2025 Research Methods & Data Analysis
The Complete Exam Bible · Semester 1, 2026/27BB2025 Overview
- The Hong Kong Polytechnic University
- Semester 1, 2026/27
- Level 2
- Statistics
- MCQs and short answers
What this course asks you to do
Research Methods & Data Analysis joins study design to statistical interpretation. The course introduction frames statistical literacy as necessary for planning valid studies, detecting error in published work and translating evidence into clinical practice.
- Assessed by Online quizzes, an oral research-proposal presentation and a final written test.
- Core method Start from the outcome, group structure and dependence before selecting an analysis.
- Proposal priority Align the knowledge gap, objective, design, measurement, sample size and analysis.
- Final-test practice Switch between exact multiple-choice distinctions and compact scenario interpretations.
How BB2025 is assessed
| Component | Weight | Format |
|---|---|---|
| Online quizzes | 20% | Four quizzes with one submission for each |
| Oral presentation (research proposal) | 40% | Research-proposal presentation with questions and answers |
| Final written test (exam) | 40% | Mix of MCQs and short answers |
The current course introduction publishes these components and the numeric rows total 100%. Active participation in the in-class activities is a minimum attendance requirement. A numeric threshold is not published, so the current course Canvas page controls the consequence and any operational instruction.
Assessment structure
The segment widths reproduce the published weights and total 100%.
Current BB2025 dates
| Date | Item | Control |
|---|---|---|
| 12 October 2026 | Research-proposal idea | Deadline shown in the current teaching timetable. |
| 23 November 2026 | Presentation PowerPoint | Submission deadline shown in the timetable and proposal guidance. |
| 30 November to 4 December 2026 | Presentation week | Presentation period stated in the proposal guidance. |
| 14 December 2026 | Final written test | Date stated in the current teaching timetable. |
Dates are as published in the current Semester 1 teaching timetable and research-proposal guidance. Confirm exact deadlines and submission settings in the live LMS.
What BB2025 covers
The chapters move from research questions, ethics and study design through statistical decisions and measurement to evidence synthesis and proposal defence.
Research Questions and Study Ethics
conceptual models, measurable outcomes, confounding paths and ethical designBegin with the decision the evidence should change
A useful research question names a population, an exposure or intervention, an outcome and a time point.
- Population, exposure, outcome and time
- Conceptual models and directional paths
- Confounding as an alternative explanation
- Operational definitions and measurable variables
- Research aims, questions and hypotheses
- Consent, privacy and confidentiality
Sampling and Experimental Study Design
assignment, comparison groups, observation timing and population coverageClassify studies from investigator control
An experiment assigns an intervention or condition. A quasi-experiment introduces a change without random individual allocation. An observational study records exposures that already exist.
- Experimental and observational assignment
- Randomised and quasi-experimental designs
- Cohort, case-control and cross-sectional timing
- Counterfactual comparisons
- Allocation and analysis units
- Probability sampling frames
Validity, Bias and Missing Data
selection, information, confounding, attrition and sensitivity analysisLocate the stage where validity was threatened
Selection bias changes who enters or remains in a comparison. Information bias changes how variables are observed.
- Internal and external validity
- Selection processes and analysis entry
- Measurement error and information bias
- Confounding pathways
- Differential attrition
- Missing-data mechanisms
Hypothesis Tests and Statistical Decisions
null models, p values, effect estimates, intervals and error trade-offsRead the test as a conditional argument
A hypothesis test assumes a null model, standardises an observed effect against expected variability and calculates how unusual that statistic or something more extreme would be under the reference model.
- Null and alternative hypotheses
- One-sided and two-sided tests
- Test statistics and reference distributions
- Conditional p value interpretation
- Effect estimates and confidence intervals
- Statistical and clinical importance
Comparing Groups with t Tests and ANOVA
independent, paired, omnibus and rank-based comparisonsSelect the comparison from structure
For continuous outcomes, the route depends on how many conditions are compared and whether observations are independent or linked. Independent t procedures compare unrelated groups.
- Independent and paired observations
- Difference scores and sign conventions
- Independent and paired t procedures
- Welch and pooled variance routes
- One-way omnibus ANOVA
- Repeated-measures ANOVA
Factorial, Repeated and Multivariate Models
interactions, repeated observations, covariates and outcome vectorsRead interactions before averages
A factorial model estimates main effects for each factor and an interaction asking whether one effect changes across levels of another.
