com.atlassian.confluence.content.render.xhtml.migration.exceptions.UnknownMacroMigrationException: The macro 'datalayer.push(arguments);' is unknown.

3.7.6.3 Data extraction

Owned by Carolyn Phillips
Mar 26, 2024
2 min read

Standardized data extraction tools that promote extraction of similar data form all of the included studies and are recommended. The review should detail what data the reviewers extracted from the included studies. The a priori protocol or this Manual with the original data extraction tool can be cited to indicate the tool used. Data extraction in a meta aggregation is a multi phase process, with the general details of papers, including the citations details, the population, phenomena of interest, and context as well as methodology, methods, settings and cultural information retrieved from papers before moving to extraction of the findings. The approach and process used to extract findings from the results of the included studies should be presented with enough detail to be readily reproducible. Indicate what data were considered findings in the review (i.e. themes, metaphors, etc.) and the process by which findings were identified (i.e. repeated reading of text).

Extracting findings is both the second phase of data extraction, and the first step in data synthesis.

  • A finding is defined as a verbatim extract of the authors analytic interpretation accompanied by either a participant voice, or fieldwork observations or other data.

Each finding extracted is to be accompanied by an illustration from the same text that informs the finding.

  • An illustration may be either a direct quotation of participant voice, field-work observations or other supporting data

Levels of credibility

As a finding is extracted and its accompanying illustration entered in the JBI SUMARI software, a level of ‘Credibility’ is allocated based on the reviewers perception of the degree of support each illustration offers for the specific finding it is associated with.

There are 3 levels of credibility as described below, and reviewers should document in this section of their review report HOW the decision was made to allocate these levels, and what (if any) issues arose during the process, or whether there was good agreement between the review team members.

  • Unequivocal (findings accompanied by an illustration that is beyond reasonable doubt and; therefore not open to challenge);

  • Credible (findings accompanied by an illustration lacking clear association with it and therefore open to challenge)

  • Not Supported (findings are not supported by the data).