LU 6 Mean Comparison

Download as pdf or txt
Download as pdf or txt
You are on page 1of 73

Nueva Ecija University of Science and Technology

College of Arts and Sciences


MATHEMATICS AND SCIENCES DEPARTMENT

Psy 102: Psychological Statistics

Lecture prepared by: JAYNELLE G. DOMINGO, MSc. MathEd


Parametric vs. Nonparametric

Parametric statistics are based on assumptions about the


distribution of population from which the sample was taken.
Nonparametric statistics are not based on assumptions, that is, the data
can be collected from a sample that does not follow a specific distribution.

Assumptions for Parametric Tests


• Data in each comparison group show a Normal (or Gaussian)
distribution.
• Data in each comparison group exhibit similar degrees of
Homoscedasticity, or Homogeneity of Variance.
Parametric vs. Nonparametric
LESSON OBJECTIVES

At the end of the lesson, the students are expected to:

ü Differentiate different statistical tools for comparing means;

ü Perform one sample, paired samples, independent sample t-tests


and ANOVA/ANCOVA of sample data set using SPSS;

ü Analyze data n relation to specific research question;

ü Interpret results and write it in standard format.


Introduction

In the previous lessons, you learned how to calculate the descriptive


statistics of a set of scores/values of variables using SPSS and the steps in
hypothesis testing. For instance, means were computed to describe the
average score of a variable, and standard deviations were computed to
describe the average distance of the scores from the mean.
As stated in the earlier module, analysis goes beyond this purpose.
Many problems require that we decide whether or not a statement about the
mean or the standard deviation of a sample set of scores is true or false, and
this is what we referred to as hypothesis testing. In the process, we use the
descriptive statistics of the sample data and make inference/conclusion of the
population where the sample is obtained.
Introduction

Specifically, we might want to find out whether a sample mean


is different from a specified constant, or we would like to compare
whether two or more than two means from dependent groups or from
two or more independent groups are the same or different. This module
introduces the different statistical test for these kinds of problems.
Introduction

Why t-test? Why not z-test?

Assuming that all other factors have been considered, it is rational to use t-
test for even large sample, because the population variance is unknown for
almost all cases in practice. When we use sample variance for population
variance, we deal with t distribution. This is right philosophically.
Numerically, when n> 30, statistic t and statistic z are closed, or t
distribution is closed to normal distribution, and more important, the
influence of the degrees of freedom on the shape of distribution gradually
becomes null.
Introduction

The one-sample t-test is used to determine whether the mean


of a single variable differs from a specified constant.

Illustration 1. A researcher might want to test whether the average


achievement score for a group of students differs from 75.

Illustration 2. A researcher might want to test whether the mean IQ of


the sample respondents differs from 100 at the 95% confidence level.
One-Sample T-Test

HOW TO RUN ONE SAMPLE T-TEST


IN SPSS

Problem 1. A study tests whether there is a difference between the


average IQ scores of the samples and 100. (100 is the specified
constant or the test variable). 40 people were randomly selected
to participate and were given IQ test.
One-Sample T-Test

Research Question: Does the IQ scores of the sample differ from 100? (Is there a
significant difference between the IQ scores of the samples and 100?)

Null Hypothesis: There is no difference between the Mean IQ of the


respondents and 100 (𝐻𝑜: 𝑀𝑒𝑎𝑛!" = 100)

Alternative Hypothesis: There is a difference between the Mean IQ of the


respondents and 100 (𝐻𝑎: 𝑀𝑒𝑎𝑛!" ≠ 100)

Test Statistic: One sample t-test


One-Sample T-Test

Step 1. With SPSS open, Click File, Open, Data, and search where the Excel file
is saved. Specify: Files of type (*.xls,*xlsx,*xlsm). Choose the excel file:
DATA SET (one sample t test). Click open, then OK.

