Business Statistics

Measures of Variation
 Measures of Variation
Why measure variation?
• To determine reliability of an average: When variation is small, the
average is reliable. It closely represents the individual values and is a
good estimate of the average in the universe.

• To serve as a basis for the control of the variability: quality assurance,

inequality of income and wealth, etc. Measurement of variation is
basic to the control of the cause of variation.

• To compare two or more data series with regard to their variability: A

low degree of variation would mean greater uniformity and
consistency and vice versa.

• To facilitate the use of other statistical measures: Testing of

hypothesis, analysis of fluctuations, techniques of production control,
cost control, quality control, all are based on measures of variation.
 Measures of Variation
Properties of a good measure of variation?
• It should be simple to understand

• It should be easy to compute

• It should be rigidly defined

• It should be based on each and every observation of the

distribution

• It should be amenable to further algebraic treatment

• It should have sampling stability

• It should not be unduly affected by extreme observations

 Measures of Variation
Methods of studying variation
• Range

• Interquartile Range or Quartile Deviation

• Average Deviation

• Standard Deviation

• Lorenz Curve
 Measures of Variation

• Absolute measures of deviation may be used to compare the

variations in two or more than two distributions provided the
variables are expressed in the same units and have almost
the same average value.

• If above is not the case, a relative measure of deviation is

used. It is the ratio of the absolute measure of deviation to
the average based on which the deviation is computed. This is
also called the coefficient of variation, meaning a pure
number that is independent of the unit of measurement.
 Measures of Variation
Range

Range is defined as the difference between the value of the

smallest observation and the largest observation of the
distribution.

Range = L – S

Coefficient of Range = L – S / L + S

It is a positional measure of variation

 Measures of Variation
Range
Merits:
• Simplest to understand
• Easiest to compute
• Quick but not accurate

Demerits:
• It is not based on each and every observation of the distribution.
• It is subject to fluctuations of considerable magnitude from sample to
sample.
• It cannot be computed in case of open-ended distributions.
• Range cannot tell us anything about the character of distribution between
the two extreme observations.
• Range is most unreliable as a guide to the variation of values within a
distribution.
 Measures of Variation
Range
Typical uses:

• Quality control

• Fluctuations in the share prices

• Weather forecasts
 Measures of Variation
Interquartile Range or Quartile Deviation

• The interquartile range includes the middle 50% of the

observations.

• One quartile of the observations at the lower end and

another quartile of the observations at the upper end are
excluded.

• It is the difference between the third quartile and the first

quartile:
Q3 – Q1
 Measures of Variation
Interquartile Range or Quartile Deviation

• Quartile Deviation (QD) = Q3 – Q1 / 2. It is a positional

measure of variation.

• The median +/- QD covers 50% of the observations in the case

of a symmetric distribution. It is approximately 50% in the
case of an asymmetrical distribution.

• Coefficient of QD = Q3 – Q1 / Q3 +Q1

• Coefficient of QD can be used to compare the deviations in

different distributions.
 Measures of Variation
Quartile Deviation
Merits:
• Useful to measure variation in open ended distributions.
• Useful to measure variation where data may be ranked but measured
quantitatively.
• Also useful in erratic or highly skewed distributions.
• QD is not affected by the presence of extreme values.

Demerits:
• QD ignores 50% of the items, the first 25% and the last 25%.
• Therefore, QD does not depend upon every observation.
• It is not capable of mathematical manipulation.
• Its value is affected by sampling fluctuations.
• It measures more of a distance on a scale around the average than the
variation around the average.
• It is more a measure of partition than a measure of variation.
 Measures of Variation
Average Deviation
• Average Deviation is obtained by calculating the absolute deviations of
each observation from the median (mean), and then averaging these
deviations by taking their arithmetic mean.

• Deviations from the median is more appropriate because the sum of

deviations of observations from the median is minimum when signs are
ignored.

• However, the absolute deviations from the mean is more popularly used
as the arithmetic mean is more popular as a measure of central tendency.
 Measures of Variation
Average Deviation
For Ungrouped data:

• AD (Med.) = / N

• Coefficient of AD (Med.) = AD / Median

For Grouped data:

• AD (Med.) = / N
 Measures of Variation
Average Deviation
Merits:
• It is relatively simple to understand and easy to compute.
• Average deviation appeals to general public who are not grounded in statistics.
• It is based on each and every observation of data.
• Average deviation is less affected by the values of extreme observation.
• Since deviations are taken from a central value, comparison about formation of
different distributions can be easily made.
• National Bureau of Economic Research uses AD in forecasting business cycles.

Demerits:
• Algebraic signs are ignored while taking deviations of the observations.
• AD is best when deviations are taken from the median, but median is not a
satisfactory measure of central tendency when degree of variability in a series is
high.
• It is not capable of further algebraic treatment.
• Rarely used in sociological and business studies.
 Measures of Variation
Standard Deviation
• Standard Deviation was introduced by Karl Pearson in 1893.

• Most important and widely used measure of studying variation.

• It is the square root of the mean of the square deviations from the arithmetic
mean. Therefore, called root mean square deviation.

• SD denoted by

• The square of SD 2 is called Variance.

• A small standard deviation mean a high degree of uniformity of the

observations as well as homogeneity of a series; and vice versa.

• A small standard deviation also means a more representative mean.

 Measures of Variation
Standard Deviation
For Ungrouped data:

• Deviations taken from actual mean

• Deviations taken from assumed mean

For Grouped data:

• Deviations taken from actual mean

• Deviations taken from assumed mean

Correct incorrect value of SD

 Measures of Variation
Standard Deviation
Mathematical Properties:

• It is possible to compute combined standard deviation of two or more groups or

data series.

• We have a separate formula for computing the standard deviation of the first n
natural numbers (positive integers). Page 144 of text book

• The sum of the squares of deviations from their arithmetic mean is always the
minimum. That is why the SD is always computed from the arithmetic mean.

• SD is independent of the change of origin but not scale.

• For a bell-shaped symmetrical distribution (normal distribution),

Mean +/- 1 covers 68.27% observations

Mean +/- 2 covers 95.45% observations
Mean +/- 3 covers 99.73% observations
 Measures of Variation
Relation between measures of variation

QD = 2/3

AD = 4/5

Mean +/- QD includes 50% of the observations

Mean +/- AD includes 57.51% of the observations

Mean +/- SD includes 68.27% of the observations

 Measures of Variation
Standard Deviation
Merits:
• It is based on every observation of the distribution.
• It is amenable to algebraic treatment. It is possible to calculate the combined SD
of two or more groups.
• It is less affected by fluctuations of sampling than most other measures of
variation.
• The square of SD, i.e. the variance is very useful to compare the variability of
two or ore distributions.
• SD is useful in many further statistical calculations.

Demerits:
• It is difficult to compute.
• It gives more weight to extreme values and less to those which are near the
mean (because of squaring the deviations).
 Measures of Variation
Coefficient of Variation

CV = SD / Mean
 Measures of Variation
Which Measure of Variation to use?
• Type of data available:

Avoid SD if observations are few in number or contain extreme values.

Avoid AD if data are generally skewed.
Avoid QD if data have gaps around quartiles.

Use QD in case of data with open ended classes.

• Purpose of investigation:

SD is most preferred unless otherwise stated.