Descriptive Statistics

Descriptive statistics employs a set of procedures that make it possible to meaningfully and
accurately summarize and describe samples of data. In order for one to make meaningful
statements about psychological events, the variable or variables involved must be organized,
measured, and then expressed as quantities. Such measurements are often expressed as measures
of central tendency and measures of variability.

Organization of data. Graphical representation of data is typically the first organizational step.
Frequency distributions, histograms, and/or frequency polygons are usually prepared in this
process.

 A frequency distribution, often the first organizational step, is an ordered arrangement

of all variables, which shows the number of occurrences in each category. Table shows a
frequency distribution concerning how much time students spent studying for an exam.
Note that the total number tallied (counted) in each category by the researcher equals the
number listed in the frequency column.

Such a frequency distribution can be presented graphically as a frequency histogram or

frequency polygon.

 Frequency histograms are bar graphs. Figure shows a frequency histogram derived from
the data in the frequency distribution in Table . The frequency (number of students)
determined from the tally is the ordinate (vertical, or Y, axis), and the number of hours
studied is the abscissa (horizontal, or X, axis). Each one‐hour interval is presented
 sequentially, and the height of each bar represents the number of students who studied
that number of hours.

 Frequency polygons are graphs in which the frequency of occurrence of the variable
measured is shown by using connected points rather than bars. Figure shows, in a
frequency polygon, the same data displayed in Figure . (Note that if the midpoints of each
of the bars in Figure were connected, the result would be this frequency polygon.)
Measures of central tendency. The three measures of central tendency, the mean, median,
and mode, describe a distribution of data and are an index of the average, or typical, value of a
distribution of scores.

 The mean, the arithmetic average of all scores under consideration, is computed by
dividing the sum of the scores by the number of scores. Based on the data in Table

1,
 The median is the point at which 50% of the observations fall below and 50% above or,
in other words, the middle number of a set of numbers arranged in ascending or
descending order. (If the list includes an even number of categories, the median is the
arithmetic average of the middle two numbers.) Based on the data in Table , the full list
of each student's study hours would be written 10, 9, 9, 9, 8, 8, 8, 8, and so on. If the list
were written out in full, it would be clear that the middle two numbers of the 40 entries
are 6 and 6, which average 6. So the median of the hours studied is 6.
 The mode is the number that appears most often. Based on the data in Table , the mode
of the number of hours studied is also 6 (8 students studied for 6 hours, so 6 appears 8
times in the list, more than any other number).
Graphical representations of the measures of central tendency may be presented in frequency
polygons that take the form of curves, which may be normal or skewed.

 Generally, if enough measures are taken of a variable and plotted as a frequency polygon,
the result is a normal curve (bell‐shaped curve), or normal distribution (Figure a). The
curve is symmetrical, and the mean, median, and mode fall at the highest point on the
curve.

 Skewed distributions are asymmetrical, with most of the scores grouped toward one
end. The mean, median, and mode fall at different points. Distributions may be skewed to
the left (negatively skewed) (Figure 3b) or to the right (positively skewed) (Figure 3c).
 The frequency distribution termed bimodal, has two peaks, which represent two equal
scores of highest frequency. In such a distribution, the mean and median may be at the
same point or different points.

Measures of variation. Variability refers to the extent that scores differ from one another and
from the mean. Widely used measures of variability are the range, variance, and standard
deviation.

 The range describes the spread of scores in a distribution. It is calculated by subtracting

the lowest from the highest score in the distribution. (In the example of hours of study,
the range is 10 − 1 = 9 hours.)
 The variance is a measure of variation from the mean of the squared deviation scores
about the means of a distribution. Using the data from Table as an example, the variance
for the entire distribution is computed by

 determining the mean of the distribution of data

  subtracting the mean from each score to determine the deviation score for that item
(Table , column 1)

TABLE 2 COMPUTATION OF DEVIATION SCORE, VARIANCE, STANDARD

DEVIATION: HOURS STUDIED FOR AN EXAM

 squaring each deviation score (to eliminate minus signs) and multiplying it by the
frequency of that score to account for the total number of scores (Table , column 2)
 summing the results of the previous multiplication step (Table , last entry in column 2) to
arrive at the total of all squared deviation scores and dividing by ( N − 1) ( N = number of
scores)
Some variance computations simply use N, but ( N− 1) is considered to produce a more precise
measurement. The variance gives one indication of how much the scores differ.

 The standard deviation (SD) is the square root of the variance.

TYPES OF DATA

The type of data collected determines the appropriate measurement scale, and the measurement
scale, in turn, determines the appropriate statistical procedure for analyzing particular data and
drawing conclusions from that data. Each type of measurement scale has a specific use.

Nominal scales. Nominal scales are composed of sets of categories in which objects are
classified. For example, a nominal scale dealing with household pets might include the
categories dogs, cats, birds, and fish. Data used in the construction of a nominal scale
are frequency data, the number of subjects in each category (in this case, the number of animals
for each type of pet

Ordinal scales. Ordinal scales indicate the order of the data according to some criterion. For
example, a researcher might ask people to rank their preference for types of household pets, with
1 as the most preferred and 4 as the least preferred (resulting in, perhaps, 1‐dogs, 2‐cats, 3‐birds,
4‐fish). Ordinal scales tell nothing about the distance between units of the scale (for example,
although dogs may be preferred to cats, no information is available about the extent of that
preference) and supply information only about order of preference.

Interval scales. Interval scales have equal distances between scale units and permit statements
to be made about those units as compared to other units (that is, one unit may be a certain
number of units higher or lower than another), but they do not allow conclusions that one unit is
a particular multiple of another because on interval scales there is no zero. That is, the scale does
not allow for the complete absence of the phenomenon being measured. For example, if you
refer to the interval scale used on a thermometer, you can say that 88 degrees is 2 degrees higher
than 86 degrees, but you cannot accurately say that 88 degrees is twice as hot as 44 degrees
because there is never a situation of no heat at all. (The zero on a thermometer doesn't indicate a
complete lack of heat, only one more unit on the scale, which continues downward.) Interval
scales, then, permit a statement of “more than” or “less than” but not of “how many times more.”

Ratio scales. Ratio scales have equal distances between scale units as well as an absolute zero.
If you're measuring the height of two trees and tree A is 36 inches tall and tree B is 72 inches tall,
you can accurately say that B is twice as tall as A. There is a condition of zero height. Most
measures encountered in daily living are based on a ratio scale.
Continuous and discontinuous scales. Measures may also be categorized according to
continuity and discontinuity. A continuous scale is one in which the variable under
consideration can assume an infinite number of values. A person's height, for example, might be
expressed in an infinite number of ways, ranging from feet, to inches, to tenths of inches, to
hundredths of inches, and so forth according to how small or large a measurement one wants to
make. On the other hand, discontinuous, or discrete, scales express the measurement of the
variable under consideration in a finite number of ways, as, for example, in a frequency
distribution such as the number of students in a psychology department or the number of players
on a team.