Minimum number of "good" items in validation set:
Hint: a broad initial search will make it easier to test multiple approaches.
1) Start the process
Search using any PubMed search terms you like — from a couple of keywords to the most complex search you can manage.
2) Sort your Cites
Examine a random sample of the citations returned by your search and sort them into "Good" and "Bad" piles. Good items will make up your validation set!
3) Test your Searches
PubMed Search tester now uses that validation set to automatically test successive iterations of your search strategy until you are satisfied with the balance of sensitivity and precision.
Not sure if you picked correctly? Choose an item below to see details and change its sort status.
"Good" documents: 0
"Bad" documents: 0
Skipped documents: 0
Refine your search
| Initial Search | Total found | Number screened | Good cites (Ratio) | Bad cites (Ratio) |
|---|
| Test search | Total found | Good cites found | Good cites missed | Sensitivity | Bad cites found | Precision |
|---|
See the about page for details about how sensitivity and precision are calculated.