- Main effects and interaction terms
- Simple effects and planned contrasts
- Difference-in-change interpretation
- Within-person covariance
- Sphericity and corrected inference
- ANCOVA baseline adjustment
Correlation, Regression and Categorical Data
slopes, prediction, contingency tables and paired binary outcomesLet the outcome scale set the language
Correlation treats two continuous variables symmetrically and summarises their pattern. Regression distinguishes an outcome from explanatory variables and estimates expected change or prediction.
- Pearson and Spearman correlation
- Scatterplots and nonlinear patterns
- Simple and multiple linear regression
- Conditional coefficient interpretation
- Residuals and influential observations
- Collinearity and predictor roles
Reliability, Validity and Diagnostic Accuracy
agreement, measurement error, diagnostic thresholds and predictive valuesConsistency does not guarantee a correct target
Reliability concerns repeatability under specified raters, occasions, instruments and populations. Validity concerns whether evidence supports the intended score interpretation and use.
- Internal consistency and item behaviour
- Test-retest and inter-rater designs
- Intraclass correlation selection
- Standard error of measurement
- Minimum detectable difference
- Construct and criterion validity
Survival, Epidemiology and Evidence Synthesis
censoring, hazards, appraisal, heterogeneity and qualitative evidenceTime-to-event outcomes preserve partial follow-up
Survival analysis combines whether an event occurred with when it occurred.
- Censoring and risk sets
- Kaplan-Meier survival estimates
- Hazard ratios and proportionality
- Descriptive and analytical epidemiology
- Systematic review methods
- Meta-analysis and pooled estimands
Research Proposals and Critical Appraisal
knowledge gaps, aligned methodology, feasibility and oral defenceMake every section justify the next
The proposal presentation is an argument from a knowledge gap to a feasible study. The literature review should narrow to an unresolved clinical question.
- Knowledge gaps and focused rationales
- Objectives, questions and hypotheses
- Original human clinical study boundary
- Sampling and measurement justification
- Primary outcomes and analysis alignment
- Timeline and budget feasibility
The sequence therefore begins before software: turn a clinical problem into a measurable research question, draw a conceptual model, identify the comparison that represents what would otherwise happen, and decide how participants, exposures and outcomes will be observed.
How the statistical chapters fit together
Hypothesis testing supplies a conditional decision rule, but a defensible conclusion also needs an effect estimate, an interval and a design that supports the chosen verb.
Group-comparison chapters distinguish independent from linked observations before choosing t tests, ANOVA or rank-based alternatives. Factorial and repeated models add interactions and covariance.
Regression and categorical methods change with the outcome scale, while measurement chapters separate reliability, validity, agreement and diagnostic accuracy.
What makes the course difficult
The difficult part is not calculation by hand. The teaching material states that hand calculation is not required.
The challenge is choosing a method from a short scenario, reading output without reversing a probability, and explaining what the design still cannot prove.
A significant result can be clinically small, a reliable measure can be invalid, a high correlation can hide poor agreement, and an adjusted coefficient can remain non-causal.
The research-proposal assessment
The oral presentation requires an original human clinical research study.
The guidance excludes systematic reviews, meta-analysis, secondary database analysis, animal research and purely cellular or molecular work. The proposal must connect its literature gap to objectives, design, sampling, measurement, analysis, timeline, budget and impact.
The standalone rubric places its largest content emphasis on methodology, so coherence matters more than the number of techniques named.
How this guide uses the materials
Assessment labels and weights follow the current course introduction, while dates and coverage follow the Semester 1 timetable.
Older teaching files are used for stable statistical explanations where they remain part of the current Canvas sequence. All worked cases and practice questions here are newly authored.
The available materials contain tutorial solutions and answer activities but no past final paper or sample examination.
Participation and pass conditions
Active participation in the in-class activities is a minimum attendance requirement. A numeric threshold is not published in the available course materials.
The current course Canvas page controls the consequence and any operational instruction, so check it alongside the weighted assessment table.
Worked example · free
Choose and interpret a paired comparison
- 2Identify the observations as linked within participant.
- 3Compute the mean change from the printed values.
- 3Interpret direction and name the design limit.
