Tip #58: Exporting a Random Set of Citations from Ovid Databases to Avoid Bias

By

For evidence synthesis projects, it's critical for research teams to be in accord on eligibility criteria. I encourage my teams to perform a pilot screening of a subset of retrieved citations to identify where they may not agree before they undertake the formal screening process. This requires importing a sample of records to Covidence, our institutional evidence synthesis management tool. 

The process I describe exports a random sample of records from an Ovid database to EndNote in order to convert the exported file to a compatible filetype (XML) for import to Covidence. Regardless of the reference and synthesis management tools you use, the steps I describe in Ovid remain the same.

For pilot screening in Covidence, if I were to simply export the first 50 citations retrieved by a search, I may introduce selection bias. According to Wolters Kluwer, owner of the Ovid platform, the last item added or updated in the database is the first displayed in the results when using Advanced Search Mode. Older records will display earlier if they've recently been added to the database. While Ovid allows me to re-sort results by author, title, or other fields, risk of bias remains high because the selection of citations is not random. 

I use a random number generator to select records for piloting to avoid this risk. I like Research Randomizer, although there are others. Here's how I use it.

Let's say my Embase search returned 1697 records. I typically populate a pilot review with up to 5% of the search results. In this case 85 records.


Pro tip: viewing randomly generated numbers with Place Markers Off  (last box) returns a block of numbers separated by commas, which is perfect for inputting to Ovid. 

1. Copy the resulting set of numbers and paste into the Range box at the top of your results. 



2. Click Export to create your preferred format (file type) and fields with a set of  randomly selected citations.


3. Import the file to its final destination!

While my demonstration was specific to exporting to EndNote and screening in Covidence, Ovid allows exportation of several formats, so you could certainly export a randomly selected set of records to other reference managers, screening tools, or even a spreadsheet. 

I hope you think of other reasons to use a random number generator. If you do, let us know!

That's all I've got.

Comments

  1. Andy HicknerApril 26, 2025 at 11:11 AM

    I've worked on a couple of search filter projects recently where it would have been very helpful to have a random sample like this! Thanks for this tip, Michele.

    ReplyDelete
  2. Michele McGinnisMay 5, 2025 at 11:39 AM

    Oh, indeed. What a great example of use for random sample export Andy. Thanks for sharing.

    ReplyDelete

Post a Comment

Popular posts from this blog

Favorite Features & Sneaky Solutions: A Database Tips Lightning Round: View the recording!

By
On October 7, 2024, the Medical Library Association's User Experience (UX) Caucus held a database tips lightning round. The event recording, chat transcript, presenter slide decks, and a summary of Q&A and tips culled from the chat can be found in the files section of the  UX Caucus' OSF site ! The presenters and their topics, with timestamps of when they appear in the video*, were: Methodology and Publication Type Filters in EBSCOhost Databases (00:40) Marilia Antunez, MA (LIS), MA,  Life & Allied Health Sciences Librarian, Associate Professor, The University of Akron (OH) Tricky Truncation in ProQuest Databases (07:15) Jocelyn Boice, Associate Professor / Collection Strategies and Data Analysis Librarian, Colorado State University Libraries Embase Query Translator (12:52) Molly Higgins, Systematic Review Librarian, USDA NESR What’s Included in Emtree’s /syn and /br (19:20) Amy Nadell, MSLIS, MA , AHIP-D, Librarian, Banner Health Fun with PMIDs (24:55) Jenni...
Read more

Tip #1: Bulk export from Google Scholar

By
Image
This post was updated on October 10, 2022 with new instructions for #3 under "For EndNote users..." For years, I kept hearing from colleagues that I should be searching Google Scholar as part of my systematic review projects. I didn't, for 3 reasons: I didn't know GS' search syntax GS notoriously does not allow bulk export to reference managers I didn't have a sense of how many results I should export.   Last fall, I finally said enough was enough and it was time to learn how to wrangle this beast. It turns out that there is now a third-party tool that facilitates bulk export from Google Scholar, Harzing's Publish or Perish . Here's how you get started: Download PoP for free Paste your query into the Keywords field Set your maximum number of results. I follow the guidelines on p. 135 of Wichor Bramer's dissertation , which is a great handbook for expert searching. Bramer suggests the first 200 results, or 100 if you are getting low numbers of res...
Read more

Tip #23: PubMed's [tiab] vs. [tw]

By
Image
Search field tags are a quick way to specify the exact fields you would like to search when using keywords. For example, I can tell PubMed to search for my term only in the title field or only in the MeSH Terms field. You can view a list of available search fields from the drop down menu in PubMed's Advanced Search Builder. Specifying field tags for keywords is also a good practice in PubMed because keywords searched without field tags will be translated through Automatic Term Mapping which may reduce the precision of your results. Here are a few sample searches in my search history, including #4 which wasn't limited to a field (notice how it was translated in the expanded search details): The fields that are included in the [title/abstract] fields are pretty self-explanatory (words and numbers included in a citation's title, collection title, abstract, other abstract and author keywords), but what is actually included in the [Text Word] (can also be searched as [tw]) fi...
Read more