Professional Documents
Culture Documents
In a study to test the significance of effectiveness of negotiation for meaning towards the
improvement of students classroom interaction skill, a researcher has randomly selected
10 students for the research. Data on the students classroom interaction skill was taken
before the introduction of the negotiation for meaning as well as after the treatment (post-
training to 4 months follow-up).
Null hypothesis:
There is no evidence that students classroom interaction skill changes
significantly during three different training (pre- training, post- training and 4
months follow-up).
Alternative hypothesis:
There is evidence that students classroom interaction skill changes significantly
during three different training (pre- training, post- training and 4 months follow-
up).
Assessing Normality
Cases
Median 3.0000
Variance 4.053
Minimum 1.00
Maximum 8.00
Range 7.00
Median 5.0000
Variance 4.568
Range 7.00
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
Conclusion: Using Shapiro-Wilk test for normality, because the sample size is small, n =20.
Since all the variable significant value greater than alpha level, we fail to reject null hypothesis
and accept null hypothesis. Thus the data does not violate the normality assumption. That means
the data is normally distributed.
Null hypothesis:
There is no evidence that students classroom interaction skill changes significantly during three
different training (pre- training, post- training and 4 months follow-up).
Alternative hypothesis:
There is evidence that students classroom interaction skill changes significantly during three
different training (pre- training, post- training and 4 months follow-up).
General Linear Model
Within-Subjects Factors
Measure: MEASURE_1
Dependent
train Variable
1 Pretraining
2 Posttraining
3 Fourmonths
Descriptive Statistics
Multivariate Testsa
Partial Eta
Effect Value F Hypothesis df Error df Sig. Squared
a. Design: Intercept
Within Subjects Design: train
b. Exact statistic
Epsilonb
a. Design: Intercept
Within Subjects Design: train
b. May be used to adjust the degrees of freedom for the averaged tests of significance. Corrected tests are displayed in the Tests of Within-
Subjects Effects table.
The Mauchys test of spherecity indicated Chi-square statistics has a value of 4.22 and a p-value
of 0.121. Therefore we fail to null hypothesis, accept null hypothesis and concluded that met the
assumption of sphericity.
Alternative hypothesis:
There is evidence that students
classroom interaction skill changes
significantly during three different
training (pre- training, post- training and
4 months follow-up).
.05 .000 Reject null hypothesis,
Wilks Lambda test accept alternative
hypothesis.
The Wilks Lambda test is used since sphericity assumed. The test statistic (F) equals 12.851,
with a corresponding p-value .000. Therefore, we reject null hypothesis, accept alternative
hypothesis: There is evidence that students classroom interaction skill changes significantly
during three different training (pre- training, post- training and 4 months follow-up). Mean of pre
training is lowest (M = 3.5000, SD = 2.01311), which before training is lowest. After training,
the mean increase, post training (M = 5.5500, SD = 2.48098). However, after four months, the
mean dropped, the score after 4 months follow up (M = 5.4000, SD = 2.13739). Hence, there is
evidence that students classroom interaction skill changes significantly during three different
training (pre- training, post- training and 4 months follow-up), Wilks Lambda = .421, F (2, 18)
= 12.851, p = .000.
c. Calculate and interpret an appropriate effect size
The Partial eta squared given in the Multivariate Test output box indicated the value is .588, this
result suggests a large effect size, which represent 58.8%.
2. Using the SURVEY.SAV file, select a set of variables (YOUR CHOICE!!) suitable to
conduct a Factorial- ANOVA (2 factor).
Null hypothesis:
Ho1: There is no significant difference between genders in life satisfication.
Ho2: There is no significant difference in life satisfication for three age groups.
Ho3: There is no significant difference in the effect of age on life satisfication
for males and females.
Alternative hypothesis:
Ha1: There is significant difference between genders in life satisfication.
Ha2: There is significant difference in life satisfication for three age groups.
Ha3: There is significant difference in the effect of age on life satisfication for
males and females.
Assessing Normality
Cases
Median 23.00
Variance 45.827
Minimum 5
Maximum 35
Range 30
Interquartile Range 9
Extreme Values
2 82 35
3 112 35
4 186 35
5 213 35a
Lowest 1 41 5
2 30 5
3 13 5
4 10 5
5 7 5
a. Only a partial list of cases with the value 35 are shown in the table of
upper extremes.