Key terms
- Conceptual Model
- A diagram or explicit account of variables and the directions through which they are expected to influence one another.
- Confounding
- Distortion of an exposure-outcome association by a common cause of both variables.
- Internal Validity
- The credibility of the estimated relationship for the participants and procedures actually studied.
- External Validity
- The extent to which a study result can inform another population, setting or time.
- Null Hypothesis
- The reference statement about a population parameter against which a test statistic is evaluated.
- Confidence Interval
- A model-based range of parameter values compatible with the observed estimate at a stated confidence level.
- Interaction
- A pattern in which the effect or association of one factor changes across levels of another factor.
- Regression Coefficient
- The expected outcome change associated with one predictor unit while the other included predictors are held constant.
- Reliability
- The consistency of measurements under specified raters, occasions, instruments and populations.
- Validity
- The strength of evidence supporting a score interpretation and its intended use in a defined context.
- Sensitivity
- The proportion of people with a condition who receive a positive test result.
- Specificity
- The proportion of people without a condition who receive a negative test result.
- Censoring
- A time-to-event record known only up to the last observed event-free time.
- Heterogeneity
- Variation in study effects or characteristics that changes how a synthesis should be interpreted.
BB2025 FAQ
How is the course assessed?
The current course introduction assigns twenty per cent to four online quizzes, forty per cent to the oral research-proposal presentation and forty per cent to the final written test. The numeric weights total one hundred per cent. The timetable and proposal guidance give current dates, while the live Canvas settings remain the operational source for submission instructions.
What is published about the final written test?
The course introduction describes a mix of multiple-choice questions and short answers, and the timetable places the test on 14 December 2026. The materials available here do not give a duration, section count or permitted-material rule. Prepare for method selection and interpretation, then check the current Canvas announcement for operational details.
How should I choose a statistical test?
Begin with the outcome scale, then count groups or predictors and decide whether observations are independent, paired, clustered or repeated. Next identify the population quantity the question asks for. Only then check distributional and model assumptions. This order protects against selecting a familiar procedure that estimates the wrong comparison.
What does a p value tell me?
It measures how compatible the observed statistic or something more extreme is with a specified null model under the test assumptions. It is not the probability that the null is true and it does not measure clinical importance. Report the effect estimate and confidence interval before giving the test decision.
What must the research proposal contain?
The presentation guidance requires an original human clinical study with background, rationale, objectives or hypotheses, methodology, timeline, justified budget, potential impact and references. It excludes systematic reviews, meta-analysis, secondary database analysis, animal research and purely cellular or molecular work. The planned project should remain feasible within the stated assignment constraints.
Where are marks won in the presentation rubric?
The standalone rubric rewards the background and knowledge gap, methodology, feasibility, significance, graphics, delivery, and response to questions with critical thinking. Methodology is the largest content category, so align the primary outcome, design, sampling, measurement, sample-size basis and analysis before polishing the deck.
Are past papers available in the course folder?
The available folder contains tutorial answers, laboratory exercises and video-answer links, but it does not contain a past final paper, sample final or practice examination. This guide therefore uses newly authored questions shaped only by the published mix of multiple choice and short answer, without implying access to an earlier paper.
Do I need to calculate statistics by hand?
The teaching material states that hand calculation is not required. Revision should focus on selecting a method, checking assumptions, reading software output and expressing the conclusion in defensible language. You should still be able to recompute simple printed quantities, such as a mean difference or interaction contrast, to verify direction and interpretation.
What is published about attendance and passing conditions?
Active participation in the in-class activities is a minimum attendance requirement. A numeric threshold is not published in the available course materials. The current course Canvas page controls the consequence and any operational instruction, so check it together with the weighted assessment table before making attendance or pass decisions.
Questions students ask in each BB2025 chapter
- How narrow should a research question be?
- It should be narrow enough that the population, exposure or intervention, comparison, outcome and time point can all be stated without placeholders. A good check is whether the question already implies what observations must be collected. If several incompatible designs could answer it, the wording probably still contains more than one research problem. Chapter 01 · Research Questions and Study Ethics
- What is the first question when identifying a study design?
- Ask who determined exposure or intervention. Investigator assignment points toward an experiment, while naturally occurring or chosen exposure points toward observation. Then inspect whether allocation was random, how the comparison was formed and whether observations move forward, backward or occupy one period. This reconstruction is safer than matching a familiar label. Chapter 02 · Sampling and Experimental Study Design
- What is the difference between bias and random error?
