Statistics and Research — Prometric Style Questions, Set 2

What this deck covers

Statistics and Research – Prometric Style Questions, Set 2 in 50 board-style clinical cases with worked explanations. Each case gives you a clinical vignette, four options, the correct answer, and an explanation of why each of the other three is wrong, with the guideline or textbook it follows named on the answer.

Board-style teaching questions for paediatricians and trainees preparing for board, MRCPCH and Prometric examinations. Written from published guidelines and standard paediatric textbooks, with the source named on every answer.

The 50 cases, in order

1. Counting asthma in the clinic
2. Making sure the toddlers count
3. Children who share the same air
4. The families who never wrote back
5. Every twentieth name on the list
6. A list anyone could read
7. Guessing what comes next
8. Balancing three things at once
9. Eight months apart at the start
10. What adjustment can never do
11. Fourteen children in the country
12. Switching with no gap between
13. Good enough, not better
14. Everyone gets it, in turn
15. Randomised twice over
16. Six faces, numbered
17. Doubling dilutions with a tail
18. Skewed, but nine hundred of them
19. Why 61.6 to 62.4 is wrong
20. Throwing away the numbers
21. Point nine six, and still wrong
22. Rising, but not in a line
23. Ninety per cent, mostly normal
24. Admitted or sent home
25. Half, or not half
26. The ones who did not relapse
27. Recruited over four years
28. A curve that never reaches half
29. Nought point six, explained
30. Curves that cross and part
31. The one thing still missing
32. Ten points, then five
33. The ones who stopped coming
34. Eighteen per cent unweighed
35. Stopped at the first look
36. Comparing two crude rates
37. Rates applied to one population
38. Reading a standardised ratio
39. Same summary, different shape
40. A district with almost no deaths
41. The same answer every time
42. Twenty items, one idea
43. Everyone already scores ten
44. Four items against the full test
45. Convenient, and also helpful
46. The wrong checklist downloaded
47. One step before the first child
48. Whose name goes on it
49. The same trial, twice
50. Fees from the manufacturer

Diagnoses and management options tested

Across the fifty cases you are asked to choose between options such as: A random sample drawn from the school registers, Non-response bias, The last allocation in each block is predictable, It balances unmeasured factors too, Whether families will be told, Ordinal, The range within, Spearman’s rank correlation, About a quarter of the variation is shared, The median is the time at, The smallest difference worth detecting, Multiple imputation, Direct standardisation, Indirect standardisation.

Clinical pearls from this deck

• Ask what list the participants were drawn from before you ask how many there were. A wrong frame cannot be fixed by a bigger number.
• When a small subgroup carries the research question, stratify. Enlarging a simple random sample buys precision where you did not need it.
• If you allocated by group, you must analyse by group. Counting clustered children as independent is the commonest way a nursery or school study overstates its result.
• A response rate tells you how big the hole is, never what fell into it. Compare responders with non-responders on anything the sampling frame already knows.
• Systematic sampling only fails when the list has a rhythm that matches your interval. Look at how the register is ordered before you trust it.
• Concealment is about who gets in; blinding is about what happens afterwards. A trial can be perfectly blinded and still ruined at the front door.

Work through the deck once for recognition, then again a week later to test yourself — each question gives the answer and explains why every other option fails.

More from this system: all statistics and research case decks · pediatric reference values.