Quiz Generator

Phase 2: Question Generation

Generate questions at three difficulty levels:

Easy (knowledge recall + basic comprehension):

• "What is the main identification strategy in [Paper]?"
• "Which variable serves as the instrument for X?"
• "What does the exclusion restriction require in this context?"

Medium (application + analysis):

• "If the exclusion restriction fails because [specific channel], what bias would result?"
• "Which of the following would threaten the parallel trends assumption?"
• "The authors find [result]. Which mechanism is most consistent with this?"

Hard (synthesis + evaluation):

• "A critic argues [specific objection]. Which robustness check most directly addresses this?"
• "Comparing [Paper A] and [Paper B], which identification strategy is more convincing for [question] and why?"
• "If you were to replicate this study in [different context], which assumption would be most at risk?"

Phase 3: Quality Filter (Student Reviewer)

Review each question as if you were a diligent student:

• Is the question clear and unambiguous?
• Are the distractors genuinely plausible (not obviously wrong)?
• Does the explanation help learning (not just state the answer)?
• Is there exactly one defensible correct answer?

Drop questions that fail any of these checks.

Phase 4: Output

For in-class quizzes (5-7 questions, 10 minutes)

## Quiz: [Topic]
### Lecture [N]: [Title]

**Q1.** [Stem]
- A) [Choice]
- B) [Choice]
- C) [Choice]
- D) [Choice]

*Correct: B*
*Explanation: [Why B is correct, why A/C/D are wrong]*
*Learning objective: [One sentence]*

[... repeat for all questions]

For online self-assessment (10-15 questions, mixed difficulty)

Same format but with difficulty tags and organized by topic/section.

For LaTeX/Beamer integration

\begin{frame}{Concept Check}
\textcolor{highlight}{Which of the following best describes the exclusion restriction in [Paper]?}

\pause
\begin{enumerate}[A)]
  \item [Distractor 1]
  \item \textcolor{usgblue}{\textbf{[Correct answer]}}
  \item [Distractor 2]
  \item [Distractor 3]
\end{enumerate}

\pause
\source{Based on [Author (Year)]}
\end{frame}

Question Type Templates

Identification Strategy Questions

• "What is the source of exogenous variation exploited by [Author]?"
• "Which assumption is required for the IV estimate to be consistent?"
• "What would violate the exclusion restriction in this design?"

Results Interpretation Questions

• "The coefficient on X is [value]. This means that..."
• "The first-stage F-statistic is [value]. This suggests..."
• "The authors find heterogeneous effects by [dimension]. The most likely explanation is..."

Design Comparison Questions

• "Compared to an OLS estimate, the IV estimate of [effect] is likely [larger/smaller] because..."
• "Why do the authors prefer DiD over cross-sectional comparison?"
• "What advantage does the RDD design offer over the natural experiment in [Paper]?"

Critical Thinking Questions

• "A referee argues that [specific concern]. The strongest response would be..."
• "If the treatment effect is heterogeneous, the LATE estimates..."
• "The external validity of these findings is limited because..."

Parameters

• n_questions: Number of questions (default: 6 for in-class, 12 for online)
• difficulty_mix: Distribution across easy/medium/hard (default: 2/3/1 for in-class, 3/5/4 for online)
• focus: Optional specific topic or paper to focus on
• output_format: "markdown" (default), "latex" (for Beamer), "json" (for webapp)

Integration with Course

• Questions should reference specific lectures and papers from READING_LIST.md
• Use the same notation and terminology as the slides
• Link to specific identification strategies taught in the course
• Build progressively: early quizzes test foundations, later ones test synthesis