Descriptive Statistics and Distributions Flashcards
biomedical data can be ___, ____, ____ or ____
biomedical data can be nominal, ordinal, interval or ratio
NOIR
if the measure is constrained to be positive and the mean is _____, the measure must be right skewed
if the measure is constrained to be positive and the mean is less than ~2x the standard deviation, the measure must be right skewed
describe application of the mean (average) and a con for using it
- most common measure of central tendency
- easy to model statistically; can be further averaged
- only for interval and ratio measures
- susceptible to extreme values
describe application of the median (central)
- less affected by extreme values than the mean is
- harder to model statistically: cannot be averaged
- for ordinal, interval and ratio measures
-
commonly used for skewed measurements:
- biomarker measurements
- income
- not susceptible to a few extreme values
describe application of the mode (most common)
- not at all affected by extreme values
- for all types of data: most often used for nominal data
- can be misleading if there are many categories, or nonsensical for unrounded continuous variables
- very difficult to model statistically
describe calculating the weighted mean
pooling is only for the ____
pooling is only for the mean
explain the different types of variability
- range: “total variability”
- indicates extreme values
- interquartile range, IQR: “typical variability”
- half of values fall within this range
- not affected by a few extreme values
- variance: “typical squared variability”
- easy to work with statistically
- more affected by extreme values than IQR is
standard deviation: square root of variance
describe the circumstances you would use the measures of variability
- use the range for any ordinal data
- use the interquartile range in the same settings you would use the median
- use the variance and SD in the same settings you would use the mean
describe how to calculate range
describe how to calculate the interquartile range