AI Power-Analysis Narrative: Drafting Sample-Size Justification Sections

End-of-lesson check

15 questions · take it digitally for instant feedback at tendril.neural-forge.io/learn/quiz/end-research-ai-and-power-analysis-narrative-r7a3-creators

1. In drafting a power-analysis narrative for a two-arm RCT, which task can AI reliably perform?

   1. Performing the statistical calculations that determine sample size
   2. Selecting the primary outcome variable for the study
   3. Converting a set of power-analysis assumptions into a coherent paragraph
   4. Deciding which effect size the study should target

2. A researcher is planning a two-arm RCT with a continuous primary outcome. What is the single most consequential driver of the required sample size?

   1. The number of participating research sites
   2. The assumed effect size for the primary outcome
   3. The duration of the follow-up period
   4. The choice of statistical software

3. Why cannot AI independently determine the effect-size assumption for a study?

   1. Power analysis software already contains the correct effect-size values
   2. Effect-size assumptions require approval from the funding agency
   3. The effect size must reflect clinically meaningful differences that only investigators can define
   4. AI lacks access to the raw data needed for effect-size estimation

4. What does the lesson identify as a key capability of AI in power-analysis narrative drafting?

   1. Selecting appropriate statistical tests for the primary outcome
   2. Determining whether the study design is feasible
   3. Mirroring the sensitivity-analysis approach across multiple plausible effect sizes
   4. Calculating the exact sample size needed without any investigator input

5. In a power-analysis narrative for a two-arm RCT, which of the following elements must the investigator provide to AI before drafting can begin?

   1. A list of all potential confounders
   2. The statistical code in R or Python
   3. The effect-size assumption, alpha level, and desired power
   4. The complete raw dataset for analysis

6. A researcher asks an AI to determine the sample size for their two-arm RCT. What is the most accurate description of what AI can actually produce?

   1. A decision on whether the study is ethically sound
   2. A narrative justifying the sample size based on assumptions the researcher provides
   3. A completed manuscript ready for submission
   4. The final sample-size number without any justification text

7. What is sensitivity analysis in the context of power analysis for an RCT?

   1. Calculating sample-size requirements across a range of plausible effect sizes
   2. Testing how the study performs with different random assignment methods
   3. Comparing results from multiple statistical software packages
   4. Estimating the impact of different attrition patterns after data collection

8. A student claims that AI can replace the need for a biostatistician in designing an RCT. What is the correct counterargument?

   1. AI cannot decide the effect-size assumption, which is fundamental to the calculation
   2. AI cannot generate random allocation sequences
   3. AI cannot perform interim analyses during the trial
   4. AI cannot recruit participants into the study

9. When AI drafts a power-analysis narrative for a two-arm RCT, what happens to the ownership of the effect-size assumption?

   1. It remains with the research team who must justify their choice
   2. It automatically becomes standard practice in the field
   3. It is assumed to be the largest effect size possible
   4. It transfers to the AI system that generated the text

10. What does the lesson mean when it says the effect-size assumption 'drives the whole study'?

    1. It determines whether the IRB will approve the protocol
    2. It determines the required sample size, which affects feasibility, budget, and timeline
    3. It determines which participants will be eligible for enrollment
    4. It determines which journal will ultimately publish the results

11. A researcher wants AI to draft a power-analysis narrative but has not decided on an effect-size assumption. What should the researcher do first?

    1. Use the largest effect size observed in prior pilot studies
    2. Set the effect size to 0.5 by convention
    3. Ask AI to suggest several effect sizes to choose from
    4. Determine the smallest clinically meaningful difference they want to detect

12. A learner says, 'Since AI can write the power-analysis paragraph, I don't need to understand power analysis concepts.' What is the flaw in this reasoning?

    1. Power analysis is no longer required by ethics committees
    2. AI can independently verify that the study design is ethical
    3. AI always produces accurate statistical calculations
    4. The learner must still provide the correct inputs (effect size, alpha, power) for AI to work with

13. Why might a researcher include multiple effect sizes in their power-analysis narrative rather than a single value?

    1. To confuse reviewers and demonstrate complexity
    2. To meet regulatory requirements for all possible effect sizes
    3. To show how sample-size needs change under different assumptions and to inform interpretation
    4. To ensure the study will definitely be adequately powered

14. What information must be documented in a complete power-analysis justification for a two-arm RCT?

    1. The sample size and the funding amount
    2. Effect-size assumption, alpha, power, attrition rate, and resulting sample size
    3. Only the sample size and the statistical test used
    4. The effect size and the number of co-investigators

15. A researcher provides the following to AI: effect size of 0.3, alpha of 0.05, power of 0.80, and attrition of 10%. What can AI now produce?

    1. A decision about whether to proceed with the study
    2. A completed dataset for the RCT
    3. A final analysis of the study results
    4. A narrative paragraph documenting these assumptions and the resulting sample-size estimate