Psychology HL
PREPARATION: Experimental IA design (2026 legacy)
- 1.
PREPARATION ONLY. Legacy experimental-IA design PREPARATION, not an investigation. A hypothetical group proposes a repeated-measures recall experiment: every consenting adult studies List A without category headings first, then List B with headings. Lists differ in familiarity and difficulty. Diagnose the two distinct confounds, propose a counterbalanced plan that separates heading condition from list and order, and specify a common scoring rule. No experiment is authorised.
[3 marks] · no calculatorCourse scope: Legacy course
Syllabus placement evidence: Psychology examiner instructions 2026, Internal assessment: Format of the report; Group work; Non-experimental studies. Written planning for legacy experimental investigation only; no group investigation, data collection or individual assessed report is delivered. Distinct advanced HL-track rehearsal; IA and practical requirements remain shared with SL, not additional HL-only prescription.
Worked explanation
- The second condition benefits or suffers from going second, while List B may itself be easier. Reversing order alone would not remove the association of headings with List B.
- One crossed plan has four allocated sequences: A with headings then B without; A without then B with; B with then A without; B without then A with. Across sequences, each list occurs in each condition and each condition occurs first and second.
- Use matched presentation and recall intervals, count distinct correctly recalled target words under a predefined duplicate/intrusion rule, and compare within-person condition scores. An actual legacy IA would additionally need an appropriate published basis and teacher-reviewed ethical design.
Common mistake: Counterbalancing order without crossing list identity with the heading treatment leaves material difficulty confounded with the independent variable.
Marking points
- Identifies order/practice/fatigue and unequal list difficulty as separate threats to attributing a difference to headings.
- Crosses heading condition with list identity and counterbalances condition order across participants, rather than merely swapping the same fixed lists.
- Specifies a fixed delayed-recall interval and a shared correct-word scoring rule with equal exposure and recall times.
Examiner tip: Counterbalancing order without crossing list identity with the heading treatment leaves material difficulty confounded with the independent variable.
- 2.
PREPARATION ONLY. Legacy experimental-IA ethics PREPARATION. A hypothetical memory-study plan tells adult volunteers they will be ranked publicly, although no ranking will occur, to make them try harder; the group intends to reveal the truth only after publishing named scores. Evaluate the deception and harm risks, redesign recruitment and debriefing, and specify how group work and individual reporting should be separated. Do not recruit anyone or conduct the task.
[3 marks] · no calculatorCourse scope: Legacy course
Syllabus placement evidence: Psychology examiner instructions 2026, Internal assessment: Format of the report; Group work; Non-experimental studies. Written planning for legacy experimental investigation only; no group investigation, data collection or individual assessed report is delivered. Distinct advanced HL-track rehearsal; IA and practical requirements remain shared with SL, not additional HL-only prescription.
Worked explanation
- The promised ranking can create anxiety even if it never happens; disclosing names then creates a separate confidentiality problem. Greater effort does not make these avoidable risks ethically necessary.
- Remove the public-ranking threat, explain the ordinary recall task and voluntary participation beforehand, and make withdrawal meaningful without penalty. Explain the actual purpose and data handling in a timely debrief rather than assuming a later confession repairs harm.
- An approved legacy investigation can be planned and conducted collaboratively within its rules, but each candidate writes their own report from permitted shared data. This hypothetical evaluation supplies neither ethical approval nor permission to publish anyone's identity.
Common mistake: Debriefing does not retrospectively justify avoidable distress or undo an unauthorised public release of identifiable scores.
Marking points
- Recognises unnecessary deception, threatened public evaluation and named disclosure as avoidable psychological/privacy risks, not a justified motivational manipulation.
- Replaces the threat with truthful low-risk task information, voluntary consent/withdrawal and confidential handling; explains debriefing before any agreed use of data.
- Distinguishes permitted group planning/conduct from independent final reports and prohibits publishing identifying scores without appropriate authorisation.
Examiner tip: Debriefing does not retrospectively justify avoidable distress or undo an unauthorised public release of identifiable scores.
- 3.
PREPARATION ONLY. Legacy experimental-IA analysis PREPARATION. Entirely invented paired recall scores from four adults, not collected data: without headings = 4, 5, 7, 6 words; with headings = 7, 7, 8, 6 words in the same participant order. Calculate each paired difference (with minus without), the mean and range of those differences, and explain why treating the eight scores as independent observations would misrepresent the design. Do not claim statistical significance.
