Discussion 1 – Week 2
Effect Size and Practical and Statistical Significance
Effect size refers to the strength of a relationship between two variables. It can be expressed through a difference or ratio. Odds ratios, which compare the likelihood of the same event occurring within two separate groups, reflect effect size.
Consider the following two scenarios:
- Ten thousand people participated in a survey in which they were asked whether they use a vitamin C supplement and also whether they had experienced cold symptoms the previous winter. The study demonstrated an odds ratio of 1.08 and a p-value of 0.04. Based on this, the researchers recommended the use of vitamin C supplements to avoid experiencing a cold next winter.
- Fifty people participated in a survey about stress and a diagnosis of high blood pressure. This study resulted in an odds ratio of 1.6 and a p-value of 0.07. Based on these results, the study was not published and the researchers moved on to other work.
These two hypothetical situations are extremes, but they illustrate the difference between practical and statistical significance. Practical significance is inferred from the size of the effect, while statistical significance is inferred from the precision of the estimate. The effect size reflected in the odds ratio in the first case was 0.08, or an 8% difference in likelihood of a cold based on vitamin C supplementation (small practical difference of 8%, but statistically significant, with a p-value of 0.04). The effect size in the second case was 0.6, or a 60% difference in the likelihood of a diagnosis of high blood pressure based on stress (large practical difference of 60%, but small statistical significance, with a p-value of 0.07).
For this Discussion, review this week’s Learning Resources. Consider effect size and how to determine whether practical or statistical significance exists. Select a public health research study from one of the peer-reviewed journal articles listed in the Week 2 Resources. Consider whether effect size is practically or statistically significant in the study.