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
Kolmogorov-Smirrnov test
Conclusion: Using Kolmogorov-Smirrnov test for normality, since the significant value lower
than alpha level, sig. value is .000, we reject null hypothesis and accept alternative hypothesis.
Thus the data violate the normality assumption. However, this is quite common in larger group
(Pallant, 2013).
Test the hypothesis:
Null hypothesis:
Ho2: There is no significant difference in life satisfication for three age groups.
Ho3: There is no significant difference in the effect of age on life satisfication for males and
females.
Alternative hypothesis:
Ha2: There is significant difference in life satisfication for three age groups.
Ha3: There is significant difference in the effect of age on life satisfication for males and
females.
Between-Subjects Factors
Value Label N
2 30 - 44 153
3 45+ 136
sex 1 MALES 185
2 FEMALES 251
Descriptive Statistics
Dependent Variable: Total life satisfaction
(I) age 3 groups (J) age 3 groups (I-J) Std. Error Sig. Lower Bound Upper Bound
Subset
age 3 groups N 1
30 - 44 153 21.63
18 - 29 147 22.63
45+ 136 22.95
Sig. .223
Group Statistics
Alternative hypothesis:
Ha1: There is significant difference
between genders in life
satisfaction.
Alternative hypothesis:
Ha2: There is significant difference in
life satisfaction for three age groups.
Alternative hypothesis:
Ha3: There is significant difference in
the effect of age on life satisfaction
for males and females.
The effect size for the age groups variables was small (partial eta squared = .006). This
represents only 0.6% of the variance in life satisfaction scores explained by three age groups.
According to the guidelines proposed by Cohen 1988, for interpreting these values are:
.01 = small effect, .06 = moderate effect and .14 = large effect.
The effect size for the gender groups variables was small (partial eta squared = .008). This
represents only 0.8% of the variance in life satisfaction scores explained by gender.
3. Using the BP. SAV file, select a set of variables (YOUR CHOICE!!) suitable to conduct
MANCOVA.
Null hypothesis:
Race is not the factor to the diastolic bp cold pressor and heart rate cold pressor, by controlling
age.
Alternative hypothesis:
Race is the factor to the diastolic bp cold pressor and heart rate cold pressor, by controlling age.
Cases
Median 77.0000
Variance 215.461
Minimum 34.00
Maximum 132.00
Range 98.00
Median 83.0000
Variance 236.921
Minimum 47.00
Maximum 144.00
Range 97.00
2 83 124.00
3 155 121.00
4 59 111.00
5 123 110.00
Lowest 1 22 34.00
2 33 42.00
3 112 45.00
4 127 46.00
5 75 49.00
heart rate cold pressor Highest 1 117 144.00
2 180 126.00
3 137 124.00
4 48 120.00
5 134 119.00
2 32 48.00
3 9 48.00
4 174 51.00
5 67 51.00
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
3.00 4 . 788
4.00 5 . 1133
3.00 5 . 788
6.00 6 . 233344
11.00 6 . 66678888899
19.00 7 . 0000011112333344444
21.00 7 . 555567777778888899999
31.00 8 . 0000000001111222223333333444444
23.00 8 . 55566667777888888888999
19.00 9 . 0011111122233333334
15.00 9 . 556677777888889
12.00 10 . 000011233444
4.00 10 . 5578
3.00 11 . 002
2.00 11 . 89
1.00 12 . 0
3.00 Extremes (>=124)
Between-Subjects Factors
Value Label N
3 Black, Not of
5
Hispanic Origin
4 Hispanic 4
5 White, not of
165
Hispanic origin
8 8 1
Descriptive Statistics
8 65.0000 . 1
8 88.0000 . 1
Box's M 11.299
F 1.467
df1 6
df2 594.321
Sig. .187
Multivariate Testsa
Intercept Pillai's Trace .066 5.984b 2.000 170.000 .003 .066 11.968 .876
Wilks' Lambda .934 5.984b 2.000 170.000 .003 .066 11.968 .876
Hotelling's Trace .070 5.984b 2.000 170.000 .003 .066 11.968 .876
Roy's Largest
.070 5.984b 2.000 170.000 .003 .066 11.968 .876
Root
age Pillai's Trace .035 3.068b 2.000 170.000 .049 .035 6.135 .586
Wilks' Lambda .965 3.068b 2.000 170.000 .049 .035 6.135 .586
Hotelling's Trace .036 3.068b 2.000 170.000 .049 .035 6.135 .586
Roy's Largest
.036 3.068b 2.000 170.000 .049 .035 6.135 .586
Root
race Pillai's Trace .025 .541 8.000 342.000 .826 .012 4.326 .251
Wilks' Lambda .975 .539b 8.000 340.000 .827 .013 4.313 .250
Hotelling's Trace .025 .537 8.000 338.000 .828 .013 4.299 .249
Roy's Largest
.021 .904c 4.000 171.000 .463 .021 3.616 .283
Root
Tests the null hypothesis that the error variance of the dependent variable is equal
across groups.
