Dear All,
This is an age-old question on confidence intervals. Suppose we have
two mean test scores, and desire to know if there is a statistically
significant difference between the two means. So we obtain our
confidence intervals, say, (20, 30) and (35, 45). We therefore
conclude there is a significant difference. But what if our confidence
intervals were (20, 30) and (30, 40)? What should we conclude?
Further, what if we had (20, 30) and (29, 39) or even (20, 30) and
(28, 38)? Is there still a difference? Any comments?
Raphael
|