Descriptive Statistics - Graphical Displays
- What is the goal of Descriptive Statistics?
- advice for good statistical practice: Don't neglect descriptive methods!
- Populations vs. Samples
- define population
- define sample
- What is the term for a sample which comprises the entire population?
- define parameter
- define statistic
- EXAMPLE: dataset indicated on page 15...
- What is the population?
- What is the sample?
- Name two parameters which might be of interest.
- Name two statistics which might be computed.
- Types of Variables & Descriptive Goals
- What are the 3 types of variables and what distinguishes each?
- EXAMPLE:Suppose that the students surveyed for page 15 were also asked their major. Would that be a legitimate categorical variable? Why or why not?
- Descriptive Goals:
- one categorical variable [SEE Ford Explorers dataset]
- What number/percentage fall into each category?
- How are individuals distributed amongst the categories?
- two categorical variables [SEE Ford Explorers dataset]
- Is the prevalence of one category related to the prevalence of another?
- Are there special combinations of categories?
- one quantitative variable [SEE 2000 BayState Marathon dataset]
- What patterns are there?
- shape
- center
- spread
- Are there "outliers"?
- one categorical & one quantitative variable [SEE 2000 BayState Marathon dataset]
- Does the quantitative variable tend to differ across categories?
- If the categories are ordinal, is there a relationship to the quantitative variable?
- two quantitative variables [SEE 2000 BayState Marathon dataset]
- Are the two variables "correlated"? If so, positively or negatively?
- Are there individuals whose combination of variables is an "outlier"?
- Match each term in #1 below with the appropriate term in #2 below:
- explanatory variable response variable
- predictED predictOR
- Describing Categorical Variables
- numerical summaries: good ole percentages!
- For a single variable, the percentages are obvious.
- For two (or more), percentages can be computed...
- row-wise, as Utts shows
- column-wise
- table-wise
- EXAMPLE:
- Compute the column-wise percentages for Table 2.3 (page 22).
- Compute the table-wise percentages for Table 2.3 (page 22).
- EXAMPLE:
- Ford Explorers summary
- graphical displays:
- pie charts - limited utility
- bar graphs - Minitab, SPSS
- QUESTION: How will the look of a bar graph change if every individual were to "clone itself" (with the same category for its datum)?
- Intro to Describing Quantitative Variables
- 5-number summary
- What are the 5 numbers in the summary?
- What percentage of the data lie between...
- min & Q1?
- Q1 & median?
- median & Q3?
- Q3 & max?
- Q1 & Q3?
- How are the median & Q1 & Q3 calculated? Stay tuned...
- OUTLIERS
- How are they detected?
- How should they be handled? See text's discussion on pages 45-46; it's very wise.
- EXAMPLE: sample from 2000 BayState Marathon dataset
- Graphical Displays for Quantitative Variables
- stem-and-leaf plots: We'll leave these to your reading; they look very similar to histograms.
- dotplots:We'll leave these to your reading; they also look very similar to histograms.
- histograms:
- These are best suited to revealing shape. Options for "shape" are:
- symmetric, e.g. bell-shaped
- skewed, either to the right or left
- METHOD:
- by hand - see page 30
- with graphing calculator - explore on your own
- usingMinitab - [graphs/histogram]
- EXAMPLE: sample from 2000 BayState Marathon dataset
- boxplots: stay tuned ...
- COMPARISON: Read the advice on pages 35-36 and remember it for your project.