Step 2 . Click Analyze, Compare Means, One-Sample T Test

Step 3. Select IQ and put it in the Test Variable(s) box. Set Test Value to 100 ,
the specified constant.
(Note: We are testing whether the mean IQ of the sample is equal to 100)
One-Sample T-Test

Note: If we click options, the default confidence interval percentage is 95%. It


means that we are testing our hypothesis at alpha level of p<.05, that is: We
are 95% confident that when we get another sample from the same population
we will end up with the same result. You can adjust the alpha to .01 or .001. The
smaller the alpha, the more confident we are. In social sciences, the minimum is
0.05.

Step 4. Click continue, then OK. The result of the analysis is shown
below.
One-Sample T-Test
(一次样本T检验)
One-Sample T-Test

Results: The average IQ of the 40 samples is 96.63 and the Standard


Deviation is 9.63. Compared to 100, the mean difference is -3.375.
The computed t-statistics t = −𝟐. 𝟐𝟏𝟔, df (degrees of freedom) is
39: that is 𝑛 − 1 = 40 − 1 = 39, and the alpha level or p-value
Sig. (2-tailed) is 0.033, which is 𝒑 < 𝟎. 𝟎𝟓. Since 𝑝 < 0.05, we
reject the null hypothesis saying that the mean IQ of the samples is
equal to or not different to 100. The result of the one sample t-test
show that 96.63 is statistically lower than 100 at 𝑝 < .05. The
alternative hypothesis is the conclusion: There is a difference
between the mean IQ of the respondents and 100.
One-Sample T-Test

HOW TO WRITE THE RESULT OF ONE


SAMPLE T-TEST

One- sample T test was computed to determine whether


the average IQ of the respondents significantly differ from 100.
Results show that the average IQ of the respondents (Mean =
96.63, SD= 9.63) is significantly lower than 100, 𝑡("#) = −2.216,
𝑝 < 0.05. The null hypothesis is rejected.
One-Sample T-Test

Problem 2. Test whether the average weight of the sample respondents is equal
to 110 lbs.

Research Question: Does the average weight of the samples differ from 110
lbs?
Null Hypothesis: There is no difference between the Mean weight of the
respondents and 110 lbs. (𝐻𝑜: 𝑀𝑒𝑎𝑛#$%&'( = 110)
Alternative Hypothesis: There is a difference between the Mean weight of the
respondents and 110 lbs. (𝐻𝑎: 𝑀𝑒𝑎𝑛#$%&'( ≠ 110)
Test Statistic: One sample t-test
One-Sample T-Test

Using the same procedure enumerated above, we run a one


sample t–test using weight as the test variable and 110 as test
value.

The results are as follow:


One-Sample T-Test
(一次样本T检验)
One-Sample T-Test

Results: The average weight of the 40 samples is 110.70 and the


Standard Deviation is 3.212. Compared to 110, the mean difference
is very low, which is only 0.70. The computed t-statistics 𝑡 = 1.378,
df (degrees of freedom) is 39: that is 𝑛 − 1 = 40 − 1 = 39, and the
alpha level or p-value Sig. (2-tailed) is 0.176, which is 𝒑 > 𝟎. 𝟎𝟓.
Since 𝑝 > 0.05, we retain the null hypothesis saying that the mean
weight of the samples is equal to or not different with 110. The
result of the one sample t-test show that 110.70 is statistically
equal to 110 at 𝑝 > 0.05.
One-Sample T-Test

HOW TO WRITE THE RESULT OF ONE SAMPLE T-TEST

One- sample T test was computed to determine whether the


average weight of the respondents significantly differs from 110. Results
show that the average weight of the respondents (𝑀𝑒𝑎𝑛 = 110.70,
𝑆𝐷 = 3.312) is statistically equal to 110, 𝑡("#) = 1.378, 𝑝 > 0.05. The
null hypothesis is retained.
Paired Sample T-Test

The paired sample t-test procedure compares the means of two


variables for a single group. The procedure computes the differences between
values of the two variables for each case and tests whether the value differs
from 0.

Example 1. In a study on high blood pressure, all patients are measured at the
beginning of the study, given a treatment, and are measured again. Thus,
each subject has two measures, often called before and after measures.

