Chapter 11: Inferential Statistics

Activity 11.1: Probability

Activity 11.2: Learning to Read a t-Table
Activity 11.3: Calculate a t-Test
Activity 11.4: Perform a Chi-Square Test
Activity 11.5: Conduct a t-Test
Activity 11.6: The Big Game

Activity 11.1:

Probability

1. Take a coin and test the hypothesis that the coin is dishonest, that is, that it comes
up heads more times than it comes up tails.
2. Flip the coin four times. Record the number of heads here:
3. Based on just this amount of data, would you accept or reject the hypothesis?
 Accept
 Reject
 Why
4. Suppose that you had gotten four heads (maybe you did). Would you then accept
or reject the hypothesis? Why?
5. Now flip the same coin four more times and record the number of heads. Do this
16times.Note: Each set of four flips can be considered a sample. For each sample,
the possible number of heads is 0, 1, 2, 3, or 4.)

Flip Number Number of Heads Flip Number Number of Heads

1 9
2 10
3 11
4 12
5 13
6 14
7 15
8 16

6. Next, tally the number of times each possible outcome occurred. Then change each to a
percent by dividing by 16.

Number of times outcome

Outcome occurred Percent
 0 heads
1 head
2 heads
3 heads
4 heads

7. If you were to use several coins and many more samples, you would (almost
certainly) arrive at a table that presents the results to be expected with an honest coin.
It would be very similar to that shown below.

Outcome Percent Probability

0 heads 6 .06

1 head 25 .25

2 heads 38 .38

3 heads 25 .25

4 heads 6 .06

Total 100 1.00

8. These percentages are actually probabilities. They tell us how often we would expect
each outcome to occur with an honest coin. Now, return to your original sample (the
first set of four flips of your coin). Would you now change your interpretation in any
way?

 Yes
 No
 If yes, how so?
 If not, why not?

Activity 11.2:

Learning to Read a t-Table

There are certain probabilities that researchers take as indicative of a stable non-chance
relationship. If the probability of obtaining a particular result or relationship in a sample
is less than .05 (one chance in 20), we customarily consider it to be statistically
significant -- that is, as probably not due to chance. If the probability is less than 5
percent (for example, 1 percent), we can be even more confident that we are not simply
dealing with chance. These values (1 percent and 5 percent) are frequently referred to as
levels of statistical significance.

Hence, when a research report states that a particular result (e.g., a difference in means)
or relationship (e.g., a correlation coefficient) is significant at the 5 percent level, it
means that the chances of the finding being simply a fluke (due to the particular sample
that was used) are less than 5 in 100. It means that the relationship or result is worth
noting and tentatively acceptable as a reproducible relationship for a specified
population. Note that statistical significance is not the same things a practical
significance, however.

A t-test is used to test the statistical significance of a difference between two means. A t-
table shows the value of t required for a particular result to be considered statistically
significant at various degrees of freedom (d.f.). Once the appropriate d.f. have been
determined and the t value has been calculated, the table indicates what a calculated t
value must be (the critical t value) to be considered statistically significant at different
levels of significance (e.g., .05, .01).

For example, a t value calculated for a sample having 10 d.f. using a one-tailed (i.e.,
directional) test must be at least 1.812 to be considered statistically significant at the .05
level, as shown in the normal curve and table presented below.
 What must the t value be for such a sample to be statistically at the .01 level?
 Now, suppose another sample has 25 degrees of freedom.
 What must be the t value be to be considered statistically significant at the .05
level?
 At the .01 level?

Proportion in Critical Region

Degrees of Freedom .005 .10 .05 .01
1 63.657 3.078 6.314 31.821
5 4.032 1.476 2.015 3.365
10 3.169 1.372 1.812 2.764
25 2.787 1.316 1.708 2.485
40 2.704 1.303 1.684 2.423
60 2.660 1.296 1.671 2.390
Activity 11.3:

Calculate a t-Test

A researcher wishes to compare the achievement of two groups of students who are
taught social studies by two different methods. Group I (n = 26) was taught by the inquiry
method. Group II (n = 26) was taught by the lecture method. The average (mean) score of
each group on a 100-point final examination was 85 for the lecture group (standard
deviation = 3 points) and 87 for the inquiry group (standard deviation = 2 points).

