LBOLYTC CHAPTER 3: NUMERICAL
DESCRIPTIVE MEASURES
DR. WILSON CORDOVA
MEASURES OF CENTRAL TENDENCY
CONSIDERATIONS FOR CHOOSING A
MEASURES OF CENTRAL TENDENCY
MEASURE OF CENTRAL TENDENCY
describe the “center” of a given data
set. It is a single value about which the
observation tends to cluster. For a NOMINAL VARIABLE, the mode is the only
measure that can be used.
ARITHMETIC MEAN (or simply Mean) is the sum For ORDINAL VARIABLES, the mode and the
of all total observations, denoted by x-bar: median may be used. The median provides more
information (taking into account the ranking of
categories.)
For INTERVAL-RATIO VARIABLES, the mode,
median, and mean may all be calculated. The mean
provides the most information about the distribution,
but the median is preferred if the distribution is
skewed.
MEASURES OF POSITION
PROPERTIES:
It always exists for quantitative variables. MEASURES OF POSITION are measures that
It is unique. discriminate a group of scores from another group in
It takes into account every item of the data. the same data set.
Thus, it is easily affected by extreme
values. QUANTILES divides data into an equal number of
parts.
MEDIAN is the middle value of an array, denoted by
Md. QUARTILES are values that divide a set of
observations into four equal parts. Denoted by Qi, i =
1, ..., 4.
PROPERTIES:
Not easily affected by extreme values.
DECILES are values that divide a set of observations
It always exists and is unique.
into 10 equal parts, denoted by Di, i = 1, 2, ..., 10.
MODE is the observation(s) that occur most
PERCENTILES are values that divide a set of
frequently in the data set, denoted by Mo.
observations into 100 equal parts, denoted by Pi, i =
PROPERTIES: 1, 2, ..., 100
No calculations are required (for the
ungrouped mode).
MEASURES OF VARIABILITY
It may not exist.
It may not be unique.
1 (LBOLYTC)
LBOLYTC CHAPTER 3: NUMERICAL
DESCRIPTIVE MEASURES
DR. WILSON CORDOVA
MEASURES OF VARIABILITY describe the
extent to which the data are dispersed. Since squared units of measure are often
awkward to deal with, the square root of
VARIABILITY is descriptive statistics that describe variance is often used instead.
how similar a set of scores are to each other.
The standard deviation is the square root of
The more similar the scores are to each variance
other, the lower the measure of dispersion
will be The less similar the scores are to COEFFICIENT OF VARIATION (CV) is the ratio
each other. of the standard deviation to its mean expressed in
percent.
The higher the measure of dispersion will
be In general, the more spread out a Compare variability of two populations that
distribution is, the larger the measure of are expressed in different units of
dispersion will be measurement
RANGE is the difference between the highest and
lowest value in the data set. Expressed as a percentage rather than in
terms of the units of the particular data
The range is rarely used in scientific work
as it is fairly insensitive.
It depends on only two scores in the set of SKEWNESS
data, HV and LV.
SKEW is a measure of symmetry in the distribution
Two very different sets of data can have the of scores
same range:
MEASURE OF SKEWNESS is a frequency curve
VARIANCE (s2 or σ2) is the mean squared that is not symmetrical about the mean is said to be
differences of the observations from their mean. skewed. If it tails off to the right, we describe it as
positively skewed, but if it tails off to the left, we say
This difference is called a DEVIATE or a it is negatively skewed. The relationship between the
deviation score. mean and the median is related to the direction
skewness.
The DEVIATE tells us how far a given
score is from the typical, or average, If SK < 0, then the distribution has a
score. NEGATIVE SKEW
Thus, the deviate is a measure of dispersion If SK > 0 then the distribution has a
for a given score. POSITIVE SKEW
If SK = 0 then the distribution is
STANDARD DEVIATION (s or σ) is the positive SYMMETRICAL
square root of the variance.
2 (LBOLYTC)
LBOLYTC CHAPTER 3: NUMERICAL
DESCRIPTIVE MEASURES
DR. WILSON CORDOVA
MEASURE OF KURTOSIS
KURTOSIS measures whether the scores are spread
out more or less than they would be in a normal
(Gaussian) distribution.
A distribution is said to be:
MESOKURTIC if K=3,
LEPTOKURTIC if K>3, and;
PLATYKURTIC if K< 3.
3 (LBOLYTC)