Example 2. How would you know if there was a significant increase in the post
test of the students after they were exposed to a treatment (i.e. new
method of teaching)?
Paired Sample T-Test

HOW TO RUN PAIRED SAMPLE T-TEST IN SPSS

Problem 1 : A study was conducted to determine whether there is


a difference between the pretest and posttest scores of
students who reviewed for the licensure exam. Reviewees
were given a pretest, attended several review sessions, and
were given a post test.
Paired Sample T-Test

Research Question: Is there a significant difference between pretest and post


test scores?

Null Hypothesis: There is no difference between the pretest and post test
scores.
(𝐻𝑜: 𝑀𝑒𝑎𝑛)*$($+( = 𝑀𝑒𝑎𝑛),+(($+( )

Alternative Hypothesis: There is a difference between pretest and posttest


scores.
(𝐻𝑎: 𝑀𝑒𝑎𝑛)*$($+( ≠ 𝑀𝑒𝑎𝑛),+(($+( )

Test Statistic: Paired sample t-test


Paired Sample T-Test

Step 1. With SPSS open, Click File, Open, Data, and search where the Excel file
is saved. Specify: Files of type (*.xls,*xlsx,*xlsm). Choose the excel file:
DATA SET (paired sample t-test). Click open, then OK.

Step 2. Click Analyze, Compare Means, Paired- Samples T Test.

Step 3. Select pretest and posttest and put it in the Paired Variables as Pair 1.
Note that the default option for confidence interval percentage is .05%
(We can change it to .01 or .001)

Step 4. Click OK and we have the result.


Paired Sample T-Test
(配对样本T检验)
Paired Sample T-Test

Results: The Mean pretest of the 60 respondents is 74.43 and the


Standard Deviation is 6.39 while the Mean posttest is 82.62 and
Standard Deviation is 5.12 . The mean difference (pretest-posttest)
is –7.60. The computed t- statistics 𝑡 = – 9.40, df (degrees of
freedom) is 59: that is 𝑛– 1 = 60– 1 = 59), and the alpha level or
p-value Sig. (2-tailed) is 0.000, which is p<0.05, even less than
p<0.001. Since p<.05, we reject the null hypothesis saying that the
mean pretest is not different with the mean posttest. The result of
the paired sample t-test shows that 82.03 is significantly higher
than 74.43 at p<0.001.
Paired Sample T-Test

HOW TO WRITE THE RESULT OF PAIRED- SAMPLE T-TEST (OR DEPENDENT


SAMPLES)

Paired sample t-test was computed to determine whether the Mean


posttest of the students who reviewed for the licensure examination
significantly differs with the Mean pretest. Results show that the Mean
posttest (𝑀𝑒𝑎𝑛 = 82.03, 𝑆𝐷 = 5.12) is significantly higher than the Mean
pretest (𝑀𝑒𝑎𝑛 = 74.43, 𝑆𝐷 = 6.39) , 𝑡(./) = −9.4, 𝑝 < 0.001. The null
hypothesis is rejected. It implies that the review sessions have significant effect
on the posttest scores.
Paired Sample T-Test

Problem 2. An intervention program was designed to help


students improve Math performance. They were given a test
before the intervention program. After the intervention program
was implemented, the same students were given another test.

Research Question: Is there a significant difference on the average


Math scores of the students before and after the intervention
program?
Paired Sample T-Test

Null Hypothesis: There is no significant difference on the average


Math scores of the students before and after the intervention
program.
(𝐻𝑜: 𝑀𝑒𝑎𝑛1$2,*$ = 𝑀𝑒𝑎𝑛32($* )

Alternative Hypothesis: There is a difference between the scores


before and after
(𝐻𝑎: 𝑀𝑒𝑎𝑛%&'()& ≠ 𝑀𝑒𝑎𝑛*'+&) )

Test Statistic: Paired sample t-test


Paired Sample T-Test

Using the same data set, we perform the same procedure , using before
and after as paired variables.