Use Table D.2 in the Appendix in the text to calculate a t-test for the difference in means.
Follow the steps in Table D.2 to fill in the table below (we have filled in the first two
lines).

Inquiry Group Lecture Group

Mean 87 85
Standard deviation (SD) 2 3
Standard error of the mean (SEM)
Standard error of the difference (SED)

t = (fill in the blank space) Degrees of freedom (d.f.) is obtained by n1 + n2 – 2 =

26 + 26 – 2 = 50

Consult the table below to determine if the obtained t-value is statistically significant at
the .05 level.

Proportion in Critical Region

Degrees of Freedom .005 .10 .05 .01

1 63.657 3.078 6.314 31.821
5 4.032 1.476 2.015 3.365
10 3.169 1.372 1.812 2.764
25 2.787 1.316 1.708 2.485
40 2.704 1.303 1.684 2.423
60 2.660 1.296 1.671 2.390

The difference between the inquiry and the lecture groups was (write the answer here) or
was not (write the answer here) statistically significant.
*Note: This exercise is designed to illustrate the procedure. When the number of cases in
each group, as here, is less than 30, a somewhat different formula should be used. It can
be found in any basic statistics text.

Activity 11.4:
Perform a Chi-Square Test

Chi-square is the most commonly used statistic for determining whether a relationship
between two categorical variables is statistically significant. The formula for calculating
chi-square is:

Suppose a researcher wishes to determine whether, at selected high schools, there is a

relationship between number of students enrolling in physical education courses and
participation in intramural sports. The data might look like that shown in Table 11.4.

Table 11.4

Number of students Number of students

enrolling in physical participating in
University education courses intramural sports Totals
Alpha 70 (60) 30 (40) 100
Beta 130 70
Kappa 160 140
Totals 360

The numbers in each of the cells in the table represent the observed frequencies (fO). To
obtain chi-square, go through the following steps:

1. Add up (total) both columns and rows. For example, the total number of students
enrolling in physical education courses is 360. Enter the totals in the table.

2. Calculate the proportion of the total frequency that falls in each row and column.
Thus, you should see that Alpha University has 100/600, or 1/6 of the total number
of students.

3. Multiply each row proportion by its column total. These are expected frequencies
(fΕ). Thus, for Alpha University, the expected frequencies for the number of students
enrolling in physical education courses is 1/6(360) = 60. For the number of students
participating in intramural sports, it is 1/6(240) = 40. Enter these in parentheses, as I
did for Alpha University, in each of the other cells.

4. For each cell, subtract the expected frequency (fΕ) from the obtained frequency (fO),
square the result, and then divide it by (fΕ). Fill in the results below.
 (70 – 60)2/60 = 102/60 = 100/60 = 1.67

(30 – 40)2/40 = -102/40 = 100/40 = __________

(130 – 120)2/120 = 102/120 = 100/120 = __________

(70 – 80)2/80 = -102/80 = 100/80 = __________

(160 – 180)2/180 = -202/180 = 400/180 = __________

(140 – 120)2/120 = 202/120 = 400/120 =

5. Next, total these six values, as symbolized by

to obtain a chi-square value of __________ (You should get 11.80)

6. Now, to determine whether this value is statistically significant, compare the

calculated value of chi-square to the values in the chi-square table below. To
determine the degrees of freedom (d.f.), multiply the number of rows minus one (r –
1) times the number of columns minus one (c – 1). In this case, it would be (3 – 1) x
(2 – 1) = 2. The chi-square table indicates that, with two d.f., a value of _________
is required for a result to be statistically significant. Is the value you obtained (11.80)
statistically significant? Yes __________ No __________

LEVELS
____________________________________
DEGREES OF
FREEDOM .10 .05 .02 .01 .001

1 2.71 3.84 5.41 6.64 10.83

2 4.60 5.99 7.82 9.21 13.82
3 6.25 7.82 9.84 11.34 16.27
4 7.78 9.49 11.67 13.28 18.46
5 9.24 11.07 13.39 15.09 20.52
10 15.99 18.31 21.16 23.21 29.59
20 28.41 31.41 35.02 37.57 45.32
25 34.38 37.65 41.57 44.31 52.62
30 40.26 43.77 47.96 50.89 59.70

To be significant, the X2 obtained from the data must be equal to or larger than the value
shown in the table.
Activity 11.5:
Conduct a t-Test

Identify a group of at least 10 people (classmates, friends, neighbors, etc.).

