OCA-531 RM — Model Answers and Marking Guide

Q1 answer key

Award 1 mark for the option and 1 mark for a valid reason.

No.	Answer	Reason and closest-distractor check	Module
1	D	Good research should be systematic, controlled where appropriate, and empirical; high cost is not required.	1
2	B	Basic research expands knowledge/theory; applied research targets an immediate practical problem.	1
3	B	Constructivism treats meaning and reality as socially constructed; positivism assumes an observable objective reality.	1
4	A	Iterative narrowing improves clarity and feasibility; remaining broad obstructs measurement.	2
5	B	The fish head states the focal effect/problem and the bones organise possible causes.	2
6	B	Repeatedly asking why moves from symptoms toward root causes; it is not a sampling procedure.	2
7	C	Random assignment and control support causal inference better than observational association.	3
8	C	“Reduces” specifies the direction; a two-tailed alternative would allow either change.	3
9	D	Reliability concerns consistency; confounding, selection, and history threaten internal causal validity.	3
10	B	Ordered response categories establish rank but not guaranteed equal intervals.	4
11	A	Alpha measures how consistently items in a multi-item scale relate; it does not establish validity.	4
12	C	Construct validity concerns whether the intended construct is measured; the others are reliability methods.	4
13	A	Strata are formed and sampled internally; selecting whole groups describes cluster sampling.	5
14	B	Larger random samples reduce standard sampling error but do not automatically remove systematic bias.	5
15	B	p(1−p) is maximal at 0.5, producing the largest conservative sample.	5
16	B	The median resists extreme values; the mean is pulled toward them.	6
17	C	Magnitude 0.80 is strong and the minus sign is negative; correlation alone is not causation.	6
18	B	The coefficient 2 is expected y change per one-unit x increase; 3 is the intercept.	6
19	C	Independence factorises the joint probability; adding applies to mutually exclusive alternatives.	7
20	A	Bayes revises a prior using likelihood of observed evidence; it does not prove causation.	7
21	A	Repeated-sample means approach normality under standard conditions; raw data need not be normal.	7
22	B	Since p>α, evidence is insufficient to reject; this does not prove or “accept” H₀.	8
23	B	A two-sample t-test compares independent means when population variance is unknown; paired data require pairing.	8
24	B	Alpha is the chosen false-positive probability; beta is Type II error and 1−β is power.	8
25	B	These are Wallas's creativity stages, not the complete research process.	9
26	B	IEEE uses citations such as [1]; author-date is typical of styles such as APA.	9
27	B	Belmont identifies respect for persons, beneficence, and justice; the other lists are not its principles.	10
28	B	Inventing observations is fabrication, one of fabrication/falsification/plagiarism misconduct.	10
29	B	CS research must go beyond implementation to correctness, analysis, and empirical comparison.	11
30	A	Conclusions fail to transfer if the simulation model is unrealistic; formatting is not a validity threat.	11

Q2 model answers

1. Basic research seeks general knowledge or theory without an immediate use requirement. Applied research uses systematic inquiry to solve a specific practical problem. Marks: each definition 1. (Module 1)
2. Check time, resources, ethics, data availability, and scope. A focused research question must be achievable within all five constraints. Marks: five checks 1; purpose 1. (Module 2)
3. Internal validity asks whether the observed effect is truly caused by the studied factor; external validity asks whether findings generalise beyond the study. Marks: each definition 1. (Module 3)
4. Nominal: blood group; ordinal: satisfaction rank; interval: Celsius temperature; ratio: age or mass. Marks: four scales 1; suitable examples 1. (Module 4)
5. Probability sampling gives population units known nonzero selection chances, enabling sampling-error inference. Non-probability sampling uses convenience, judgement, quota, or referrals and has unknown inclusion chances. Marks: each type 1. (Module 5)
6. Type I error rejects a true null hypothesis with probability α; Type II error fails to reject a false null with probability β. Power is 1−β. Marks: each error 1. (Module 8)
7. State dataset/sample, comparison baselines, performance metrics, and statistical analysis. Together they make the proposed claim testable. Marks: four elements 1; purpose 1. (Module 11)

Q3 model answers

1. Example: “Among university students aged 18–25, does using social media for more than three hours daily predict higher validated anxiety scores after six months compared with use of three hours or less?” This identifies population, exposure/comparator, measurable outcome, and time frame and can be tested observationally while controlling confounders. Rubric: population 1; exposure/comparator 1; outcome 1; time 1; coherent question 1. (Module 2)
2. Items rated 1–5 from strongly disagree to strongly agree: “Resources meet my needs,” “Staff are helpful,” “Study spaces are comfortable,” and “Digital access is reliable.” Reverse-code any negatively worded item. Pilot the scale and assess internal consistency with Cronbach's alpha (about 0.70 or higher is commonly acceptable), obtain expert review for content validity, and test construct validity against related measures. Rubric: four relevant items 2; response scale 1; reliability 1; validity 1. (Module 4)
3. Two rolls give 6×6=36 equally likely ordered outcomes. Sum 7 occurs for (1,6),(2,5),(3,4),(4,3),(5,2),(6,1), six outcomes. Thus P(sum=7)=6/36=1/6≈0.1667. Rubric: space size 1; outcomes 2; fraction 1; result 1. (Module 7)
4. Define the problem as excessive energy use and identify a gap in current profiling. Propose an analyser/optimisation, a representative app dataset, a strong baseline, battery/latency metrics, and paired statistical comparison. Include expected contribution, timeline, risks, reproducibility, user consent for telemetry, minimised/anonymised data, and secure retention. Rubric: problem/gap 1; method 1; evaluation 2; ethics/reproducibility 1. (Modules 10–11)