You are on page 1of 19

The Statistical Imagination

Chapter 4. Measuring
Averages

2008 McGraw-Hill

Central Tendency Statistic


A statistic that provides an
estimate of the typical, usual,
or normal score found in a
distribution of raw scores
Provides a sense of average
2008 McGraw-Hill

The Three Measures of


Central Tendency
The three central tendency statistics are
the mean, the median, and the mode
Each has strengths and weaknesses that
make it useful with certain score
distributions
The relative values of the three central
tendency statistics inform us of the shape
of a score distribution
2008 McGraw-Hill

Central Tendency Statistics


and Levels of Measurement
The mean and median are
appropriate with interval/ratio
variables
The mode is appropriate with
variables of all levels of measurement
With nominal variables, means and
medians are meaningless
2008 McGraw-Hill

The Mean
The mean is the sum of all scores in
a distribution divided by the number
of scores observed; the arithmetic
average
Most useful central tendency statistic
because it allows for many
mathematical operations
Applies to interval/ratio variables
2008 McGraw-Hill

Calculating the Mean


Sum the individual scores of the
interval/ratio variable X, then divide
this sum by the number of
observations (i.e., the sample size, n)
Weaknesses of the mean: Its
calculation is affected by outliers and
skews
2008 McGraw-Hill

Combined Mean of Two Groups


To get the combined mean of two
groups of different sizes (n), multiply n
times the group mean for each group
and sum to obtain the total sum of X;
divide this sum by the total n
Do NOT simply average the two group
means
2008 McGraw-Hill

The Median
The median is the middle score in a ranked
distribution; the score for which half of the
cases fall above and half fall below
It is equal to the 50th percentile
It is a location score
Useful with interval/ratio variables
Best central tendency statistic to report
when a distribution of scores is skewed
2008 McGraw-Hill

Calculating the Median


Rank the scores from smallest to largest
Divide the sample size by 2 to get near the
middle score in the ranked distribution
If n is odd, the median will be an actual
case in the sample; however, if n is even,
the median is located between two middle
scores and is calculated by taking the
mean of those two scores
2008 McGraw-Hill

Weaknesses of the Median


The median is insensitive to the values of
the scores in a distribution. Thus, two
distinctly different score distributions may
have the same median
The median is sensitive to a change in
sample size. If new cases are added, the
median may drastically change

2008 McGraw-Hill

The Mode
The mode is the most frequently occurring
or most popular score in a distribution
It is useful with variables of all levels of
measurement
The mode is a score, X , not a frequency, f
Do not confuse the mode with the majority
of scores
The mode is easy to spot in charts
2008 McGraw-Hill

Calculating the Mode


Compile scores into a frequency
distribution
Identify the value of X with the most
cases
On a histogram, it is the score of X for
the highest column; on a polygon, the
highest peak; on a pie chart, the largest
slice
2008 McGraw-Hill

Weaknesses of the Mode


The mode is the least useful measure
of the three because of its narrow
informational scope
The mode is insensitive to the values
of scores in a distribution
The mode is also insensitive to
sample size
2008 McGraw-Hill

Frequency Distribution Curves


A substitute for a frequency histogram or
polygon in which we replace these graphs
with a smooth curve
The area under the curve represents the
total number of subjects in a population, and
is equal to a proportion of 1.00 or a
percentage of 100 percent
The relative locations of the mean, median,
and mode on the X-axis are predictable for
certain shapes of distribution curves
2008 McGraw-Hill

Three Common Shapes of


Frequency Distributions
1. Normal

distribution or normal

curve
2. Negatively skewed distribution
3. Positively skewed distribution

2008 McGraw-Hill

Features of the Normal Curve


The mean, median, and mode of the
variable are equal and centered in the
curve
The curve is symmetrical and bell-shaped
When a distribution is not skewed or
otherwise oddly shaped, the mean is the
central tendency statistic of choice
2008 McGraw-Hill

Features of a Negatively
Skewed Distribution
Has extreme scores in the low or
negative end
The mean will have the lowest
value of X, the mode the highest,
and the median will fall between

2008 McGraw-Hill

Features of a Positively
Skewed Distribution
Has extreme scores in the high or positive end
The mean will have the highest value of X, the
mode the lowest, and the median will fall
between
When a distribution is skewed, the median is
the statistic of choice. It minimizes error in
describing a skewed distribution because it
falls between the mean and the mode
2008 McGraw-Hill

Statistical Follies
Mixing subgroups of subjects can result
in a distorted mean
For example, the average age of firstgraders (around 6 years old) and their
mothers (averaging around 31 years
old) is 18.5 years, an age no one in
either group approximates
2008 McGraw-Hill

You might also like