1. Divide the group into two roughly equal sub-groups, based on some characteristic
such as gender, age, height, etc.
2. Ask each person independently to pick a number from 1 to 9.

3. Calculate the mean of these numbers for each group.

4. Write the difference between the means of the two groups here.
5. Calculate a t-test for the difference in means using Table D-2 in the appendix of
the text. Use the portion of the t-table shown below to determine the statistical
significance of the difference in means that you obtained for the two groups.

Proportion in Critical Region

.005 .10 .05 .01
Degrees of Freedom
1 63.657 3.078 6.314 31.821
5 4.032 1.476 2.015 3.365
10 3.169 1.372 1.812 2.764
25 2.787 1.316 1.708 2.485
40 2.704 1.303 1.684 2.423
60 2.660 1.296 1.671 2.390

6. Were the results statistically significant?

 Yes
 No
7. What basic assumption must be met to justify using a t-test
8. Was it met?
 Yes
 No
 Explain why it was or was not.

Activity 11.6:
The Big Game

Ms. Jones has recently failed Bobby Thomasinki, Central High School’s star pitcher, on
is last exam in U.S. History, making him ineligible to play in the school’s championship
baseball game that will occur in two weeks. Bobby’s parents have asked for a conference
with you to discuss the matter. Bobby received a score of 16 out of 20 points on the exam
(which consisted of 20 true-false questions). Ms. Jones believes that Bobby must have
cheated, as he had failed the three previous exams she had given in the course and had
never received more than 8 points on any of these other true-false exams. Bobby claims
that he was lucky, that he passed the exam by guessing. Ms. Jones says this simply could
not happen, and presents the information in the table below as evidence to support her
belief. A total of 500 of Ms. Jones students have taken this exam over the last five years.
Bobby’s parents say it could happen, and that he therefore should be allowed to play in
the championship game. You think that, given the facts of the case, it was indeed unlikely
that Bobby could have passed the exam by guessing. Use probability to support your
decision.

PAST PERFORMANCE OF STUDENTS ON THIS QUIZ

NO. OF NUMBER PROPORTION NUMBER NUMBER PROPORTION

CORRECT OF OF OF CORRECT OF OF
RESPONSES STUDENTS STUDENTS RESPONSES STUDENTS STUDENTS

0 0 .000 11 79 .158
1 0 .000 12 61 .122
2 1 .002 13 39 .078
3 1 .002 14 18 .036
4 2 .004 15 7 .014
5 8 .016 16 1 .002
6 18 .036 17 1 .002
7 37 .074 18 0 .000
8 58 .116 19 0 .000
9 81 .162 20 0 .000
10 88 .176

1. Bobby received a score of 16. Do you think he was just guessing? (Hint: What is the
probability of a student getting a score at least as high as Bobby’s score, based on the
past performance of students who have taken this exam? Do you think these past
records reflect more than guessing?)

2. What would you decide in Bobby’s case? Was he guessing? What other possibilities
might explain his high score? Would you allow him to play in the big game?

Problem Sheet 11: Inferential Statistics

1. An appropriate inferential technique for my study would be:

2. Indicate whether you would use a parametric or a nonparametric technique and
why.
3. Indicate whether you would or would not do a significance test and why.
4. Indicate whether you would or would not calculate a confidence interval because:
5. I would not use any inferential techniques in my study because:
6. The type of sample used in my sample is:
7. The use of this type of sample in my study places the following limitation(s) on
my use of inferential statistics: