← MOST study notes index · CADIO / NYU School of Global Public Health
MOST from a Methodological Perspective: Module 7
Study notes · CADIO / NYU School of Global Public Health
Reading the answer key in this module (source notes; flagged, not silently corrected). The payload represents the answer key twice: an item-level field naming the single answer the quiz uses, and an option-level corrects list with per-answer flags. The two disagree on 11 of the 15 items. The marks below follow the item-level field, because that is what the published quiz marks and scores: completing the assessment in a browser on 14 September 2026 with these answers scored 100%, and every revealed answer matched the marks used here. The option-level data is preserved and flagged below as a source anomaly. For question 2 the option-level list adds an extra correct option; it is shown unmarked, following the scored key. The source provides no feedback text for any item, so none is reproduced and none is inferred. Review pointers are my own cross-references to the module notes (agent synthesis).
What this module is, in one paragraph. Module 7 is the course’s final assessment; its overview sentence (quoted below) states that it covers content from Modules 1 to 6 across fifteen questions. There is no teaching material here, the module is the examination itself, so these notes reproduce all fifteen items with the answer key the course scores, the option marks exactly as the published quiz applies them, and nothing invented on top. Use it as a self-test: answer each item from memory first, then check the keyed option and the review pointer, which names the lesson that taught the topic. The remaining sections add the scored key in the course’s own words, a synthesis of what the assessment covers (including the answer-key data anomaly), a glossary of the tested concepts, and the sources.
Course overview: “The final assessment covers content from Modules 1-6 across fifteen questions.”Course description, course.json (source text)
Final Assessment: the fifteen questions with the scored answer key
Each item below reproduces the published question text and all four options exactly; the option marked ☑ is the answer the course scores as correct, taken from the item-level answer key in course.json. The course provides no feedback text for any item. “Review pointer” lines are my own cross-references to the module notes where the topic is taught (agent synthesis, not source text).
Source data inconsistency (all fifteen items). Source data inconsistency: the course payload’s option-level corrects list and per-answer flags disagree with the scored answer key in 11 of the 15 items: the scored key is absent from the corrects list in 10 of them, and the corrects list carries an extra option in 1 of them. The answer marks reproduced here follow the scored key (the item-level correct field, which is what the published quiz marks and scores); the disagreement is documented in the evidence file and neither source field is altered. For question 1 the disagreeing option-level data marks option A, while the scored key (option B) states the intervention EASE definition taught in Module 1.
Question 1
MULTIPLE CHOICE: Intervention EASE refers to (select ONE):
- ☐ The direct connection between the conceptual model and the intervention components.
- ☑ A strategic balance of effectiveness, affordability, scalability, and efficiency.
- ☐ How easy it is to implement an intervention across a wide variety of settings.
- ☐ How readily an intervention can be adapted to suit a particular environment.
Review pointer (agent synthesis): taught in Module 1, Lesson 5: Intervention EASE.
Question 2
MULTIPLE CHOICE: The three phases of MOST are (select ONE):
- ☐ Pilot study, ERCT, and dissemination.
- ☐ Screening, refining, and confirming.
- ☐ Conceptual model, pilot study, optimization trial.
- ☑ Preparation, optimization, and evaluation.
Source inconsistency (question 2). Source inconsistency: this is a single-answer question, but the payload’s option-level corrects list flags an additional option as correct. The mark shown follows the scored key; the extra flagged option is recorded as a source anomaly in the evidence file, and no intended answer is inferred.
Review pointer (agent synthesis): taught in Module 1, Lesson 6: The three phases of MOST.
Question 3
MULTIPLE CHOICE: Which ONE of the following experimental designs is NOT typically used in the optimization phase of MOST?
- ☐ Factorial design
- ☑ ERCT
- ☐ Sequential multiple-assignment randomized trial (SMART)
- ☐ Micro-randomized trial
Review pointer (agent synthesis): taught in Modules 1 and 5: Designs used in the optimization phase versus the evaluation phase.
Question 4
MULTIPLE CHOICE: The resource management principle states that (select ONE):
- ☑ The investigator must make the best use of research resources to gain the most, and the most reliable, scientific knowledge.
