Lesson #34 Review and Summary
Important Topics - Descriptive statistics - Basic probability, independence - Sensitivity and specificity - Relative risk, odds ratio - Binomial distribution - Normal distribution, standardizing - Reading tables (Z, t, F, 2 )
- Inference (C.I. and hypothesis testing) - one mean(one-sample t-test) - two paired means(paired t-test) - two ind. mean(pooled t-test) - several means(ANOVA) - one proportion (Z-test for one proportion) - two proportions - cat. var. association(contingency table) (Z-test for two proportions) - num. var. association(correlation/regression)
A colleague had data on body fat for runners and swimmers, and performed a pooled t-test to compare the two types of athletes. I made similar measurements for runners and swimmers, but I also included volleyball players. Now that I have three groups, how do I go about testing for a difference?
Our team has developed a new medical technique that we think will lead to better survival rates than a standard technique. We have two groups of patients; one group which received the standard technique, and another which received our new technique. How can we obtain statistical evidence that our technique is better?
We have collected data on 20 women who are taking a certain drug, and want to see if the average age of menopause of such women is greater than 55 years old. How does one make such a statement?
At work, we wanted to look at the effectiveness of a clean-up effort at a nearby lake. A pollution index was measured at a number of sites around the lake before the clean-up effort, and again six months after. The data was analyzed using an ANOVA, where the two time periods were the two treatments. Is this OK, or is there a better method?
I work as a volunteer for the Red Cross, and one of my supervisors wants me to look for a relationship between blood type and whether or not the person has ever donated blood. I have a random sample of 150 people that I took from a large university class, where I asked people these two questions. What kind of analysis should I use?
A physician in my office wishes to quantify the relationship between the amount of moderate to strenuous exercise and systolic blood pressure. The exercise variable is converted to METS, which is a quantitative variable. I have suggested she use a correlational analysis. Do you have any further suggestions?
I recently saw a study which said that 30% of teenagers in a rural area eat a nutritious diet. I expect that this percentage is even lower in the urban area where I live. How can I go about finding out if this is true? I have asked a bunch of teenagers at a local school about their diet.
My wife works for the Department of Mental Health. She says they want to see if there is a difference in the average levels of a certain blood chemical between people with depression and those not suffering from depression. How does one go about performing such an analysis?