Following our refinements to our database of data sources and the lessons of last week, we dove further into the pool of data-synthesis articles we identified previously from NCEAS and Web of Science.
Data extraction is (probably) the part of a systematic review that takes the most effort. It is also the part that provides you with data to analyze and synthesize! The typical workflow for data extraction involves downloading the original article if you haven’t done so already and then moving through your data sheet column by column and entering any relevant information from the article. Some columns in the datasheet can be very quick to fill in, for example, journal name and publication year. Other columns (i.e. pieces of data) can be very challenging to locate in an article or non-existent. One example of a challenging piece of data to extract is the location of each article’s data output. Sometimes researchers store their data in supplemental material sections or online repositories. We did our best to track down where data output gets stored, but we also found myriad broken links and other data dead ends.
This week, we continued our work on the citation best practices document. Each time one of us noticed a manuscript that provided particularly clear citations for their data sources we added it to our growing best practices document. We’re looking forward to compiling this list and sharing it soon. We are also working on an illustration that highlights the steps required to complete a systematic review. Previous guidelines for systematic reviews (Koricheva et al. 2013; Moher et al. 2009) leave out some of the critical but more detailed steps (e.g. choosing reference manager software) and we provide those in our illustration.