Unit 8: Inferential Statistics: Test of Difference

Unit 8: Inferential Statistics: Test of

Difference

Introduction
In the previous unit, we dealt with the process of hypothesis testing. The process
is fundamental to how statistics are used to connect empirical observations to
theories. It is now that we can apply it to two commonly used statistical tests for
comparing means. In this unit, you will learn about the different statistical tests
used for testing differences. In the testing hypothesis of comparison involving
one or two groups or variables, we will utilize one sample t-test, independent
samples t-test and paired samples t-test. Moreover, we have Analysis of
Variance (ANOVA) for comparing means of three or more groups. Emphasis
will be given to the interpretation of statistical analysis results, thus SPSS output
tables are provided to ease you from a mathematical approach (minimal
coverage/use of formula) and can focus and enjoy learning Statistics.

Learning Outcomes
At the end of this unit, you are expected to:
A. Use appropriate statistical test in testing the difference between and
among means;
B. Employ hypothesis testing in comparing mean difference; and
C. Interpret statistical analyses results utilizing SPSS outputs.

1
 Unit 8: Inferential Statistics: Test of Difference

Topic 1: T-test

Learning Objectives

Upon the completion of this topic, you are expected to:

1. Identify the type of t-test to be employ in a given study;
2. Employ t-test following the hypothesis testing procedure; and
3. Interpret t-test analyses results from SPSS output tables.

Presentation of Content

T-test
Some studies require the investigation on the differences between two means.
T-test is a statistical test that can be used to achieve this research objective. As
a parametric test, t-test assumes the normality of data distribution and the
other assumptions. Depending on the nature of samples, t-test can be classified
as follows:
a. One-Sample T-test
b. Independent Samples T-test
c. Paired Samples T-test

Interpreting the Results

In the interpretation, you need to report the following information:
1. The means and standard deviations for each group.
2. The t-computed value, degrees of freedom, and probability value.
3. Remember that the Levene’s test when significant, you report the
bottom row; if not significant, report the top row.

One-Sample T-test
By using a one-sample t-test, it is possible to test whether the mean of a
particular variable differs from a specified value (Wagner, 2015). Specifically,
is employed when we are comparing the population mean and the sample
mean. It will determine whether that difference is statistically significant or
not. The formula in determining the t-computed value is:
𝑥̅ − 𝜇 (𝑥̅ − 𝜇)√𝑛 − 1
𝑡= 𝑠 𝑜𝑟 𝑡=
⁄ 𝑛−1 𝑠
√

𝑑𝑓 = 𝑛 − 1

2
 Unit 8: Inferential Statistics: Test of Difference

Where:
𝑥̅ = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑚𝑒𝑎𝑛
𝜇 = 𝑝𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑚𝑒𝑎𝑛
𝑠 = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛
𝑛 = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒

How to run one-sample t-test using SPSS?

To run one-sample t-test using SPSS, follow these steps:
Analyze  Compare Means  One-Sample T-Test  (Input the test
value.)  Options  (Set the confidence interval and criterion for
excluding missing cases.)  Continue  OK

You can also watch a tutorial video on how to run one-sample t-test at:
https://youtu.be/C2Qa5d9ij0Y

Example:
Based from the records, Alice found out that the average highest year of
school completed by most of their students’ parents is 12 which corresponds to
the completion of high school education. She then randomly selected 120
respondents to test her hypothesis at 𝑎 = 0.05 that it should be higher.
Analysis using SPSS generated the following outputs:
One-Sample Statistics

N Mean Std. Deviation Std. Error Mean

Highest year of school

120 12.2500 1.83454 .16747
completed

One-Sample Test

Test Value = 12

95% Confidence Interval

Sig. (2- Mean of the Difference

t Df tailed) Difference Lower Upper

Highest year of
1.493 119 .138 .25000 -.0816 .5816
school completed

Solution:
To answer the problem, we will test the hypotheses that:
𝐻0 : There is no difference between the mean highest year school
completed of the parent population and of the sample.
𝐻𝑎 : The mean highest year school completed of the parent population
and of the sample differ.

3
 Unit 8: Inferential Statistics: Test of Difference

The first table reveals that the mean highest year of school completed by the
sample respondents is 12.25 with standard deviation of 1.83. One-sample test
results showed no significant difference between the two groups (𝑡 =
1.493, 𝑑𝑓 = 119, 𝑝 > 0.05). It means that the 𝐻0 should not be rejected,
hence, the average highest year of school completed by the sample
respondents is equal to the population mean.