The results are as follow:


Paired Sample T-Test
(配对样本T检验)

Using the same data set, we perform the 但是,分析超出


same procedure , using before and after as 了此目的。许多问题要求
paired variables. 我们决定关于分数样本集
的平均值或标准偏差的陈
述是对还是错。在统计中,
此陈述称为假设,而关于
该陈述是对还是错的决策
The results are as follow: 称为假设检验。
Paired Sample T-Test

Results. The mean difference is small, only −1.13. The t-value is


also small, −1.52 and the p-value is 0.134 (which is greater
than .05). Since 𝑝 > 0.05, we retain the null hypothesis
stating that the two means are equal. It means that the Mean
score before (73.88) does not differ with the Mean score after
(75.02).
Paired Sample T-Test

HOW TO WRITE THE RESULT OF PAIRED SAMPLE T-TEST (OR


DEPENDENT SAMPLES)

Paired- sample T test was computed to determine the


significant difference on the average Math scores of the students
before and after the intervention. Results show that the scores before
the intervention (𝑀𝑒𝑎𝑛 = 73.88, 𝑆𝐷 = 6.20) is equal to the scores
after the intervention (𝑀𝑒𝑎𝑛 = 75.02, 𝑆𝐷 = 6.56), 𝑡(%#) = −1.13, 𝑝 >
0.05. The null hypothesis is retained. It implies that the intervention
program does not affect Math performance.
Independent Sample T-test

The independent sample t-test compares means for two


groups of cases. Ideally, for this test, the subjects should be
randomly assigned to two groups, so that any difference in
response is due to the treatment (or lack of treatment) and not to
other factors.
Independent Sample T-test

Example 1. Patients with diabetes are randomly assigned to a placebo group


and a treatment group. The placebo subjects receive an inactive pill, and
the treatment subjects receive a new drug that is expected to lower sugar
levels. After the subjects are treated for two months, the two-sample or
independent-samples T-test is used to compare the sugar levels for the
placebo group and the treatment group. Each patient is measured once
and belongs to one group.
Example 2. Students were randomly assigned as Group 1 and were taught using
the new method of teaching (treatment group) and Group 2 who were
taught the traditional teaching method (control group). After some time,
the students were given an achievement test. Is Group 1 better than
Group 2 on the test?
Independent Sample T-test

HOW TO RUN INDEPENDENT SAMPLE T-TEST IN SPSS

Problem 1: A study was conducted to determine whether there is a


difference between the average Math scores of students who
were taught the new method of teaching (Group 1) and the
traditional method of teaching (Group 2).
Independent Sample T-test

Research Question: Is there a significant difference on the Math scores between those
who were taught the traditional and new method of teaching?

Null Hypothesis: There is no significant difference on the Math scores between students
who were taught the traditional and the new method of teaching
(𝐻𝑜: 𝑀𝑒𝑎𝑛!"#$%!%&'#( = 𝑀𝑒𝑎𝑛')* ,)!-&$ )

Alternative Hypothesis: There is a significant difference on the Math scores between


students who were taught the traditional and the new method of teaching
(𝐻𝑎: 𝑀𝑒𝑎𝑛!"#$%!%&'#( ≠ 𝑀𝑒𝑎𝑛')* ,)!-&$ )

Test Statistic: Independent sample t-test


Independent Sample T-test

Step 1. With SPSS open, Click File, Open, Data, and search where the Excel file is saved.
Specify: Files of type (*.xls,*xlsx,*xlsm). Choose the excel file: DATA SET (independent
sample t test) .Click Open, then OK.

Step 2. Click Analyze, Compare Means, Independent -Samples T test

Step 3. Select Math score and put it as Test Variable(s), select group and put it in Grouping
Variable. The default option for confidence interval percentage is .05 (We can change it
to .01 or ,.001)

Step 4. Click Define groups. Since 1 is the code we used for Group 1 (new method) and 2 for
Group 2 (traditional), we will use the same numbers in defining the groups.

Step 5. Click Continue, then OK. We have the result.