- Bias shifts an estimate systematically through selection, measurement, confounding or analysis. Random error makes estimates vary across repeated samples. Larger samples usually reduce random error, but they do not remove a systematic process and can make a biased estimate appear very precise. Diagnose the mechanism before recommending more observations. Chapter 03 · Validity, Bias and Missing Data
- Does a small p value show that the null is unlikely?
- Not from a standard frequentist test. The p value conditions on the null model and asks about the observed statistic or something more extreme. It does not reverse that relationship to calculate a probability for the hypothesis. To interpret the result, add the effect estimate, confidence interval, study design and clinical context. Chapter 04 · Hypothesis Tests and Statistical Decisions
- How do I decide between an independent and paired test?
- Ask whether each value in one condition has a meaningful link to one value in the other. Repeated measures from the same participant and deliberately matched pairs are linked; separate treatment groups are independent. Pairing changes the unit of analysis to the within-pair difference, so it must come from the design rather than convenience. Chapter 05 · Comparing Groups with t Tests and ANOVA
- What does a significant interaction mean?
- It means the effect or association of one factor varies across levels of another under the fitted model. It does not prove that every simple contrast is non-zero. Show the cell means or profile plot, estimate the planned simple effects with intervals and avoid presenting an overall main effect as the whole conclusion. Chapter 06 · Factorial, Repeated and Multivariate Models
- Does a high correlation mean one variable predicts another well?
- Not necessarily. Correlation summarises a sample's linear or monotonic association and does not evaluate calibration, prediction error or performance in new data. Outliers, subgroup mixing and restricted ranges can change it substantially. A prediction model needs explicit direction, validation and error assessment beyond a single pairwise coefficient. Chapter 07 · Correlation, Regression and Categorical Data
- Can a measure be reliable but invalid?
- Yes. Repeated measurements can cluster closely around a biased target, or a coherent item set can measure the wrong construct. Reliability limits how well a score can support many uses, but validity requires additional evidence from theory, relationships, known groups, reference measures and the actual decision context. Chapter 08 · Reliability, Validity and Diagnostic Accuracy
- What does censoring mean in a survival analysis?
- It means the exact event time is unknown beyond the participant's last event-free observation. The participant still contributes information up to that point. Censoring is not the same as never experiencing the event, and it can bias inference if leaving follow-up is related to prognosis after the available model information is considered. Chapter 09 · Survival, Epidemiology and Evidence Synthesis
- What kind of project is allowed for the presentation?
- The guidance requires an original human clinical research study. It excludes systematic reviews, meta-analysis, secondary analysis of existing databases, animal research and purely cellular or molecular research. The proposed study may differ from a future dissertation. Use the assignment boundary to reject an attractive topic before investing in an ineligible design. Chapter 10 · Research Proposals and Critical Appraisal
How to study for the exam
Build revision around decisions rather than lecture order. For each scenario, write the outcome scale, number of groups or predictors, dependence structure and target parameter before naming a procedure. Then interpret output in a fixed sequence: effect and direction, interval, p-value decision, clinical meaning and design limit.
Alternate one multiple-choice distinction with one short answer so recognition and construction develop together. For the proposal, keep a single alignment sheet linking the knowledge gap, objective, primary outcome, design, recruitment, sample-size assumptions and analysis. If one row changes, update every downstream row.
Rehearse the fifteen-minute presentation with interruptions and prepare the strongest objection to feasibility, measurement and analysis. A good response may concede a real limitation while explaining why the chosen compromise remains proportionate and what sensitivity analysis or pilot evidence would reduce the uncertainty.
Use figures only when they reveal relationships: a conceptual model for causal paths, a flow diagram for selection and loss, and an effect plot for estimates and intervals. Before any current deadline or examination instruction, verify the live course Canvas page rather than relying on a saved copy.
Your AI Statistics tutor for BB2025
Stuck on a hard BB2025 question? Sia is AskSia’s AI Statistics tutor — ask any BB2025 Research Methods & Data Analysis question and get a clear, step-by-step explanation grounded in how the course is actually taught and assessed. Read this whole study guide free, then take your hardest questions to Sia.
