IB Diploma Programme Psychology HL (first assessment 2021; current guide). This topic is the HL-ONLY research extension: the paradigms behind qualitative and quantitative research, the analysis and trustworthiness criteria of qualitative work, and the PAPER 3 skill — reading an unseen summary of a study and evaluating it methodologically and ethically. Everything here builds on topic 03's methods; the new work is judging them in use. The worked scenario is original, and its numbers are carried through the page.

Must-know for the HL extension:

Quantitative research 量化研究 tests hypotheses: numbers, controls, large samples, statistical decisions — it answers HOW MUCH and WHETHER there is an effect, and its quality criteria are reliability (would the measure repeat?) and validity (does it measure the construct, and do conclusions hold internally and externally?). Qualitative research 质性研究 explores meanings: how people understand, experience and account for their world — it answers WHAT and HOW it feels, through thick description rather than counts. The paradigms are not rivals on one scale: a questionnaire of 200 students finding $r = +0.42$ between study time and grades (a real-scenario style finding) says the two move together and NOTHING about mechanism; follow-up interviews asking WHY are not "weaker" — they answer the question the correlation cannot.

Qualitative SAMPLING is deliberate, not random: Purposive sampling 目的性抽样 selects information-rich cases (the six teachers who redesigned the curriculum); Snowball sampling 滚雪球抽样 recruits through participants' networks (hard-to-reach groups); convenience takes who is accessible. The criterion is SATURATION: new interviews adding no new themes, not sample size: the sixth interview that repeats the first five is a reason to stop, and a sample of one school is a reason to hedge transfer, not to collect the whole city.

Thematic analysis 主题分析 moves from raw talk to claims responsibly: transcribe, read wholly, CODE the meaningful units, group codes into candidate THEMES, check themes against the whole data set and against disconfirming passages, define and name each theme with telling extracts. CONTENT ANALYSIS quantifies qualitative material — counting how often categories appear — and inherits the quantitative duty of RELIABILITY: two raters coding the same 20 statements and agreeing on 16 show 80% inter-rater agreement, a level to report and improve by sharpening the coding manual (not by averaging away disagreement).

The trustworthiness of qualitative findings has four criteria: Credibility 可信度 (do the findings reflect the participants' reality — served by triangulation, member checking, prolonged engagement), transferability 可迁移性 (would they travel — served by THICK DESCRIPTION of the context, not by claims of representativeness), Dependability 可靠性 (would the process repeat dependably — served by an audit trail of decisions) and confirmability 可证实性 (do the data, not the researcher's preferences, support the conclusions — served by reflexivity 反身性, the researcher's declared role and assumptions). The bias family is the mirror image: the OBSERVER EFFECT (behaviour changing because a researcher is present), INTERVIEWER BIAS (leading questions, selective listening) and researcher expectancy — reflexivity names these rather than pretending them away.

Paper 3 gives unseen STIMULUS MATERIAL (a summary of a study: its aim, method, sample, procedure, data and findings) and asks three questions. QUESTION 1 is reading: identify the method and its features, the sampling, the data type; state what a finding in the stimulus suggests (and does not). QUESTION 2 is methodological discussion: strengths and limitations of the design, sample, measure and analysis — each named from the stimulus, each with its consequence. QUESTION 3 is evaluation: judge the study's credibility, transferability and ethics together, and RECOMMEND a modification with its trade-off — the marks live in the integration, not in reciting criteria.

Worked example (the scenario's spine). A researcher wants to know why students avoid asking questions in class. Design: six semi-structured interviews with volunteers from one year group (opportunity sample), thematic analysis producing three themes (fear of judgement, timing of pauses, perceived teacher preference for the already-confident). Evaluation a Paper 3 answer would make: credibility served by direct quotes and a described coding process, weakened by no member checking and single-researcher coding; transferability hedged — one school, one year group, volunteers — and hedged honestly through thick description; ethics — consent and anonymised transcripts present, but interviews on classroom anxiety touch a sensitive area, so distress management and the right to skip questions belong in the procedure. Modification: add a second coder with an agreement check (moving coding toward dependability) and member checking of themes (toward credibility), at the cost of time and of participants editing what they now recognise as their own themes.

• Quantitative criteria (reliability, validity) against qualitative trustworthiness (credibility, transferability, dependability, confirmability) — named and distinguished (paradigms).
• Purposive, snowball and convenience sampling with saturation as the stopping rule (sampling).
• Thematic analysis's six steps; content analysis with inter-rater reliability (16/20 = 80%) (analysis).
• The bias family — observer effect, interviewer bias, expectancy — answered by reflexivity and triangulation (bias).
• Paper 3's three questions and what each rewards; the worked scenario's evaluation and modification (Paper 3).

Boundary notes: Paper 3 is HL ONLY and its marks reward using the STIMULUS: every point names a feature from the summary, and criteria recited with no reference to the study cap the band. All scenario numbers are original.