- ☐ An optimization trial should always be done as cheaply as possible.
- ☐ It is important to establish a budget for intervention implementation.
- ☐ It is a good idea to hire a financial professional to manage research resources.
Review pointer (agent synthesis): taught in Module 1, Lesson 6: The resource management principle.
Question 5
MULTIPLE CHOICE: The continual optimization principle states that (select ONE):
- ☐ Once an intervention has been optimized, its effectiveness will continue.
- ☐ It is not possible to conduct more than one optimization trial in a single optimization phase of MOST.
- ☑ Optimization is an ongoing process; interventions can and should be improved repeatedly over time.
- ☐ Only continuous variables should be used as outcomes in an optimization RCT.
Review pointer (agent synthesis): taught in Module 1, Lesson 6: The continual optimization principle.
Question 6
MULTIPLE CHOICE: Consider a factorial optimization trial, in which each factor has two levels, “no” and “yes.”The term main effect, as used in this course, can be interpreted as (select ONE):
- ☐ The effect of a factor when all of the other factors in the experiment are set to “no.”
- ☐ The effect of a factor when all of the other factors in the experiment are set to “yes.”
- ☐ The effect of a factor as compared to that of the other factors in the experiment.
- ☑ The effect of a factor on average, across all levels of all of the other factors in the experiment.
Review pointer (agent synthesis): taught in Module 3, Lessons 1 and 2: The main effect under effect coding.
Question 7
MULTIPLE CHOICE: Which ONE of the following best summarizes the role of the ERCT in MOST?
- ☐ It is recommended that investigators working within the MOST framework never conduct ERCTs.
- ☐ An ERCT should be conducted first, before an intervention is optimized, to establish that there is an overall treatment effect.
- ☐ The ERCT is one type of optimization RCT design.
- ☑ The ERCT is used in the evaluation phase of MOST.
Review pointer (agent synthesis): taught in Modules 1 and 2: The role of the ERCT.
Question 8
MULTIPLE CHOICE: Which ONE of the following statements comparing factorial experiments and ERCTs is NOT correct?
- ☐ In an ERCT, effects are estimated by comparing the means of individual experimental conditions, whereas in a factorial experiment, effects are estimated by comparing the means of particular combinations of experimental conditions.
- ☐ Factorial experiments enable estimation of interactions between intervention components, whereas this is typically not possible in an ERCT.
- ☐ The ERCT is powered to test the effectiveness of the intervention as a package, whereas a factorial experiment is typically not powered for this.
- ☑ As compared to an ERCT, a factorial experiment will typically require a much larger sample size to achieve adequate power.
Review pointer (agent synthesis): taught in Modules 3 and 4: Factorial versus ERCT power and sample size.
Question 9
MULTIPLE CHOICE: Which ONE of the following statements is NOT correct?
- ☐ Statistical power is an important consideration when planning an optimization RCT.
- ☑ Statistical power is an important consideration when planning a pilot study.
- ☐ Optimization trials are not pilot studies.
- ☐ It is a good idea to conduct a pilot study to prepare for an optimization RCT.
Review pointer (agent synthesis): taught in Module 4, Lesson 1: Pilot studies and statistical power.
Question 10
MULTIPLE CHOICE: Which ONE of the following statements related to the optimization objective is correct?
- ☑ The optimization objective expresses what is meant by intervention EASE in a particular study
- ☐ The optimization objective provides a definition of the best possible intervention
- ☐ Only a single optimization objective can be applied to the results of an optimization trial.
- ☐ The optimization objective is selected in the evaluation phase of MOST
Review pointer (agent synthesis): taught in Module 1 (preparation phase): The optimization objective.
Question 11
MULTIPLE CHOICE: Which ONE of the following is NOT characteristic of the factorial optimization trial?
- ☑ It is necessary to identify a single control condition for the entire experiment, which can be challenging.
- ☐ When the design is perfectly balanced and effect coding is used properly in data analysis, the effect estimates are uncorrelated.
- ☐ Effect estimates are made based on combinations of experimental conditions.