a. Design: Intercept + age + race
Type III
Sum of Mean Partial Eta Noncent. Observed
Source Dependent Variable Squares df Square F Sig. Squared Parameter Powerc
The Kolmogorov-Smirnov indicates all the dependent variables sig. values are greater than alpha
level = .05. Therefore, we fail to reject null hypothesis and accept null hypothesis. Thus the data
does not violate the normality assumption. That means the data is normally distributed.
Alternative hypothesis:
There is significant difference in the mean score for pre-test and post-test.
Assessing Normality
teachinggroup
Cases
Descriptives
Median 55.5000
Variance 36.614
Minimum 45.00
Maximum 62.50
Range 17.50
Median 47.0000
Variance 79.789
Minimum 35.00
Maximum 62.50
Range 27.50
Extreme Values
2 17 62.50
3 14 60.00
4 16 58.00
5 20 56.00
Lowest 1 11 45.00
2 13 46.50
3 12 52.00
4 19 54.00
5 18 55.00
2 4 59.50
3 2 51.00
4 7 50.00
5 8 48.00
Lowest 1 5 35.00
2 10 37.50
3 9 40.00
4 6 43.50
5 3 46.00
a. Only a partial list of cases with the value 65.00 are shown in the table of
upper extremes.
b. Only a partial list of cases with the value 65.00 are shown in the table of
lower extremes.
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
post
Histograms
Stem-and-Leaf Plots
.00 4 .
2.00 4 . 56
2.00 5 . 24
3.00 5 . 568
3.00 6 . 022
2.00 3 . 57
4.00 4 . 0368
3.00 5 . 019
1.00 6 . 2
Shapiro-Wilk test
Null hypothesis: There is no significant effect of control group and treatment group on anxiety level.
Alternative hypothesis: There is significant effect of control group and treatment group on anxiety level.
Univariate Analysis of Variance
Between-Subjects Factors
Value Label N
1.00 new tm 10
Descriptive Statistics
Dependent Variable: post
6.691 1 18 .019
Tests the null hypothesis that the error variance of the
dependent variable is equal across groups.
a. Design: Intercept + Group + pre + Group * pre
teachinggroup
Dependent Variable: post
a. Covariates appearing in the model are evaluated at the following values: pre =
63.6750.
Pairwise Comparisons
Dependent Variable: post
(I) teachinggroup (J) teachinggroup (I-J) Std. Error Sig.b Lower Bound Upper Bound
Univariate Tests
Dependent Variable: post
Partial Eta
Sum of Squares df Mean Square F Sig. Squared
The F tests the effect of teachinggroup. This test is based on the linearly independent pairwise
comparisons among the estimated marginal means.
Examination of the Shapiro-Wilk statistics and histograms for each group indicated that the
ANCOVA assumption of normality was supported. Scatterplots indicated that the relationship
between the covariate (pre-test) and dependent variable (post-test) was linear. The assumptions
of homogeneity of variances were supported by the absence of the significant IV-by-covariate
interaction, F (1, 16) = .130, p = .723. However the Levenes test, F (1, 18) = 6.691, p = .02.
