Lecture Authoring Guide

Standards for structuring lecture notebooks across the Business Decisions course. Each lecture group follows a specific pedagogical pattern.

Measure Impact Lectures

Reference Header (REQUIRED)

Every measure-impact notebook MUST begin with a reference block:

# [Lecture Title]

> **Reference:** *Causal Inference: The Mixtape*, Chapter X: [Chapter Title] (pp. XX-XX)

This lecture introduces [framework/method]. We apply these concepts using
the Online Retail Simulator to answer: **[business question]?**

Source Material

The Mixtape chapters are available at _external/books-mixtape/. Use these PDF chapters as the primary source for the theory sections. The theory content should closely follow the book's:

• •Notation and terminology
• •Order of topic presentation
• •Key examples and intuitions
• •Mathematical derivations

Structure Pattern: Theory → Application

Part I: Theory (from Mixtape chapter)

• •Introduce framework/method with formal definitions
• •Present notation (potential outcomes, estimators)
• •Explain identification assumptions
• •Include key equations and intuition

Part II: Application (from understand-domain)

• •Frame a business question (e.g., "Does improving product content quality increase sales?")
• •Use Online Retail Simulator to generate data with known ground truth
• •Demonstrate the method with the running example
• •Show what works and what fails (selection bias, etc.)
• •Connect results back to theory

Narrative Style for Theory Sections

Measure Impact Theory sections require detailed, textbook-quality exposition:

Prose Depth

• •Each concept gets thorough explanation with multiple paragraphs
• •Explain the "why" behind each method, not just the "what"
• •Build intuition before presenting formal definitions
• •Connect new concepts to previously introduced material

Mathematical Exposition

• •Use LaTeX equations for all mathematical notation
• •Show derivations step-by-step, not just final results
• •Explain each term in equations when first introduced
• •Present formal definitions with precise notation

Structure

• •Use tables to organize definitions, parameters, and comparisons
• •Include worked examples where concepts benefit from concrete illustration
• •Decompose complex ideas into named components (e.g., "Baseline Bias", "Differential Treatment Effect Bias")
• •Use subsections to break down major concepts into digestible parts

No Summary Sections

Lectures should focus purely on content. Do NOT include:

• •"Conclusion" sections
• •"Key Takeaways" sections
• •"What's Next" sections
• •"Summary" sections

Let the content speak for itself. Students absorb concepts through the theory and application—wrap-up sections add length without value.

Chapter-to-Lecture Mapping

Note: Generalized Roy Model draws from external sources (Heckman & Vytlacil).

Understand Domain Lectures

Different pattern—introduce tools/systems, not theory-first.

Structure Pattern: Tool → Exploration

1. •Introduce the tool/system and its purpose
2. •Show configuration and usage
3. •Explore generated data interactively
4. •Connect to business questions that motivate later lectures

Cross-References

• •Code: Use /review-code for Python standards
• •Writing: Use /review-writing for prose and formatting

Lecture Summary Guidelines

When writing lecture summaries in index.md files, follow this standardized pattern.

Standard Format (2-3 Sentences)

Sentence 1: We introduce/discuss/present [method/framework/tool] as [its role/purpose].

Sentence 2: This section [formalizes/explores/examines/emphasizes] [what problem/concept it addresses].

Sentence 3 (optional): The goal is to [learning objective] OR [connection to broader framework].

Examples

Potential Outcomes Model:

We introduce the potential outcomes framework as the foundational model for causal inference.
This section formalizes the fundamental problem of causal inference—missing counterfactuals—and
explains why randomization resolves it. The goal is to establish a precise language for defining
causal effects and understanding what can and cannot be identified from data.

Instrumental Variables:

We introduce instrumental variables as a method for causal inference when unobserved confounding
is present. The focus is on identification assumptions, interpretation of local average treatment
effects, and the role of instruments within the Generalized Roy framework.

Key Characteristics

Tone:

• •Conceptual and framework-focused (not action/implementation-focused)
• •Theoretical emphasis on understanding (not building/implementing)

Content:

• •Emphasize identification assumptions, learning objectives, frameworks
• •Connect to broader course context (Generalized Roy Model, business decisions)
• •Explain what problem the method solves

Structure:

• •Always 2-3 sentences (no more, no less)
• •Use verbs: introduce, discuss, present, explore, examine (avoid: implement, build, create)
• •Focus on "what you understand" not "what you produce"

Common Patterns

Do:

• •Start with "We introduce [X] as [purpose]"
• •Use "This section formalizes/explores/examines..."
• •End with "The goal is to..." or "The focus is on..."
• •Mention identification assumptions where relevant
• •Connect to Generalized Roy framework (for Measure Impact)

Don't:

• •List technical components without context
• •Use action verbs like "implement," "build," "create"
• •Write single-sentence summaries
• •Focus on tools/outputs instead of concepts/understanding
• •Exceed 3 sentences

Review Checklist

When reviewing a measure-impact lecture, verify:

Structure

• • Reference header present with correct chapter and page numbers
• • Theory section covers Mixtape chapter concepts
• • Application section uses Online Retail Simulator
• • Business question clearly stated

Content Alignment

• • Theory matches the referenced Mixtape chapter
• • Application demonstrates the method with simulated data
• • Results connect back to theory (e.g., selection bias shown)

Completeness

• • Key definitions and notation introduced
• • Identification assumptions explained
• • At least one worked example with code
• • Interpretation of results discussed