Note: From the table, the p-value or the probability value is located under the
“Sig. (2-tailed)”. This value should be less than or equal to the set level of
significance to conclude that the two groups’ means differ.

Independent Samples T-test

Independent samples t-test enables us to compare the means of a particular
dependent variable across its independent groups. For this type, we can use the
formula:
𝑥̅1 − 𝑥̅2 (𝑛1 − 1)𝑠12 + (𝑛2 − 1)𝑠22
𝑡= 𝑎𝑛𝑑 𝑠𝑝 = √
1 1 𝑛1 + 𝑛2 − 2
𝑠𝑝 √𝑛 + 𝑛
1 2

𝑑𝑓 = 𝑛1 + 𝑛2 − 2

How to run independent samples t-test using SPSS?

To run independent samples t-test using SPSS, follow these steps:
Analyze  Compare Means  Independent-Samples T-Test  (Drag
the dependent variable into the “Test Variable(s)” area.)  (Drag the
independent variable into the “Grouping Variable” area.)  Define
Groups  (Select the groups to be compared by placing the
corresponding codes.)  Continue  Options  (Set the confidence
interval and criterion for excluding missing cases.)  Continue  OK

Data analysis employing independent samples t-test can also be watched at:
https://youtu.be/-qGFZFOQx7Q

Example:
From the 120 respondents, Alice is interested to which sex has higher number
of years completed. At 𝑎 = 0.05, she analyzed the data in SPSS and resulted
to the following. Is there a significant difference between the numbers of years
completed?

4
 Unit 8: Inferential Statistics: Test of Difference

Group Statistics

Std. Error
Sex N Mean Std. Deviation Mean

Highest year of school Female 56 12.4464 1.55995 .20846

completed Male 64 12.0313 1.76355 .22044

Independent Samples Test

Levene's Test for

Equality of

Variances t-test for Equality of Means

95% Confidence

Interval of the

Sig. (2- Mean Std. Error Difference

F Sig. t df tailed) Difference Difference Lower Upper

Highest year of Equal variances

.132 .717 1.357 118 .177 .41518 .30590 -.19058 1.02094
school assumed

completed Equal variances

1.368 117.983 .174 .41518 .30340 -.18563 1.01599
not assumed

Solution:
We will test the hypotheses that:
𝐻0 : The mean highest years completed by the male and female
respondents are equal.
𝐻𝑎 : The mean highest years completed by the male and female
respondents differ.

Descriptive statistics presented in the first table shows that the mean highest
years of school completed of the male and female respondents were 12.45
with 𝑆𝐷 of 1.56 and 12.03 with 𝑆𝐷 of 1.76 respectively. Independent samples
test analysis results indicated no significant difference between the male and
female respondents (𝑡 = 1.357, 𝑑𝑓 = 118, 𝑝 > 0.05). Hence, there is not
enough evidence to reject the null hypothesis. It is concluded that the means of
the number of years completed by male and female respondents are equal.

Note: If Levene’s test is significant (𝑝 < 𝑎), equal variances cannot be

assumed (it is violated) so the values from the bottom row of the table should
be used.

5
 Unit 8: Inferential Statistics: Test of Difference

Paired Samples T-test

This type of t-test can be employed when, for the given data set, the means of
all the values for each two variables are to be compared. For paired sample t-
test, we can use the formula:
𝑑̅
𝑡= 𝑎𝑛𝑑 𝑑𝑓 = 𝑛 − 1
𝑠𝑑
√𝑛 − 1

Where:
𝑑̅ = 𝑚𝑒𝑎𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑎𝑖𝑟𝑒𝑑 𝑜𝑏𝑠𝑒𝑟𝑣𝑎𝑡𝑖𝑜𝑛𝑠
𝑠𝑑 = 𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑 𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑡ℎ𝑒 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒𝑠
𝑛 = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑠𝑖𝑧𝑒

How to run paired samples t-test using SPSS?

To run independent samples t-test using SPSS, follow these steps:
Analyze  Compare Means  Paired-Samples T-Test  (Drag the
two variables into the “Paired Variables” area under “Variable1”
and “Variable2”. Additional pairs may be added in subsequent rows.)
 Options  (Set the confidence interval and criterion for excluding
missing cases.)  Continue  OK

You can watch a video on how to run using the software at:
https://youtu.be/vll22ZnFOP0

Example:
Teacher Alice believes that she can deliver the lesson effectively by
employing Collaborative Approach in teaching their topics on hypothesis
testing. She random sampled 50 students and administered the pre-test. After
the 2-week discussion, she gave the post-test and analyzed the data. She run
the test using SPSS and generated the following tables:

Paired Samples Statistics

Mean N Std. Deviation Std. Error Mean

Pair 1 Post-test 16.0600 50 2.27147 .32123

Pre-test 14.1800 50 2.27399 .32159

Paired Samples Correlations

N Correlation Sig.

