Since its launch in 2017, the Clinical and Translational Science Institute’s Sample Size calculator website (sample-size.net) has gained traction not only at UCSF, but nationwide – it had 300,000 sessions in the past year with users from 1,200+ universities. It is the standard site used for UCSF courses, including Designing Clinical Research and the undergraduate medical education courses on Epidemiology, Biostatistics, and Population Science (EBPS). The website has recently gone through a redesign with added improvements based on user feedback as well as new calculators to help researchers.
“The site redesign should make it even more useful to clinical researchers. We have added many new calculators, a tool to help you find the right calculator, and a summary page that shows all the calculators on the site,” said Michael Kohn, MD, MPP, member of CTSI’s Study Design consultation unit (part of the Consultation Services program) and the Sample-Size website, and professor of Epidemiology and Biostatistics.
Kohn worked with site co-developer Josh Senyak on the site redesign. "Josh and I have added many new calculators, a tool to help you find the right calculator, and a summary page that shows all the calculators on the site,” Kohn said.
The standard calculators help investigators determine:
1) the precision with which they can estimate a parameter given their sample size,
2) the sample size that they need to estimate a parameter with a certain precision,
3) the minimum detectable effect size given group sizes, and
4) the required group sizes given the minimum detectable effect size.
The site also has additional calculators that will be useful to individuals studying and using diagnostic tests (e.g., for COVID-19) such as this one, which estimates the true prevalence of disease in the population:
https://sample-size.net/prevalence-calculator/
It runs the calculations in two different ways, according to two slightly different mathematical models. (The website team calls this the "seroprevalence" calculator as a short-hand, since it's immediately relevant to current COVID-19 seroprevalence studies.)
“I am proud of our seroprevalence calculator, which (for example) calculates the true sero-prevalence of antibodies to SARS-CoV-2 based on a study that uses an imperfect antibody test. This calculator doesn't just calculate the point estimate; it also correctly calculates the confidence interval, which has been a subject of controversy,” added Kohn.
This is a really useful site!
-Sample Size website User
The team has also added new tools that help researchers find the calculator that best suits their needs. This page shows all the calculators on the site, organized in a three-part taxonomy that illustrates relationships among the calculators:
https://sample-size.net/all-calculators-on-this-site/
A related tool is this interactive page where users answer questions about their research study, and the page provides a selection of calculators which may be appropriatefor the user's problem:
https://sample-size.net/calculator-finder/
Both tools were inspired by feedback the team received during a user testing process.