Independent Sample T-test
(独立样本T检验)
Independent Sample T-test

Results: The Mean score in Math of Group 1, n=51 (new method of teaching) is
78.96 and the Standard Deviation is 8.93, while the Mean score of Group
2, n=49 (traditional method) is 76.84 and Standard Deviation is 7.98. The
mean difference (Group 1 – Group 2) is 2.214. The computed t-statistics
𝑡 = 1.253 (when we assumed that variances of Group 1 and 2 are equal
as shown by the Levene’s test for Equality of Variances*), df (degrees of
freedom) is 98: that is 𝑛 − 2 = 100 − 2 = 98, and the alpha level or p-
value Sig. (2-tailed) is 0.213 which is p>0.05. Since p>0.05, we retain the
null hypothesis saying that the mean score of Group 1 is not different with
the mean scores of Group 2. The result of the independent sample T-test
shows that 78.96 is statistically equal to 76.84.
Independent Sample T-test

*The Levene’ test for Equality of Variances is a test of homoscedasticity – the


characteristic of the distribution wherein the variance of one group should be
roughly the same for another group. Meaning, the variance of group 1 should
be equal to the variance of group 2. The Levene’s test , 𝐹 = 2.033, 𝑝 = 0.157,
indicates that there is no significant difference between the variances of Group
1 (SD=8.93) and Group 2 (SD=7.98) . Therefore, the data is homoscedastic. If the
data is homoscedastic, it is also assumed that the scores are normally
distributed (normal curve).
Independent Sample T-test

HOW TO WRITE THE RESULT OF THE INDEPENDENT SAMPLE T-TEST

To determine whether the scores in Math of Group 1 (taught


with the new method) differs with Group 2 (taught with the traditional
method), independent sample T-test was performed. Results show that
the scores of Group 1 (𝑀𝑒𝑎𝑛 = 78.96, 𝑆𝐷 = 8.92) is statistically the
same as Group 2 scores (𝑀𝑒𝑎𝑛 = 76.84, 𝑆𝐷 = 7.98), 𝑡(#&) = 1.25, 𝑝 >
0.05. The null hypothesis is retained. It implies that the new method of
teaching does not affect Math performance.
Independent Sample T-test

Problem 2. A study was conducted to determine the effect of an


intervention program to improve students’ intrinsic motivation. Students
were randomly assigned as Group 1 (control group) and Group 2
(experimental group). After several sessions, students’ intrinsic
motivation were measured using a standardized test. (Students’
responses on a 5–point Likert scale were averaged, the higher average
score, the higher intrinsic motivation).
Independent Sample T-test

Research Question: Is there a significant difference on the intrinsic motivation between


the control group and the experimental group?

Null Hypothesis: There is no significant difference on the intrinsic motivation of the


control group and experimental group
(𝐻𝑜: 𝑀𝑒𝑎𝑛.&'!"&( = 𝑀𝑒𝑎𝑛)/0)"%,)'!#( )

Alternative Hypothesis: There is significant difference on the intrinsic motivation of the


control group and experimental group
(𝐻𝑎: 𝑀𝑒𝑎𝑛.&'!"&( ≠ 𝑀𝑒𝑎𝑛)/0)"%,)'!#( )

Test Statistic: Independent sample t-test


Independent Sample T-test

Using the same procedure, we run an independent- samples T test, with


intrinsic motivation as test variable, and Group_A as grouping variable.

The results are as follow:


Independent Sample T-test
(独立样本T检验)

Using the same procedure, we run an 但是,分析超出


independent- samples T test, with intrinsic 了此目的。许多问题要求
motivation as test variable, and Group_A as 我们决定关于分数样本集
的平均值或标准偏差的陈
grouping variable.
述是对还是错。在统计中,
此陈述称为假设,而关于
该陈述是对还是错的决策
The results are as follow: 称为假设检验。
Independent Sample T-test

Results: The Mean score in intrinsic motivation of Group 1, n=38 (control


group) is 2.54 and the Standard Deviation is .47, while the Mean score of
Group 2, n=42 (experimental group) is 4.39 and Standard Deviation is 0.40
. The mean difference (Group 1 – Group 2) is –1.85. The computed t-
statistics is –18.944 (when we assumed that variances of group 1 and 2
are equal as shown by the Levene’s test), df (degrees of freedom) is 78:
that is 𝑛 − 2 = 80 − 2 = 78), and the alpha level or p-value, Sig. (2-
tailed) is 0.000, which is p<0.001. Since p<0.05, we reject the null
hypothesis saying that the mean score of Group 1 is not different with the
mean scores of Group 2. The result of the independent-samples T test
shows that 4.39 is significantly higher than 2.54 at p<0.001.
Independent Sample T-test