- ☐ The design yields estimates of the main effect of each factor and interactions between factors.
Review pointer (agent synthesis): taught in Module 3, Lesson 5: Experimental control in the factorial design.
Question 12
MULTIPLE CHOICE: Which ONE of the following statements is INCORRECT?
- ☐ When an investigator is working from the conclusion-priority perspective, it is usually important to select a Type I error rate no larger than .05.
- ☐ An investigator working from the decision-priority perspective must consider how painful or expensive Type II errors are relative to Type I errors, and may wish to adjust these error rates accordingly.
- ☑ In the conclusion-priority perspective, investigators usually try to keep the Type I and Type II error rates about the same.
- ☐ The decision-priority perspective is usually appropriate for the optimization phase of MOST.
Review pointer (agent synthesis): taught in Module 6, Lesson 4: Type I and Type II error priorities in the conclusion and decision perspectives.
Question 13
MULTIPLE CHOICE: Which ONE of the following statements is TRUE?
- ☐ Clustering in data occurs only when participants are in pre-existing groups, such as families or schools.
- ☑ Contamination between experimental conditions tends to reduce treatment effects.
- ☐ When conducting within-cluster random assignment, it is usually a good idea to implement only a few experimental conditions within each cluster.
- ☐ Within-cluster random assignment can help prevent contamination.
Review pointer (agent synthesis): taught in Module 5, Lesson 1: Contamination between experimental conditions.
Question 14
MULTIPLE CHOICE: Which ONE of the following statements about factorial optimization trial is INCORRECT?
- ☑ It is necessary to assume that the components are completely independent and do not interact.
- ☐ It is necessary to assume that the components can be experimentally manipulated independently.
- ☐ If a pilot test indicates that an experimental condition cannot be implemented, it is necessary to redesign the experiment.
- ☐ Implementation of a particular level of a factor should be exactly the same across experimental conditions.
Review pointer (agent synthesis): taught in Module 6, Lesson 1: The independent-manipulability assumption.
Question 15
MULTIPLE CHOICE: Which of the following is NOT an example of a situation in which a cluster structure is likely to be present?
- ☐ The optimization trial is taking place in a school.
- ☑ Participants may occasionally interact while waiting to complete measures.
- ☐ One of the candidate components being examined is to be delivered in a group setting.
- ☐ The intervention is to be delivered to married and cohabiting couples.
Review pointer (agent synthesis): taught in Module 5, Lesson 1: Cluster structures in practice.
The scored keys in the course’s own words
Each block quotes verbatim, from the source payload, the option that the published quiz scores as correct for that question. This is the same key as the marks above, as a revision list.
Question 1 key: “A strategic balance of effectiveness, affordability, scalability, and efficiency.”Scored answer key, question 1 (course.json)
Question 2 key: “Preparation, optimization, and evaluation.”Scored answer key, question 2 (course.json)
Question 3 key: “ERCT”Scored answer key, question 3 (course.json)
Question 4 key: “The investigator must make the best use of research resources to gain the most, and the most reliable, scientific knowledge.”Scored answer key, question 4 (course.json)
Question 5 key: “Optimization is an ongoing process; interventions can and should be improved repeatedly over time.”Scored answer key, question 5 (course.json)
Question 6 key: “The effect of a factor on average, across all levels of all of the other factors in the experiment.”Scored answer key, question 6 (course.json)
Question 7 key: “The ERCT is used in the evaluation phase of MOST.”Scored answer key, question 7 (course.json)
Question 8 key: “As compared to an ERCT, a factorial experiment will typically require a much larger sample size to achieve adequate power.”Scored answer key, question 8 (course.json)
Question 9 key: “Statistical power is an important consideration when planning a pilot study.”Scored answer key, question 9 (course.json)
Question 10 key: “The optimization objective expresses what is meant by intervention EASE in a particular study”Scored answer key, question 10 (course.json)
Question 11 key: “It is necessary to identify a single control condition for the entire experiment, which can be challenging.”Scored answer key, question 11 (course.json)
Question 12 key: “In the conclusion-priority perspective, investigators usually try to keep the Type I and Type II error rates about the same.”Scored answer key, question 12 (course.json)
Question 13 key: “Contamination between experimental conditions tends to reduce treatment effects.”Scored answer key, question 13 (course.json)
Question 14 key: “It is necessary to assume that the components are completely independent and do not interact.”Scored answer key, question 14 (course.json)
Question 15 key: “Participants may occasionally interact while waiting to complete measures.”Scored answer key, question 15 (course.json)
Course Feedback Survey (lesson 2)
The second lesson is a short feedback survey. The course text, quoted:
Survey invitation: “We value your feedback and use it to improve this course.” “Please take a moment to complete this short survey:”Course Feedback Survey lesson, text block (course.json)
Survey button: “GO TO SURVEY”Course Feedback Survey lesson, button label (course.json)
The button links to the course survey form: https://nyu.qualtrics.com/jfe/form/SV_8C9xQE5f3hmqr66.
