PSY153 Statistics in

Behavioral Sciences I
Chi-square: Differences between samples of frequency data / Probability

Burak Emre Gürsoy, Ph.D.

Chi-square, Differences between
samples of frequency data
• Chi-square is used with nominal (category) data. A minimum of two
categories is involved.
• It tests whether the frequency counts in the various nominal
categories could be expected by chance or whether there is a
relationship.
• Chi-square is relatively uncommon in psychological research because
psychologists tend to opt for score data. However, in some
circumstances its use is necessary.
Chi-square, Differences between
samples of frequency data
• One-sample chi-square compares the frequencies obtained in each
category with a known expected frequency distribution.
• Two-sample chi-square uses a crosstabulation or frequency table for
two variables.
• This contains the frequencies in the various possible combinations of
categories for the two variables.
Chi-square, Differences between
samples of frequency data
• The disparity between the actual frequencies in the data and what
the frequencies would be if the null hypothesis were true is at the
heart of the calculation.
• The bigger the disparity, the bigger the value of chi-square and the
more one’s findings are statistically significant.
Chi-square, Differences between
samples of frequency data
• When the chi-square table has more than four cells (i.e. combinations
of categories), interpretation becomes difficult.
• It is possible to subdivide a big table into a number of smaller chi-
squares in order to facilitate interpretation.
• This is known as partitioning.
Chi-square, Differences between
samples of frequency data
• Sometimes data may violate the mathematical foundations of chi-
square too much.
• In these circumstances, the data may have to be modified to meet the
mathematical requirements, or an alternative measure such as the
Fisher exact test may be employed.
• We will not cover how to calculate it in this semester.
Table 18.1
Relationship between gender and wanting to
be a psychologist
Table 18.2
Relationship between sexual orientation and
religion
Table 18.3
Relationship between doll choice and ethnicity
Table 18.4
Stylised table for chi-square
Chi-square, Differences between
samples of frequency data
Theoretical issues
• Imagine a research study in which children are asked to choose
between two television programs, one violent and the other non-
violent.
• Some of the children have been in trouble at school for fighting and
the others have not been in trouble.
• The researcher wants to know if there is a relationship between the
violence of the preferred television program and having been in
trouble for fighting at school.
Table 18.5
Relationship between preferred TV programme
and fighting
Table 18.6
Relationship between preferred TV programme and
fighting including the marginal frequencies (column and
row frequencies)
Table 18.7
Contingency table including both observed and
expected frequencies
Significance Table 18.1
5% and 1% significance values of chi-square (two-tailed
test). Appendix F gives a fuller and conventional version
of this table

The degrees of freedom for the chi-square are

calculated using the following formula: df = (r-1)(c-
1) where r is the number of rows and c is the
number of columns.
Significance Table 18.1
5% and 1% significance values of chi-square (two-tailed
test). Appendix F gives a fuller and conventional version
of this table (Continued)
Figure 18.1
Conceptual steps for understanding the chi-
square test
Probability
• Although probability theory is at the heart of statistics, in practice
researchers need to know relatively little of this.
• The addition rule basically suggests that the probability of, say, any of
three categories occurring is the sum of the three individual
probabilities for those categories.
• The multiplication rule suggests that the probability of different
events occurring in a particular sequence is the product of the
individual probabilities.
Probability
• When any of us use a test of significance we are utilizing probability
theory.
• This is because statistical tests are based on it.
• What we meant by a 5% level of significance is that a particular event
(or outcome) would occur on five occasions out of 100.
• Although we have adopted the percentage system of reporting
probabilities in this course, statisticians would normally not write of a
5% probability.
Probability
Instead, they would express it as being out of a single event rather than
100 events.
Thus:
• .05 is an alternative way of writing 5%
• .10 is an alternative way of writing 10%
• 1.00 is an alternative way of writing 100%.
Probability
The addition rule is quite straightforward: It merely states that for several
mutually exclusive outcomes the sum of their probabilities adds up to 1.00.
• If you have a set of 150 people of whom 100 are women and 50 are men,
the probability of picking a woman at random is 100/150 or .667.
• The probability of picking a man at random is 50/150 or .333.
• However, the probability of picking either a man or a woman at random is
.667 + .333 or 1.00.
• In other words, it is certain that you will pick either a man or a woman.
Probability
The multiplication rule is about a set of events.
It can be illustrated by our set of 150 men and women, in which 100 are
women and 50 are men.
Again, the assumption is that the categories or outcomes are mutually
exclusive.
We could ask how likely it is that the first five people that we pick at
random will all be women, given that the probability of choosing a
woman on a single occasion is .667.
Probability
Probability