Table 1.

Comparison of attributes of Classic Systematic Review, Classic Weight of Evidence, and an integrated approach.

Attribute	Classic SR	Classic WoE	Integrated SR & WoE
Emphasis	Transparent and unbiased assembly of information	Hypothesis best supported by available information	Scientific rigor while accommodating many situations
Generality of results	General applicability (e.g., the chemical is a carcinogen)	General or case-specific (e.g., the chemical caused this cancer cluster)	General or case specific
Institutions	Institutions specify methods and compile results (e.g., Cochrane)	None; users define methods	Some government agencies recommend for specific applications
Consistency	Consistent methods within fields	Diverse methods even within fields	Consistent framework with diverse options
Sources of information	Published experiments from literature	Literature, purposive studies and models, data bases, etc.	Any type of information
Types of evidence	One per assessment or, if more than one, assessed separately	Usually more than one	Usually more, because most questions cannot be answered with only one type
Implications of evidence	One type of study with one implication	Multiple types of evidence have different implications for hypotheses	Usually multiple types of inferences
Meta-analysis	Standard inferential method	Seldom used	Encouraged when appropriate
Causation	Not an issue because the experiments that answer the question are inherently causal	Causal inference from heterogeneous evidence that seldom experimentally answers the assessment question	Recommends distinct assessment to establish causation and then use the causal relationship to make predictions.
Role of rating	Used to express risk of bias or other qualities, but not for inference	Implied by the concept of weighting but seldom employed	Recommended for transparency of weighting evidence and drawing inferences
Role of expertise	Expertise needed but latitude minimized by detailed methods and statistical inference	Expert knowledge and judgment are essential and explicit	Expert knowledge and judgment are essential and explicit
Tools	Software tools for literature searching, screening search results, and extracting information	No known software tools for automating steps in environmental assessments	Software tools for literature search, screening and extraction.