Synthesis (my own reading: not part of the course)
The fifteen items are the course’s own coverage check, and the balance of topics shows what the course regards as essential: seven items test the foundations from Module 1 (EASE, the three phases, the two principles, the optimization objective, and where the ERCT belongs), two test the factorial machinery from Module 3 (the effect-coded main effect and the control structure of the design), two test powering from Module 4 (the factorial-versus-ERCT economy and the pilot-study logic), three test the extended design repertoire from Module 5 (alternative optimization designs and cluster structure, including contamination), and two test responsible conduct from Module 6 (the independent-manipulability assumption and the error-rate priority question). There is no arithmetic in the assessment: the numbers that matter (per-level n, design effects, corrected alphas) are exercised in the module lessons rather than here.
The most consequential finding in preparing these notes concerns the source data rather than the questions. The payload represents each item’s answer key twice: once as an item-level field naming the single answer the quiz uses, and once as an option-level list with per-answer flags. On this module the two representations disagree on eleven of the fifteen items, in ten of them the option-level list marks a different answer, and in one (question 2) it marks the scored answer plus an additional option. The published quiz follows the item-level field: completing the assessment in a browser and answering with the item-level key produced a score of 100%, with every revealed answer matching that key, including the ten items where the option-level data pointed elsewhere. The option-level data is therefore recorded as a source anomaly and left exactly as found. The marks shown in these notes follow the item-level key, and no intended answer is inferred for any conflict.
Three individual keys are worth a second look because the unmarked options are the plausible mistakes the course is guarding against. Question 6 contrasts the effect-coded interpretation of a main effect (an effect averaged across the levels of the other factors, which Module 3 adopts) with the dummy-coded first-order reading (an effect with the other factors set to “no”); the course is explicit that the second quantity should not be called a main effect. Question 12 separates the conclusion-priority and decision-priority stances on Type I and Type II error rates; the statement keyed as incorrect is the suggestion that conclusion-priority investigators keep the two rates about the same. Question 14 tests the independent-manipulability assumption against the much stronger claim that components do not interact, which the course never requires. On question 2 the stale option-level data would have added a second correct option; the scored key keeps the single answer the course teaches for the three phases of MOST.
Figure 1: Assessment coverage map (agent synthesis)
| Question | Topic tested | Where taught |
|---|
| 1 | Intervention EASE | Module 1, Lesson 5 |
| 2 | The three phases of MOST | Module 1, Lesson 6 |
| 3 | Designs used in the optimization phase versus the evaluation phase | Modules 1 and 5 |
| 4 | The resource management principle | Module 1, Lesson 6 |
| 5 | The continual optimization principle | Module 1, Lesson 6 |
| 6 | The main effect under effect coding | Module 3, Lessons 1 and 2 |
| 7 | The role of the ERCT | Modules 1 and 2 |
| 8 | Factorial versus ERCT power and sample size | Modules 3 and 4 |
| 9 | Pilot studies and statistical power | Module 4, Lesson 1 |
| 10 | The optimization objective | Module 1 (preparation phase) |
| 11 | Experimental control in the factorial design | Module 3, Lesson 5 |
| 12 | Type I and Type II error priorities in the conclusion and decision perspectives | Module 6, Lesson 4 |
| 13 | Contamination between experimental conditions | Module 5, Lesson 1 |
| 14 | The independent-manipulability assumption | Module 6, Lesson 1 |
| 15 | Cluster structures in practice | Module 5, Lesson 1 |
My own cross-reference of the fifteen items to the lessons that taught each topic. The assessment itself states only that it covers Modules 1 to 6; module numbers refer to the methodological course and its notes pages.
