Stats Terms: Graphs and Basic Variable Terms

What type of data are pie and bar charts used for?

Categorical

General rule about pie vs bar charts?

Bar charts are better.

Weaknesses of pie charts

• Can be overkill, sometimes best just to write out data.

• Pie charts indicate you are seeing 100% of something, but it is hard to eyeball differences in percents.

• Need many tiny slices or a large other category for situations with many categories.

• Color coded pie charts/charts with keys can also be difficult to read.

Strength of pie charts

Can be good in specific situations like some voting polls where the differences between a few categories are clearly shown and percents are written out.

Strengths of bar charts (especially over pie charts)

Generally better for comparisons than pie charts, and can show changes over time.

Bar charts vs histograms.

Bar charts generally have white space between bars, histograms usually not.

Bar charts are for ___ data

Categorical

Histograms are for ___ data

quantitative

What is a right skew and what does it indicate?

A graph with a long right tail, peak of data is to the left. Indicates a number of high performing outliers.

What is a left skew and what does it indicate?

A graph with a long left tail, peak of data is to the right. Usually indicates a max value that most subjects can hit, with a number of under-performers included.

Peak of a graph

Region of maximum number of scores; indicates where the standard lies.

Range

Distance between smallest and largest number in the dataset.

How is range calculated?

Subtract smallest score from largest

What is a histogram class?

AKA bins. Can have different sizes that influence how the graph can be read (how easily shape of distribution can be determined). Classes can be 10, 50, 500, 1000, etc in size.

Symmetrical graphs

Graphs that look roughly the same to both the left and right (assuming a line drawn down the middle of the graph)

What are the extremes of a graph called?

Regardless as to whether they are symmetrical or skewed, these are called tails

Modality

Number of peaks a distribution has. One can have unimodal, bimodal, trimodal graphs.

Outliers

Values so far from middle of the data that they may not be representative of the dataset and should be considered for exclusion.

Case

Objects or subjects described by the data. So can be human or animal subjects, companies, objects, etc.

Label

Unique identifiers (name or number) given to each subject/case.

Variables

Attributes being studied. Can be categorical or quantitative

Categorical variable

A group such as sex, school, graduating class, etc. Is not something that could be averaged or multiplied

Quantitative variable

An attribute such as height, cost, grams of food intake that could be averaged and can have a unit of measurement.

Different cases can have different ___ of a variable

values

What can't be determined by a histogram?

Data points like max, min, range. You need to look at dataset for those values. Histograms give shape of distribution.