Stat 20: some review for midterm 1 Michael Lugo September 27, 2010 The exam covers Chapters 1 through 5 and 8 through 11 of the text. Here I’ll briefly summarize the things you should know from each chapter. Chapter 1: Controlled Experiments • what is a controlled experiment? • some important words: placebo, control, treatment, confounder, doubleblind • don’t need to know the details of specific experiments Chapter 2: Observational Studies • what is an observational study? • what are some pitfalls of observational studies? (adherence, Simpson’s paradox, doctors preferentially seeing patients who are in trouble, and so on) Chapter 3: Histograms • how to draw a histogram. Make sure you can label the axes correctly! • what happens to the histogram when we transform the data? • how to determine the portion of a population which lies in some interval [a, n] (area under a curve) • different kinds of variables: discrete vs. continuous, qualitative vs. quantitative 1 Chapter 4: Averages and standard deviations • how to compute them. (and the RMS) • relationship between averages and histograms (“balancing” interpretation) • how do the average and the SD change with a change of scale? Chapter 5: The normal approximation • standardizing data • how to find areas under the (standard) normal curve • finding areas under the normal curve for a given mean and SD • percentiles and the normal curve Chapter 8: correlation • drawing and interpreting scatter diagrams • the elliptical “cartoon” of a scatter diagram. where are the SD, regression lines on such a cartoon? • interpretation of correlation coefficients: what does it mean if r is positive? negative? close to 1? close to 0? • how to compute correlation coefficients (there are two ways; either one is fine) Chapter 9: more about correlation • what happens to r when data undergoes a change of scale? • when is it appropriate to use r as a measure of association? (only in the linear case) • ecological correlation • association is not causation 2 Chapter 10: regression • how to make predictions using regression • regression line is the smoothed graph of averages • regression to the mean, and the regression fallacy • why are there two regression lines? Chapter 11: RMS error for regression • what is the RMS error of a prediction from regression? • using the normal curve in a vertical strip 3