Glossary
- Intervention EASE
- The strategic balance of effectiveness, affordability, scalability and efficiency that an optimized intervention is meant to achieve; the definition the course builds Module 1 around.
- Three phases of MOST
- Preparation, optimization, and evaluation.
- Resource management principle
- Make the best use of research resources; select the optimization trial design that most efficiently addresses the research questions.
- Continual optimization principle
- Optimization is an ongoing process: interventions can and should be improved repeatedly, and the evaluation phase feeds back into preparation.
- Optimization objective
- The statement, chosen in the preparation phase, that expresses what intervention EASE means for a particular study.
- Factorial optimization trial
- An experiment that crosses factors so that main effects and interactions can be estimated efficiently; the workhorse optimization design (often 2^k).
- Evaluation RCT (ERCT)
- A package-versus-control randomised experiment; the evaluation-phase design, not typically an optimization-phase design.
- Main effect (effect coding)
- The effect of a factor on average across all combinations of the other factors’ levels; in a balanced, effect-coded factorial the effect estimates are uncorrelated.
- Independent manipulability
- The necessary assumption that all factors can be manipulated independently: every combination makes sense and each level is implemented consistently. It does not mean the components fail to interact.
- Conclusion versus decision priority
- Two stances on error rates: conclusion-priority keeps strict scientific conventions (Type I control); decision-priority, appropriate to the optimization phase, prioritises power and decisions based on observed effect sizes.
- Pilot study
- A feasibility study with no hypothesis testing and no power analysis; optimization trials are not pilot studies, though pilot testing an optimization trial first is recommended.
- Cluster structure (nesting)
- Participants belonging to clusters such as schools, clinics or families; clusters may be pre-existing or experiment-induced, and they change the arithmetic of power and contamination.
- Contamination
- Treatment information or resources bleeding across experimental conditions, which tends to reduce treatment effects.
- Alternative optimization designs
- Designs such as SMART and micro-randomized trials used in the optimization phase alongside the factorial family; the ERCT is not among them.
Sources
- Course materials: CADIO, MOST from a Methodological Perspective, Module 7 (“Final Assessment”), share link share.articulate.com/g4PWX_CeVMVcvcJICQ925 (resolved from bit.ly/most-mp-assessment; resolution re-verified in a browser on 14 Sep 2026), two lessons: a fifteen-question single-choice final assessment and a course feedback survey. The module contains no video or slide assets.
- Course feedback survey: https://nyu.qualtrics.com/jfe/form/SV_8C9xQE5f3hmqr66 (linked from the second lesson).
- Module notes cross-references: Module 1, Module 2, Module 3, Module 4, Module 5, Module 6 (same site, methodological course).
- Verification record: evidence.json in the build directory (course.json sha256, live-quiz audit, checker output, negative-control matrix), retained with the task handoff.
Prepared 14 September 2026. The fifteen assessment items are reproduced from the course payload (course.json) exactly, including question wording, option text and the scored answer key; the marks follow the item-level key that the published quiz applies, verified live on 14 September 2026 (the assessment was completed in a browser; score 100%; every revealed answer matched the marks used here). The payload’s option-level corrects list and flags disagree with that key on eleven of the fifteen items; the disagreement is flagged above and recorded in the build evidence, not corrected. This module has no videos, slide decks, transcripts or images, so no narration quotes or slide figures exist for it; the quoted source text is verified against the course payload itself. The fail-closed checker verified complete question/option coverage, exact normalised question and option text, the marks, quote provenance and the required anomaly notes; private negative controls were exercised. The checker does not establish the scientific correctness of all prose; review pointers, the synthesis and the glossary are labelled agent synthesis.