HOW TO WRITE THE RESULT OF THE INDEPENDENT SAMPLE T-TEST

To determine whether the intrinsic motivation of the control group


differs with the experimental group, independent-samples T test was
performed. Results show that the average score in intrinsic motivation of
students who were given the intervention program (𝑀𝑒𝑎𝑛 = 4.39, 𝑆𝐷 = 0.40)
is significantly higher than the average score of those who did not receive the
intervention (𝑀𝑒𝑎𝑛 = 2.54, 𝑆𝐷 = 0.47), 𝑡(45) = −18.944, 𝑝 < 0.001. The
null hypothesis is rejected. It implies that the intervention program was
effective in improving students’ intrinsic motivation.
One-way ANOVA (Analysis of Variance)

The one-way ANOVA produces a one-way analysis of


variance for a quantitative dependent variable by a single factor
(independent) variable. Analysis of variance is used to test the
hypothesis that several means (2 or more means) are equal. In
addition to determining that differences exist among the means,
we may want to know which pair/s of mean differ.
One-way ANOVA (Analysis of Variance)

Example 1. Is there a significant difference on the test scores of


students who had self-review, pair review or group review?

Example 2. Is there a significant difference on the teaching


performance of university teachers in terms of educational
attainment/qualification (BS, MS or PhD)?
One-way ANOVA (Analysis of Variance)

HOW TO RUN ONE-WAY ANALYSIS OF VARIANCE (ANOVA) IN


SPSS

Problem 1: A study was conducted to determine whether there is a


significant difference on the Science performance among students
who had self-review, pair-review and group review. After the test
and scores were recorded, the students were asked what type of
review technique they had. (Codes were 1-self, 2-pair, 3-group)
One-way ANOVA (Analysis of Variance)

Research Question: Is there a significant difference on the Science scores among


students who had self-review, pair-review and group review? (or Does type of
review affect Science performance?)
Null Hypothesis: There is no significant difference on the Science scores among students
who had self-review, pair-review and group review. (or Type of review does not
affect Science performance.)
(𝐻𝑜: 𝑀𝑒𝑎𝑛1)(2 = 𝑀𝑒𝑎𝑛0#%" = 𝑀𝑒𝑎𝑛3"&40 )
Alternative Hypothesis: There is significant difference on the Science scores among
students who had self-review, pair-review and group review or Type of review
affects Science performance.
(𝐻𝑎: 𝑀𝑒𝑎𝑛1)(2 ≠ 𝑀𝑒𝑎𝑛0#%" ≠ 𝑀𝑒𝑎𝑛3"&40 )
Test Statistic: One-way ANOVA
One-way ANOVA (Analysis of Variance)

Step 1. With SPSS open, Click File, Open, Data, and search where the Excel file is saved.
Specify: Files of type (*.xls,*xlsx,*xlsm). Choose the excel file: DATA SET (one-way
anova). Click Open, then OK.

Step 2. Click Analyze, Compare Means, One-Way ANOVA.

Step 3. Select Science score and put it on the Dependent List. Select review technique
and put it as the Factor. The default option for confidence interval percentage is
.05% (You can change it to .01 or .001)

Step 4. Select Options and Click Descriptive & Homogeneity of variance test, Then click
Continue
One-way ANOVA (Analysis of Variance)

Step 5. Click Post Hoc*, then Select 1 (Scheffe and Bonferroni are the most
commonly used). Click Continue

*Post hoc analysis is used to determine which pair of means are equal or
different. If the ANOVA result is significant, it indicates that at least 1 pair of
means is different (the highest value and the lowest value are significantly
different). It could be 2 pairs, or 3 pairs. If the ANOVA is not significant, no need
to analyze the post.

Step 6. Click OK, and we have the result.


