1. Design of studies

Controlled experiments

• Study units, population, sample, variable, response, factors
• treatment, treatment group, control group, placebo.
• blind and double-blind studies, randomized study, stratified experiment

Observational Studies

• Controlling for variables, confounding factors, stratified study
• Distinguishing between Controlled Experiments and Observational Studies

Sample Surveys

• Sampling unit, population, parameter, sample, statistic
• Sources of bias: selection bias, nonresponse bias, response bias,
• A large sample does not protect against selection bias
• Sampling procedures --- simple random sample, stratified sample, quota sample
  --- describe, identify, strengths and weaknesses

2. Displaying data

• qualitative vs. quantitative, discrete vs. continuous
• Categorical variables: nominal, ordinal; ordered categorical, interval scale

• Understanding histograms: the shape tells us the relative counts/percent in different intervals
  --- Area respresents the percent in specified intervals
  --- Scale on the vertical axis are percent per horizontal-axis unit.
• Forms of histogram display: bars of equal/different widths, back-to-back, stacked
  Pie charts, Bar charts, Stem-and-leaf diagrams

3. Summary statistics, and the Normal curve

• Measures of center: Median, Average, Mean, Percentiles.
• Measures of spread: Standard Deviation (SD), Range, inter-quartile range,
• How to define them, find their values, ``guesstimate'' them from histograms.

The normal curve

• Standard units (SD units), z-scores
• Normal approximation for data
• Normal curve arithmetic: Finding z-scores, percentiles, percentages

4. Correlation and Association

• read, interpret:
• describe basic shapes, identify outliers, non-linearity
• Identify independent and dependent variables in a problem and on scatter diagrams

• Associations may be positive, negative, weak, strong
• Correlation coefficient, r : a number between -1 and +1
• Interpretation of r -- measures the degree of clustering of the scatter diagram about the SD line
• Sensitivity of the correlation coefficient to outliers and to non-linearity

• for data on 2 variables: 2 means, 2 SD's, and r
• Association is not cause! -- confounding factors, time trends, etc.