These are notes from a lecture I gave to a grad writing class in EcoEvo at UCI.
- What is a review paper?
- Critical evaluation of published literature to develop new conclusions or insights
- Covers a general topic (> 100 papers)
- Targeted toward broad audience, especially students
- What is a great review paper?
- Evaluates the implicit but unexamined assumptions of an important body of work
- Lays the foundations for a new research direction
- Provides a new understanding of a rapidly evolving field
- Why write a review paper?
- To provide intellectual leadership in a research area
- To resolve contradictory evidence
- To get cited
- Identifying a good subject. Look for:
- Many research papers cite exclusively an earlier opinion paper that was based on minimal evidence
- “xxxxx is a matter of debate.”
- “Researchers have often noted [pattern x], but the underlying mechanisms are unclear.”
- “Responses of xxxxx to yyyyy are inconsistent among studies.”
- In several papers on a topic, authors can’t really explain their results & invoke a scapegoat mechanism
- Reviews versus standard papers
- Structure of standard papers:
- Introduction
- Hypothesis
- Methods
- Results
- Discussion
- Conclusion
- Structure of reviews:
- Introduction greatly expanded
- Instead of hypothesis, identify unknown/problem/question
- No methods or results (unless it’s a synthesis or meta-analysis)
- Discussion expanded
- Conclude with new hypothesis/conceptual framework (sometimes)
- Structure of standard papers:
- Nuts & bolts
- No standard organization for review papers
- Your topics must be logically organized, though
- Thematic
- Chronological
- Methodological
- First step: Develop outline of topics to be covered in review
- Early in the process, submit outline to journal editors as a proposal
-
- Problems to avoid
- Lack of analysis and commentary
- Not stating the unknown/problem/question
- Lack of logical organization of topics
- Not objectively written
- Referencing errors
- Contribution to field unclear
- How can you ensure your review contributes new knowledge?
- Develop new conceptual framework
- Identify and prioritize new ways forward
- Propose new underlying mechanisms
- Resolve a controversy
- Integrate knowledge from multiple disciplines
- Conduct synthesis or meta-analysis
- A synthesis tests relationships/patterns by compiling data from multiple publications
- A meta-analysis is a formal statistical approach that combines results from several studies
- Appropriate to test general trends:
- “Does x consistently alter y across studies?”
- “How much does x alter y?”
- “What factors modify the effect of x on y?”
- Need many studies that have tested the same question (>25)
- Appropriate to test general trends:
- Basic approach to a meta-analysis:
- Convert results from each study into consistent metric: “Effect size”
- Example: % change relative to control
- Weight each study by number of replicates or variance
- Calculate weighted average across studies
- For specifics, see Gurevitch & Hedges 2002
- Problems to avoid