Articles - Journaltalk

Power Analysis is Essential: High-Powered Tests Suggest Minimal to No Effect of Rounded Shapes on Click-Through Rates

About this article

Author
  • Ron Kohavi
  • Jakub Linowski
  • Lukas Vermeer
  • Fabrice Boisseranc
  • Joachim Furuseth
  • Andrew Gelman
  • Guido Imbens
  • and Ravikiran Rajagopal
Keywords A/B test,online controlled experiment,winner’s curse,replication crisis,reproducibility crisis,Sample Ratio Mismatch (SRM),call-to-action (CTA) buttons,virtual element shapes.
Volume Number 23
Issue Number 1
Pages 139–172
File URL Power Analysis is Essential: High-Powered Tests Suggest Minimal to No Effect of Rounded Shapes on Click-Through Rates
File Format
Access
Publication year 2026

Flag this article

Join this discussion.

1 comment

  1. A fourth replication of the pattern was reported by us recently at https://bit.ly/roundedCornersPatternMetroPost, adding evidence to this replication effort.

    With a sample of over 7.4 million users (vs. under 1,000 in the original study), the estimated lift was -0.07% (p-value 0.83), so consistent with the null hypothesis of no effect.

    by Ron Kohavi

Log in to Journaltalk to discuss this article!

Don't have a Journaltools account? Sign up now.

Required

Log in to Your Account

Member login

Jt Article Discussions

Most recent article-specific discussions at Journaltalk.

03 Apr

The Ideological Profile of France’s Economic Bestsellers
Cry Wine and Sell Vinegar: Rejoinder to Miao and Wang
Specification Searching in the Race between Education and Technology
Green Vanities in Europe
For Housing Supply Deregulation Over Against Three Recent Papers
Globalization and the China Shock: A Reassessment of Autor, Dorn, and Hanson
Journaltalk: Opening the journals to civil voices everywhere!

All contents © 2026 by Daniel Klein unless otherwise attributed. All rights reserved.