One-way ANOVA (Analysis of Variance)
(单向方差分析)
One-way ANOVA (Analysis of Variance)
(单向方差分析)
One-way ANOVA (Analysis of Variance)

Results 1. Descriptive statistics show that the Mean of Group 1 (self-review, n=16) is
77.25, Group 2 (pair review, n=31) is 80.16, and Group 3 (group review, n=13) is
80.69. Levene’s test show that the data is homoscedastic, (F=0.577, p>0.05,
indicating that there are no significant differences on the variance of the three
groups.

Results 2. ANOVA Table shows that the F-value (Between Groups) is 0.668, and p=value
(Sig.) is 0.517, which is greater than 0.05. The df between groups is 2 (3 groups –
1) and within group is 57 (n–2–1 = 60–2–1). Since the p-value>0.05, we retain the
null hypothesis stating that the three means are equal. Statistically, 77.25, 80.16
and 80.69 are equal. Note: There is no need to examine post hoc analysis because
the overall ANOVA indicates no significant differences.
One-way ANOVA (Analysis of Variance)

HOW TO WRITE THE RESULT OF ONE-WAY ANOVA

One way analysis of variance was performed to determine the effect of type of review
to Science scores. Results show that type of review has no significant effect to Science sores,
𝐹(",$%) = 0.668, 𝑝 > 0.05. The average scores of those who self-reviewed (𝑀𝑒𝑎𝑛 = 77.25, 𝑆𝐷 =
11.03), pair reviewed (Mean=80.16, SD = 8.71) and group reviewed (𝑀𝑒𝑎𝑛 = 80.69, 𝑆𝐷 = 7.39)
are equal. The null hypothesis is retained. It implies that the type of review does not affect the
performance of the students in Science.
One-way ANOVA (Analysis of Variance)

Problem 2. Determine the effect of type of review to the students’ English scores.

Research Question: Is there a significant difference on the English scores among


students who had self-review, pair-review and group review?
Null Hypothesis: There is no significant difference on the English scores among students
who had self-review, pair-review and group review.
(𝐻𝑜: 𝑀𝑒𝑎𝑛1)(2 = 𝑀𝑒𝑎𝑛0#%" = 𝑀𝑒𝑎𝑛3"&40 )
Alternative Hypothesis: There is significant difference on the English scores among
students who had self-review, pair-review and group review.
(𝐻𝑜: 𝑀𝑒𝑎𝑛1)(2 ≠ 𝑀𝑒𝑎𝑛0#%" ≠ 𝑀𝑒𝑎𝑛3"&40 )
Test Statistic: One-way ANOVA
One-way ANOVA (Analysis of Variance)

Using the same data set, we run an ANOVA with English as the
dependent variable.

The results are as follow:


One-way ANOVA (Analysis of Variance)

Results: The Mean of Group 1 (self review) is 78.56, Group 2 (pair review) is 79.10 and Group 3
(group review) is 90.15. The data is homoscedastic as shown in the Levene’s test, (F = 0.374,
p>0.05). The ANOVA model is significant, F-value = 6.599, p=0.003, which is less than 0.05.
Since p-value <0.05, we reject the null hypothesis stating that there is no difference between
the means. The result of the One Way ANOVA indicates that there is significant difference.

Since the ANOVA is significant, we examine the post hoc analysis to determine which pair/s
of means are different. The multiple comparison using Scheffe shows that the difference
between Group 1 (self) and 2 (pair) is not significant (p=0.985, which is p>0.05). However, the
difference between Group 1 (self) and Group 3 (group) is significant (p=0.011, which is
p<0.05). The difference between Group 2 (pair) and Group 3 (group) is also significant
(p=0.005, which is p<0.05). Of the three pairs of means, two are found to be significantly
different: Group 1 vs Group 3 and between Group 2 vs Group 3.
One-way ANOVA (Analysis of Variance)

HOW TO WRITE THE RESULT OF ANOVA (WITH POST HOC)

