Chapter 35 — Self-Check Quiz

20 questions. Answers with explanations at the end. Aim for 16+.

Questions

1. The structure the chapter says every product in it shares is:

• a) Deliberate fraud throughout
• b) A real finding, a partly-real measurement, and an unevidenced outcome claim stacked on top
• c) Sound science poorly marketed
• d) Three unevidenced claims

2. The Weizmann glucose-response work established that:

• a) The glycaemic index is entirely wrong
• b) Individual postprandial glucose responses to identical foods differ substantially, are reasonably stable within a person, and are partly predictable
• c) Everyone responds identically to the same food
• d) Microbiome composition determines all dietary needs

3. What that work did NOT establish is:

• a) That responses differ
• b) That eating to lower those responses improves any health outcome in non-diabetic people
• c) That the model outperformed carbohydrate content
• d) That the responses are stable

4. In §35.2's five-step chain, the unsupported link is:

• a) Step 1 — responses differ
• b) Step 4 — that lower postprandial glucose means better health in a non-diabetic person
• c) Step 2 — that they are predictable
• d) Step 3 — that a diet can be designed to lower them

5. The chapter compares this structure to:

• a) The placebo effect
• b) Chapter 2's beta-carotene template — a measurable intermediate observed to travel with an outcome, and an intervention sold before anyone tested whether moving it moves the outcome
• c) Publication bias
• d) Regression to the mean

6. The most damning practical finding about DTC genetic diet tests is:

• a) They are expensive
• b) The same sample can return materially different dietary recommendations, from different companies or even the same one
• c) They take too long
• d) They require a blood draw

7. The DIETFITS trial was unusual because it:

• a) Was very short
• b) Pre-specified the genotype hypothesis in advance, randomized, ran a year, supported adherence, and made BOTH arms eat a good diet
• c) Had no control group
• d) Tested supplements

8. DIETFITS found that:

• a) Low-carbohydrate won clearly
• b) Average weight change was similar between diets, individual variation within each arm was enormous, and neither the pre-specified genotype pattern nor insulin secretion predicted who did better
• c) The genotype pattern predicted response strongly
• d) Nobody lost weight

9. Results 2 and 3 of DIETFITS matter together because:

• a) They cancel out
• b) Individual variation is real and large — the genetic test simply did not explain it
• c) They show genetics is irrelevant to all nutrition
• d) They show diets don't work

10. Consumer microbiome tests have no:

• a) Scientific basis at all
• b) Agreed reference standard for a "healthy" microbiome — the comparison population is the company's own customer base
• c) Regulatory approval anywhere
• d) Sequencing capability

11. For non-diabetic CGM use, the chapter's overall verdict is:

• a) ✅ Well supported
• b) 🟡 Unclear — real educational value briefly, no demonstrated outcome benefit, and severe interpretation risks
• c) ❌ Not supported in any sense
• d) 🟢 Probably beneficial

12. "Flatter is better," pursued honestly in a non-diabetic person, recommends:

• a) More vegetables
• b) Avoiding fruit, legumes and whole grains — because fibre-containing carbohydrate raises glucose and cheese does not
• c) More exercise
• d) Smaller portions of everything

13. A postprandial glucose rise in a person without diabetes is:

• a) Evidence of pre-diabetes
• b) A normal physiological response — regulation is working when it rises and returns
• c) Always harmful
• d) A measurement artefact

14. The blood-type diet study found that:

• a) The diets worked as predicted
• b) Some diets produced better markers, and the benefit was unrelated to the person's actual blood type — the better diets contained more plants and less processed food
• c) No diet had any effect
• d) Blood type predicted response strongly

15. According to §35.7, the biggest evidence-based lever for personalizing nutrition is:

• a) Genotype
• b) Preference, culture, budget and schedule — because adherence beats composition
• c) Microbiome composition
• d) Glucose response

16. "Some people respond differently" becomes a finding rather than a hypothesis only when:

• a) A large trial reports it
• b) A replicate crossover shows the same person responds the same way twice
• c) A mechanism is proposed
• d) Two labs agree

17. Which of these is NOT one of the four things that produce apparent responders in a trial?

• a) Measurement error
• b) Regression to the mean
• c) Differential adherence
• d) Publication bias

18. The rule of n-of-1 experiments that the chapter says nobody follows is:

• a) Pick one variable
• b) Decide in advance what result would make you stop
• c) Alternate the condition
• d) Write down a prediction

19. Personalized systems' outputs converge on conventional advice partly because:

• a) The companies are copying each other
• b) Sensible algorithms hard-code conventional constraints to stop themselves recommending cheese over lentils
• c) The algorithms are broken
• d) Regulators require it

20. §35.12 says a rule generated from a test of your own biology is harder to argue with because:

• a) It is more accurate
• b) It arrives with the authority of personal data, so disagreeing with it feels like disagreeing with your own body
• c) It is legally binding
• d) It comes from a doctor

Answers

1. b) Real finding, partly-real measurement, unevidenced outcome claim. ⚠️ The skill worth having is seeing the join. §35.0, and it recurs in every product in the chapter.

