Statistics Question?? One tailed or Two tailed test?

Question

Scores on standardized intelligence tests are normally distributed with m = 100 and s = 15. You think that experience is especially important in determining intelligence and that an enriched environment can increase a child’s IQ test scores. You randomly select a sample of 30 children to participate in an enrichment program. After their experience in the enrichment program, the IQ scores for the sample is M = 110.



a. Is there evidence that your enrichment program increased IQ test scores?



b. In answering part a., did you use a one-tailed test or a two-tailed test? Justify your choice of a one-tailed or two-tailed test. How did this choice influence your conclusion?



c. Comment on the effect size obtained with your enrichment program.

- I used a Two tailed test to calculate (a) and Cohen's d for part (c) Now I am wondering if I should have used a One tail test???

Answer 1

For testing a hypothesis in one direction, a one-tailed test should be used - to test a general correlation (where you do not know which way it will go) you use a two-tailed. Do you have a data set for the children's test scores before the enrichment program, or just the mean of 100 and standard dev. of 15? You really need both sets of scores to do a correlation, or any experiment for that matter, to be able to say your results are the result of the program. You could try using an Anova using just your average m and s and the m and s from your kids' results, or test a new set of 30 kids w/o the program and run a Chi-square with two different populations. It might help to have a larger sample size anyway; it's hard to get measurable results with such a small sample.