Pair 1 Post-test & Pre-test 50 .891 .000

6
 Unit 8: Inferential Statistics: Test of Difference

Paired Samples Test

Paired Differences

95% Confidence

Std. Interval of the

Std. Error Difference Sig. (2-

Mean Deviation Mean Lower Upper t df tailed)

Pair Post-test -
1.88000 1.06215 .15021 1.57814 2.18186 12.516 49 .000
1 Pre-test

Solution:
The null and alternative hypotheses for the given problem are:
𝐻0 : There is no significant difference between the pre-test and post-test
mean scores of the respondents.
𝐻𝑎 : There is a significant difference between the pre-test and post-test
mean scores of the respondents.

Paired samples statistics revealed that in the pre-test, the mean score of the
respondents is 14.18 with SD of 2.27 while their score in the post-test
is 16.06 with SD of 2.27. Analysis employing paired samples t-test resulted to
a statistically significant difference between the pre-test and post-test
scores (𝑡 = 12.516, 𝑑𝑓 = 49, 𝑝 < 0.05). Therefore, the null hypothesis is
rejected. The post-test mean score is significantly higher than the pre-test
mean score. Moreover, the approach employed in teaching hypothesis testing
is effective.

7
 Unit 8: Inferential Statistics: Test of Difference

Topic 2. Analysis of Variance

Learning Objectives

Upon the completion of this topic, you are expected to:

1. Explain the concepts and purpose of Analysis of Variance (ANOVA);
4. Employ analysis of variance by following the hypothesis testing
procedure; and
5. Interpret ANOVA results from SPSS output tables.

Presentation of Content

Analysis of Variance
Analysis of Variance (ANOVA) is another inferential statistics for comparing
means like the t-test. However, it involves test for significance between mean
scores of at least two groups across one or more than one variable. It can be
used to test statistical significance using categorical independent variables in
conjunction with a continuous dependent variable.

One-way ANOVA
The analysis of variance to compare the means of three or more groups in the
light of one independent variable. It has two estimates of the population
variance enumerated as:
1. Between-Group Variance
2. Within-Group Variance

Between-Group Variance
This estimate involves the computation of variance by using the means of the
groups or between the groups.

Within-Group Variance
This estimate includes all the data in computing for the variance. It is not
affected by the differences in the means.

Note: If found out that there is no difference in the means, the between-group
variance estimate is approximately equal to the within-group variance
estimate, and the F-computed value will be approximately equal to 1.

F-computed Value
ANOVA is based on the comparison of variance between groups to the
variance within groups, emerging as the 𝐹 ratio or 𝐹 statistic. For the
computation, let us consider the following ANOVA table:
8
 Unit 8: Inferential Statistics: Test of Difference

Degrees
Source of Sum of Mean
of F-statistic
Variation Squares Squares
Freedom
(∑ 𝑥)2 (∑ ∑ 𝑥)2 𝑆𝑆𝑏 𝑀𝑆𝑏
Between 𝑆𝑆𝑏 = ∑ − 𝑘−1 𝑀𝑆𝑏 = 𝐹=
𝑛 𝑛𝑡 𝑘−1 𝑀𝑆𝑤
(∑ 𝑥)2 𝑆𝑆𝑤
Within 𝑆𝑆𝑤 = ∑ ∑(𝑥 2 ) − ∑ 𝑛𝑡 − 𝑘 𝑀𝑆𝑤 =
𝑛 𝑛𝑡 − 𝑘
Total 𝑆𝑆𝑡 = 𝑆𝑆𝑏 + 𝑆𝑆𝑤 𝑛𝑡 − 1

How to carry out one-way ANOVA using SPSS?

