Step 3: Metadata prescreen

The metadata prescreen refers to the inclusion or exclusion of records based on titles and abstracts (if available). Its main purpose is to reduce the number of records by excluding those that are clearly irrelevant to the review objectives. When in doubt, records can be retained (included provisionally) to decide in step 5, i.e., the screen based on full-text documents.

The prescreen is typically conducted manually, but other modes of semi-automated and fully-automated prescreening are also possible:

• Manual prescreens can be split between team members. The splits can be overlapping or non-overlapping, whereas an overlapping prescreen can be useful to analyze inter-coder reliability (also referred to as a parallel independent prescreen).

• Semi-automated prescreens are typically applied in large datasets and may involve machine-learning classifiers.

• Fully automated prescreens may be useful to apply scope restrictions (e.g., related to year of publication or publication outlets).

Operations

• colrev prescreen