One way analysis of variance was performed to determine the effect of type of review to
English scores. Results show that type of review has significant main effect to English scores, 𝐹(",$%) =
6.599, 𝑝 < 0.01. The average scores of those who group reviewed (𝑀𝑒𝑎𝑛 = 90.15, 𝑆𝐷 = 10.96) was the
highest. Post hoc analysis using Scheffe showed that those who group reviewed scored significantly higher
than those who pair reviewed (𝑀𝑒𝑎𝑛 = 79.10, 𝑆𝐷 = 9.53) and self- reviewed (𝑀𝑒𝑎𝑛 = 78.56, 𝑆𝐷 =
9.67). On the other hand, there is no significant difference between self and pair review. The null
hypothesis is rejected. It implies that the type of review affects the performance of the students in
English.
One-way ANCOVA (Analysis of Covariance)

The Analysis of Covariance (ANCOVA) is a one way-ANOVA


with one or more covariates. A covariate is a variable that might
potentially affect the dependent variable, and the aim of the
analysis is to remove its effect by making it constant.
One-way ANCOVA (Analysis of Covariance)

HOW TO RUN ANALYSIS OF CO-VARIANCE (ANCOVA) IN SPSS

Example. A researcher aims to determine the effect of type of review to


the students’ English scores. English test scores is influenced by verbal IQ
scores and the researcher will control its effect. (The independent
variable is type of review, the dependent variable is English score, and
the covariate is the verbal IQ scores)
One-way ANCOVA (Analysis of Covariance)

Research Question: Is there a significant difference on the English scores among


students who had self-review, pair-review and group review, controlling
for the effects of verbal IQ scores?
Null Hypothesis: There is no significant difference on the English scores among
students who had self-review, pair-review and group review, when verbal
IQ scores are controlled.
Alternative Hypothesis: There is significant difference on the English scores
among students who had self-review, pair-review and group review, when
verbal IQ scores are controlled
Test Statistic: One-way ANCOVA
One-way ANCOVA (Analysis of Covariance)

Step 1. With SPSS open, Click File, Open, Data, and search where the Excel file
is saved. Specify: Files of type (*.xls,*xlsx,*xlsm). Choose the excel file:
DATA SET (ancova). Click Open, then OK.
Step 2. Click Analyze, General Linear Model, Univariate
Step 3. Select English score and put it on the Dependent Variable. Select review
technique and put it on the Fixed Factor(s), and put the Verbal IQ scores
as Covariate(s).
Step 4. Select Options. Display Means for (OVERALL) and review technique.
Click Descriptive statistics and Estimates of effect size. Then click
Continue.
Step 5. Click OK, and we have the result.
One-way ANCOVA (Analysis of Covariance)

Result 1: The Mean of Group 1 (self review) is 78.56, Group 2 (pair review) is 79.10 and
Group 3 (group review) is 90.15. This is the same result for the descriptive
statistics.

Results 2. ANOVA Table shows that there is a significant main effect of Verbal IQ scores
to English scores, 𝐹(6,89) = 1197.64, and p-value (Sig.) is 0.000. The independent
variable (review technique) does not affect English scores, 𝐹(;,89) = 0.020, and p–
value (Sig.) is 0.980.
The independent variable (review technique) affects English test scores, when the
model does not include the covariate (Verbal IQ scores). Using analysis of
covariance (ANCOVA), the effect of review technique is no longer significant. In
other words, English test scores are influenced by verbal IQ scores, and not the
type of review conducted.
One-way ANCOVA (Analysis of Covariance)

HOW TO WRITE THE RESULT OF ANCOVA

One way analysis of covariance was performed to determine


the effect of type of review to English scores, controlling for the effects
of Verbal IQ scores. Results show that verbal IQ scores affect English test
scores, 𝐹(;,%=) = 1997.64, 𝑝 < 0.001. However, type of review does not
affect English test scores, 𝐹(>,@A) = 0.020, 𝑝 > 0.05.
References

Dela Rosa, E. D. (2019). Learning Module in Statistics with SPSS


Applications. Philippine Copyright 2019.

Deauna, M. C. (2011). Applied Educational Statistics 1. Quezon


City, Philippines C & E Publishing 2011

You might also like