To carry out one-way ANOVA test using SPSS, follow these steps:
Analyze  Compare Means  One-Way ANOVA  (Drag the
dependent variable into the “Dependent List” area.)  (Drag the
independent variable to the “Factor” area.)  Options  (Select
needed descriptive statistics under “Statistics” and set the confidence
interval and criterion for excluding missing cases under “Missing
Values”.)  Continue  (Under “Equal Variances Assumed” and
“Unequal Variances Assumed”, select the “Post Hoc” tests to be
used.) Continue  OK

You can also watch a video on how to carry out Analysis of Variance using
SPSS at: https://youtu.be/rS3k8ONVN-o

Post Hoc Test

When the ANOVA is found to be significant, we must examine which two
groups differ significantly from the total number of groups; so post hoc tests
look at mean difference between different pairs. The Levene’s test will allows
us to determine which post hoc test we will use. The post hoc test for unequal
variances assumed must be used if the Levene statistic is significant but the
test for equal variances assumed if not.

Interpreting the Results

These information should be included in the write-up:
1. Means and standard deviations of both groups (needed when
significant difference is found).
2. F-statistic, between-groups 𝑑𝑓 and within-groups 𝑑𝑓, and p-value.
3. Interpretation from the post hoc test used (remember that the post hoc
is dependent on the Levene’s test)

9
 Unit 8: Inferential Statistics: Test of Difference

Example:
Mr. Manuel hypothesized that the students of the College of Teacher
Education enrolled for the school year 2020-2021 have equal knowledge on
the free tuition fee in tertiary education (R.A. 10931). He randomly selected
200 respondents from the current three year levels (there were no seniors
during this S.Y.) and conducted the survey. Test his hypothesis at 𝑎 = 0.5.
Refer to the succeeding SPSS output tables for the data analysis results.
Descriptives
Knowledge on R.A. 10931

95% Confidence
Interval for Mean

Std. Std. Lower Upper

N Mean Deviation Error Bound Bound Minimum Maximum

Freshman 60 2.9492 .47430 .06123 2.8266 3.0717 2.15 4.30

Sophomore 70 4.0343 .54237 .06483 3.9050 4.1636 3.05 4.95
Junior 70 4.4214 .51101 .06108 4.2996 4.5433 3.20 5.00
Total 200 3.8443 .79416 .05616 3.7335 3.9550 2.15 5.00

Test of Homogeneity of Variances

Knowledge on R.A. 10931

Levene Statistic df1 df2 Sig.

1.786 2 197 .170

ANOVA
Knowledge on R.A. 10931
Sum of Squares df Mean Square F Sig.

Between Groups 73.918 2 36.959 141.136 .000

Within Groups 51.588 197 .262
Total 125.506 199

10
 Unit 8: Inferential Statistics: Test of Difference

Multiple Comparisons
Dependent Variable: Knowledge on R.A. 10931
Tukey HSD

Mean 95% Confidence Interval

(I) Year (J) Year Difference (I- Std. Lower Upper

Level Level J) Error Sig. Bound Bound

Freshman Sophomore -1.08512* .09003 .000 -1.2977 -.8725

Junior -1.47226* .09003 .000 -1.6849 -1.2596

Sophomore Freshman 1.08512* .09003 .000 .8725 1.2977
Junior -.38714* .08650 .000 -.5914 -.1829
Junior Freshman 1.47226* .09003 .000 1.2596 1.6849

Sophomore .38714* .08650 .000 .1829 .5914

*. The mean difference is significant at the 0.05 level.

Solution:
In this problem, we will test the hypotheses that:
𝐻0 : The mean scores (knowledge on RA 10931) of the three year levels
are equal.
𝐻𝑎 : At least two of the three year levels differ in mean scores
(knowledge on RA 10931).

Data analysis employing one-way ANOVA resulted to a statistically

significant differences among the three year levels 𝐹 (2, 197) = 141.136, 𝑝 <
0.05). The null hypothesis is then rejected which means that at least two of the
three year levels significantly differ in knowledge on free tuition fee. Post hoc
Tukey test revealed that the juniors (𝑥̅ = 4.42, 𝑆𝐷 = 0.51) are more
knowledgeable than both sophomores (𝑥̅ = 4.03, 𝑆𝐷 = 0.54) and
freshmen (𝑥̅ = 2.94, 𝑆𝐷 = 0.47). Moreover, the sophomores (𝑥̅ = 4.03, 𝑆𝐷 =
0.54) are more knowledgeable than the freshmen (𝑥̅ = 2.94, 𝑆𝐷 = 0.47).

Note: To identify pair or pairs with significant mean difference from the post
hoc analysis results, determine those who have probability value of less than
or equal to the set level of significance. If the mean difference is positive, the
group in the first column has the higher mean compared to the corresponding
group in the second column.

11