Summary Measures in Statisitics
- 1.Three types of variables are commonly used in clinical research studies:
Continuous – can take on an infinite number of possible values
Binary – can take on only two possible values
Categorical – can take on only a few possible values
A histogram plots the observed values of a variable on the X-axis versus the relative frequency of these values on the Y-axis.
The arithmetic mean describes the middle of the data for symmetric distributions, including data from normal-appearing distributions.
The median refers to the value within a distribution for which exactly half of the data fall above this value and half fall below it.
Disagreement between mean and median values suggests that a distribution is not normally distributed.
A continuous variable may be divided into 3, 4, or 5 equally sized groups called tertiles, quartiles, and quintiles, respectively.
The interquartile range is defined as the 25th and 75th quantiles of a distribution.
Some techniques to describe the joint distribution between two variables include tabulation across categories, scatter plots, correlation, and quantile-continuous plots.
- 9.Correlation coefficients are interpreted as:
+1 indicates perfect positive agreement between two variables
0 indicates no agreement between two variables
–1 indicates perfect negative agreement between two variables