[3 marks]Course scope: Legacy course
Syllabus placement evidence: Psychology examiner instructions 2026, Internal assessment: Format of the report; Group work; Non-experimental studies. Written planning for legacy experimental investigation only; no group investigation, data collection or individual assessed report is delivered. Distinct advanced HL-track rehearsal; IA and practical requirements remain shared with SL, not additional HL-only prescription.
Worked explanation
- Keep participants aligned: 7 - 4 = 3, 7 - 5 = 2, 8 - 7 = 1 and 6 - 6 = 0 words. Rearranging either condition separately would destroy the pairing that defines the comparison.
- The total difference is 6 words across four people, so the mean difference is 6/4 = 1.5 words. The largest difference minus the smallest is 3 - 0 = 3 words; this is the range of differences, not the range of all raw scores.
- Each person contributes two related scores, giving four paired responses rather than eight independent participants. A real analysis must justify its paired procedure and assumptions; these small hypothetical data do not license an invented p-value or a claim that every participant improved.
Common mistake: Pooling the two conditions as eight independent people discards participant matching and overstates the independent sample size.
Marking points
- Calculates paired differences of 3, 2, 1 and 0 words.
- Calculates mean paired difference 1.5 words and range of differences 3 words.
- Explains dependence within each participant and the need for design-appropriate paired inferential analysis/assumptions; the descriptive change alone does not establish a reliable population effect.
Examiner tip: Pooling the two conditions as eight independent people discards participant matching and overstates the independent sample size.
- 4.
PREPARATION ONLY: legacy HL experimental IA. In a hypothetical imagery-versus-repetition recall experiment, a scorer knows each condition and gives partial credit for approximate answers only when expecting good performance. Allocation is already random and instructions standardised. Explain the remaining bias, design a blinded scoring procedure with an auditable rule, and explain why this is different from random allocation. Do not conduct a study; no teacher or ethics approval is claimed.
[3 marks] · no calculatorCourse scope: Legacy course
Syllabus placement evidence: Psychology examiner instructions 2026: Internal assessment - Non-experimental studies (controlled variables); Format of the report. Applies experimental control to scoring in legacy HL IA preparation. Original design rehearsal only, not an authenticated experiment, approved protocol or new-cohort proposal.
Worked explanation
- Random allocation does not stop an informed scorer from giving unequal credit after responses are produced. The bias occurs at measurement, even if the original comparison was fairly allocated.
- Remove condition labels and use a fixed answer key before scoring, retaining coded records for checking disputed decisions. Keep the allocation key separate until scores are final; blinding and allocation address different threats.
Common mistake: Treating random allocation as protection against every later source of experimenter bias.
Marking points
- Identifies expectancy-related measurement bias from condition-dependent scoring.
- Proposes coded condition-hidden responses scored against a pre-specified exact/acceptable-answer rule with documented decisions.
- Distinguishes reducing scorer bias from random allocation's role in balancing participant differences.
Examiner tip: Treating random allocation as protection against every later source of experimenter bias.
- 5.
PREPARATION ONLY: legacy HL IA descriptive analysis with entirely invented independent-group word-recall scores. Condition A: 2, 3, 3, 4, 13. Condition B: 4, 4, 5, 5, 6. Calculate each mean and median. Explain why the summaries support different descriptions of a typical score, and state why the score 13 must not simply be deleted to favour a hypothesis. No significance claim or actual investigation is requested.
[3 marks] · no calculatorCourse scope: Legacy course
Syllabus placement evidence: Psychology examiner instructions 2026: Internal assessment - Format of the report (descriptive and inferential statistics). Adds distribution-sensitive descriptive analysis to legacy HL experimental-IA preparation. Synthetic scores are not collected results and cannot complete the analysis/report requirement.
Worked explanation
- Both lists are already ordered: their middle values are 3 and 5. Their totals are 25 and 24, giving means of 5 and 4.8; the mean ordering reverses the median ordering because 13 strongly affects A's total.
- Describe the distribution and verify the high score against permitted range/original records in a real study. An unusual but valid observation is not automatically an error; deleting it after seeing its effect would bias the comparison.
Common mistake: Selecting whichever centre measure supports the prediction while hiding the other or removing an inconvenient score.
Marking points
- Calculates means A = 25/5 = 5 and B = 24/5 = 4.8 words.
- Calculates medians A = 3 and B = 5 words and explains that A's high value pulls its mean above its median.
- Requires checking accuracy and a justified transparent exclusion policy, not outcome-driven removal of a valid extreme score.
Examiner tip: Selecting whichever centre measure supports the prediction while hiding the other or removing an inconvenient score.