literature-review

name	literature-review
description	Gather and synthesize academic literature for a research question, then use findings to inform study design and analysis.

Literature Review

Systematic process for finding academic resources relevant to a research question and using them to strengthen study design.

When to use this skill

• At the start of a study, before writing plan.md
• When the user asks to ground their research in existing literature
• When designing survey instruments that should reference established constructs or scales
• When choosing analysis methods and needing precedent

Step 1: Clarify the research question

Before searching, ensure you have a focused research question. Ask the user if needed. A good research question has:

• A clear population or domain (e.g., "US consumers", "software engineers", "LLM agents")
• A specific phenomenon or relationship (e.g., "willingness to pay for privacy", "effect of framing on choices")
• A defined outcome or construct (e.g., "purchase intent", "trust", "response quality")

Write the refined research question to literature/research_question.md in the study directory.

Step 2: Search for literature

Use WebSearch to find relevant academic papers, working papers, and review articles. Run multiple searches with varied queries to get good coverage.

Search strategy

Run at least 3-5 searches using different angles:

# Direct topic search
WebSearch: "{core topic}" site:scholar.google.com

# Key constructs + methodology
WebSearch: "{construct}" survey experiment methodology

# Established scales or measures
WebSearch: "{construct}" validated scale measurement instrument

# Recent reviews or meta-analyses
WebSearch: "{topic}" systematic review OR meta-analysis

# Domain-specific searches
WebSearch: "{topic}" site:ssrn.com
WebSearch: "{topic}" site:arxiv.org
WebSearch: "{topic}" site:nber.org

What to look for

1. Foundational papers — Seminal work that defined the constructs or theory
2. Methodological precedents — Studies that used similar designs (surveys, experiments, conjoint, etc.)
3. Validated instruments — Established question scales or measurement approaches
4. Recent work — Papers from the last 3-5 years showing the current state of knowledge
5. Contradictions or gaps — Areas where findings conflict or where evidence is thin

Step 3: Read and extract key information

For each promising source, use WebFetch to access the content where possible (preprints, open-access papers, abstracts). Extract:

• Citation (authors, year, title, venue)
• Research question / hypothesis
• Method (sample size, design, instruments used)
• Key findings (effect sizes, main results)
• Relevance (how it connects to the current study)

Step 4: Write the literature summary

Create literature/review.md in the study directory with the following structure:

# Literature Review: {Research Question}

## Overview
{2-3 paragraph synthesis of what the literature says about this topic}

## Key Themes

### Theme 1: {name}
{Summary of findings across papers}
- Source 1 (Author, Year): {key finding}
- Source 2 (Author, Year): {key finding}

### Theme 2: {name}
...

## Methodological Insights
{What methods have been used to study this? What worked well?
What sample sizes were typical? What validated scales exist?}

## Gaps and Opportunities
{What hasn't been studied? Where do findings conflict?
What would a new study contribute?}

## Implications for Study Design
{Specific recommendations for how to design the current study,
informed by the literature:}
- Suggested constructs to measure
- Recommended question types or scales
- Sample size considerations
- Potential confounds to control for
- Analysis approaches used in prior work

## References
{Full list of cited sources}

Step 5: Connect to study design

After completing the review, use the findings to:

1. Inform the plan — Reference literature findings in plan.md to justify design choices
2. Adopt validated measures — Use established scales rather than inventing new questions where possible
3. Set expectations — Note expected effect sizes or base rates from prior work
4. Identify controls — Include covariates or conditions that the literature suggests matter
5. Frame the contribution — Articulate what the study adds beyond existing work

File organization

study_root/
  literature/
    research_question.md    # Refined research question
    review.md               # Full literature review
    sources/                # (optional) saved PDFs or key excerpts
  plan.md                   # References literature findings

Tips

• Breadth first, then depth. Start with broad searches to map the landscape, then drill into the most relevant papers.
• Don't over-search. 10-20 well-chosen sources are better than 50 superficial ones. Stop when you start seeing the same papers cited repeatedly.
• Distinguish empirical from theoretical. Note which sources report original data vs. which are reviews or opinion pieces.
• Note methodology details. Sample sizes, response rates, and effect sizes from prior work directly inform your study design.
• Be honest about gaps. If the literature is thin in an area, say so — that's a strong motivation for the study.
• Use author names. When searching, once you find a key paper, search for the authors' other work — they often have related studies.