2. b) Responses differ substantially, are stable, and are partly predictable. ⚠️ A real result, replicated in its core observation, and it complicates the glycaemic index properly — the GI was always a population average, and this work showed how much individual variation that average conceals. §35.1.

3. b) That acting on them improves outcomes. ⚠️ This is the whole chapter. §35.1, §35.2.

4. b) Step 4. ⚠️ In diabetes, glycaemic control is an established target with hard outcome evidence (Ch 26). ⚠️ In non-diabetic people, whether the size of a postprandial excursion independently predicts disease — after accounting for what is eaten, weight, activity, fasting glucose and HbA1c — is genuinely unclear. §35.2.

5. b) The beta-carotene template. ⚠️ Observed food → inferred compound → intervention sold before the test. Ch 2's recurring cautionary structure, in a new costume. §35.2.

6. b) The same sample returns different advice. ⚠️ Different companies use different panels, effect-size assumptions and proprietary algorithms, with no agreed standard. ⚠️ A test that gives different answers to the same DNA is not measuring what it claims to measure. §35.3.

7. b) Pre-specified, randomized, a year, adherence-supported, both arms good diets. ⚠️ The last is crucial: both arms were told to maximize vegetables and minimize added sugar and refined flour. §35.3b.

8. b) Similar averages, enormous individual variation, and neither predictor worked. ⚠️ It is a rare case of a commercially attractive hypothesis being specified in advance and falsified by the people best placed to have found it if it were true. §35.3b.

9. b) Variation is real and large; the test didn't explain it. ⚠️ This pairing is routinely misreported in both directions — as "diets are all the same" and as "everyone is different so buy a test." Neither follows. §35.3b.

10. b) No reference standard. ⚠️ Plus method dependence (Ch 27 §27.3b), within-person variability, and advice that converges across customers on eat more plants, fibre, variety and fermented food — which is good advice, and is what you would have been told without the test. §35.4.

11. b) 🟡 Unclear. ⚠️ Three honest positions held simultaneously: it shows you something real, there is no outcome evidence in non-diabetic people, and the interpretation problem is severe. §35.5.

12. b) Avoiding fruit, legumes and whole grains. ⚠️ A person optimizing for a flat curve reduces exactly the foods Parts II, III and V spent the book recommending — which is Chapter 2's error with a subscription attached. §35.5, and Theo's report scored cheese and salami green.

13. b) Normal physiology. ⚠️ Regulation is working when glucose rises and returns. ⚠️ There is a version of this technology that teaches people to be alarmed by health. §35.5, §35.12.

14. b) Unrelated to blood type; the better diets had more plants. ⚠️ The cleanest available demonstration of the chapter's recurring finding: the personalization did nothing and the underlying dietary quality did everything. §35.6.

15. b) Preference, culture, budget and schedule. ⚠️ Chapter 10's threshold — adherence beats composition — which makes matching the diet to what you will actually eat the largest lever available. ⚠️ It requires no test, produces larger effects than anything in §35.3–35.6, and cannot be sold as a subscription. §35.7.

16. b) Replicate crossover, twice. ⚠️ Evidence for true individual response is reasonably good for postprandial glucose and much weaker or absent for many outcomes marketed on the responder concept. §35.8.

17. d) Publication bias. ⚠️ The four are measurement error, regression to the mean, differential adherence, and genuine individual response — and only the last is what the products are selling. §35.8.

18. b) Decide in advance what would make you stop. ⚠️ It is what separates an experiment from a justification. §35.9.

19. b) Hard-coded conventional constraints. ⚠️ Part of the "personalized" output is conventional advice, built in to prevent the algorithm recommending a worse diet with flatter curves. ⚠️ The other two reasons: the underlying pattern really is robust, and the data does not yet support more than marginal personalization. §35.10.

20. b) The authority of personal data. ⚠️ A conventional diet's rules can be argued with; a rule generated from your own biology cannot easily be. ⚠️ For someone whose restriction has become rule-governed (Ch 34 §34.9), a device generating continuous personalized justifications for narrowing the diet is close to the worst available purchase. §35.12.

Scoring

Three items worth checking regardless of score.

Question 4. ⚠️ Step 4 is the whole sector. If you can spot it in an advertisement, the chapter has done its job.

Question 15. ⚠️ The honest answer to "what should I personalize?" — and it is free.

Question 19. ⚠️ This is the observation Chapter 37 is built on, so notice it now.