What is an effect size?

An effect is the result of something. It is an outcome, a result, a reaction, a change in Y brought about by a change in X.

What is an effect size? An effect size refers to the magnitude of the result as it occurs, or would be found, in nature, or in a population. Although effects can be observed in the artificial setting of a laboratory or a sample, effect sizes exist in the real world.

Effect sizes are ubiquitous. You can find them in newspapers, college brochures, shop windows, Facebook ads, product packaging, church newsletters, blogs, tweets, TV commercials—just about anywhere. Here are some everyday examples of effect size:

  • lose 20 pounds in four weeks on the South Beach diet
  • learn how to speak Swahili in six months
  • make $2,300 a day working from home
  • improve test performance through meditation
  • fast-track your career with an MBA
  • list your property with us and sell your home within a week
  • read this book and improve your publication prospects

When researchers estimate effect sizes by observing representative samples, they generate an effect size estimate. This estimate is usually expressed in the form of an effect size index.

For a good introduction to effect sizes – how to report them, how to interpret them – check out my book Effect Size Matters


8 thoughts on “What is an effect size?

  1. Donna R September 23, 2010 / 2:34 am

    What does a negative effect size tell me?

    • Paul Ellis September 23, 2010 / 8:00 am

      The sign reveals the direction of the effect. A negative r indicates a negative correlation; a negative d indicates the effect is bigger for the second group.

  2. Anoop March 20, 2012 / 1:06 pm

    just ordered the book.

    I am not sure if you are checking this site.But can we use effect size to measure different measures? For example, we measure muscle growth via lean body mass, CSA, tape measure. So can we use effect size to combine these measurements and show the effect size?

    Thanks you

    • Paul Ellis March 25, 2012 / 1:46 pm

      I think you are confusing the measurement of effects with the measurement of variables. Do you have a particular effect in mind?

      • Anoop March 25, 2012 / 2:15 pm

        Hi Paul, I was reading a meta-analysis about hypertrophy and it used effect size. And the author wrote this:

        “Another problem with determining the effects of set volume on hypertrophy is the many ways in which hypertrophy can be measured. Studies have used wholebody lean mass (11,26), regional lean mass (27,35), muscle thickness (31,40), muscle cross-sectional area (31,35), or muscle circumference (30–32) to measure hypertrophy. Different regions of a particular muscle may also be measured (40). Thus, comparisons across studies can be difficult. The calculation of a standardized effect size (ES) can aid in the comparison across studies (3).”

        From what I understand, he means effect size can be used to measure different measures of muscle growth and be compared. Is this true? If you are releasing a second edition , I have some suggestions.

      • Paul Ellis March 25, 2012 / 3:08 pm

        Now I understand. This is the apples and and oranges problem I discuss on p.98 of the book. The author is interested in the effect of X (set volume – whatever that is) on Y (hypertrophy) but since the measurement of Y varies across studies, he is not sure that direct comparisons are possible. The solution he proposes (calculating standardized ES) does not deal with this particular problem but the problem of comparing results effects (not variables) that have been reported in different metrics.

  3. Anoop March 26, 2012 / 5:53 am

    Hi Paul, I checked page 98, but it didn’t say much. So are you saying we can use effect size as the author has used or you can’t? The measurements are in different units. I thought effect sizes measurements need to be for the same units.

    • Paul Ellis March 26, 2012 / 9:52 am

      Pages 99-101 discuss the reasons and methods for converting study-specific ES estimates into a common metric. Page 98 lists several tactics for making sure you don’t have an apples and oranges problem in the first place. If you need further information, any good meta-analysis textbook will have chapters on these issues. Some are listed in the Bibliography.

Comments are closed.