effect size and significance

[coco34]

Does anyone knows about the relationship between the effect size and the significance? I was given following question and no clue how to answer it:

You hear that a study (randomized, controlled trial with two groups) found a small sized effect of a drug (n = 72) against placebo (n = 72) of d = 0.2. Is this effect significant with α=0.05? (You can solve this problem just by looking at the formulae. You should not need the help of a computer.)

 

[css]

Hi coco34, If you want to receive help, you should probably provide more information about your question (i.e. which formulas were provided to you). In any case, regarding your first question: if the 95% confidence interval between two means does not include the value 0, the difference is statistically significant at 0.05. To answer the second question, one needs to know which level of type II error is being accepted (often, 0.8)

 

[noetsi]

If you mean statistical significance then what CSS said is true, for linear regression for logistic regression this would be one not zero. But many people mean substantive significance when they talk about effect size (literally does this matter) and substantive and statistical significance are totally different. You can be statistically significant and still have a very small effect size if you have enough statistical power.