Therefore a more stringent alpha be used (.01) to ensure the variances are equal and violated the
assumption.
There was a significant difference between the control group and treatment group on post test on
the anxiety level, F (1, 16) = 12.214, p = .003, partial eta squared = .433. The pairwise
comparisons were significant between control group (oldtm) and treatment group (newtm), p <
.01. The treatment group (M = 46.675, SD = 1.822) is lower than control group (M = 55.672, SD
= 1.819) after receive treatment, which used new teaching method, reduced their anxiety level.
Null hypothesis:
Ho1: There is no significant difference for interaction between the types of program conducted and
also across time.
Ho2: There is no significant difference change in participants confidence level across the three time
periods.
Ho3: There is no significant difference in the confidence level for participants in the maths skills group
and the confidence building group.
Alternative hypothesis:
Ha1: There is significant difference for interaction between the types of program conducted and
also across time.
Ha2: There is significant difference change in participants confidence level across the three time
periods.
Ha3: There is significant difference in the confidence level for participants in the maths skills group
and the confidence building group.
Assessing Normality
Cases
Descriptives
Median 20.00
Variance 28.828
Minimum 11
Maximum 29
Range 18
Interquartile Range 8
Median 26.00
Variance 27.068
Minimum 14
Maximum 34
Range 20
Interquartile Range 10
Tests of Normality
Kolmogorov-Smirnova Shapiro-Wilk
confidence time1
confidence time2
confidence time3
Decision matrix
Kolmogorov-Smirnov test
Statistics Alpha value Sig. value Conclusion
Ho: Population distribution is normal.
Null hypothesis:
There is no significant difference in the confidence level for participants in the maths skills group and the
confidence building group.
Alternative hypothesis:
There is significant difference in the confidence level for participants in the maths skills group and the
confidence building group.
Within-Subjects Factors
Measure: MEASURE_1
Dependent
factor3 Variable
1 confid1
2 confid2
3 confid3
Between-Subjects Factors
Value Label N
2 confidence
15
building
Descriptive Statistics
Box's M 8.522
F 1.254
df1 6
df2 5680.302
Sig. .275
Multivariate Testsa
Partial Eta
Effect Value F Hypothesis df Error df Sig. Squared
Epsilonb
Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is proportional to
an identity matrix.
a. Design: Intercept + group
Within Subjects Design: factor3
b. May be used to adjust the degrees of freedom for the averaged tests of significance. Corrected tests are displayed in the Tests of
Within-Subjects Effects table.
Tests the null hypothesis that the error variance of the dependent variable is
equal across groups.
a. Design: Intercept + group
Within Subjects Design: factor3
Profile Plots
Decision matrix
Alternative hypothesis:
Ha1:
There is significant difference for
interaction between the types of program
conducted and also across time.
Alternative hypothesis:
Ha2:
There is significant difference change in
participants confidence level across the
three time periods.
Alternative hypothesis:
Ha3:
There is significant difference in the
confidence level for participants in the
maths skills group and the confidence
building group.
According to the Boxs Test of equality if a covariance matrix, the significant value is .275,
greater than .001, we fail to reject the null hypotheses, and we have not violated the assumption
of covariance matrices.
The Levenes Test of Equality of Error Variances indicates all the significant values were larger
than alpha value (.05), therefore fail to reject null hypothesis, which there is homogeneity of
variances and assumption of homogeneity of variances s not violated.
A mixed between-within subject analysis of variance was conducted to assess the impact of two
different approaches (Math skills, confidence building) on participants scores of confidents
level, across three time periods (pre-intervention, post intervention and follow up). There was
significant interaction between the types of program conducted and also across time, Wilks
lambda = . 793, F (2, 27) = 3.53, p = .043. The partial-eta squared = .207.
There was a substantial main effect of time, Wilks lambda = .248, F 92, 27) = 40.897, p = .000,
partial eta squared = .752, with both groups showing a increasing in level of confidence across
the three time periods. However the main effect comparing the two types of intervention was not
significant, F (1, 28) = 1.255, p = .272, partial eta squared = .043. This indicates no difference in
the effectiveness of the two different approaches to increase confident level among students.
There was significant interaction between the types of program conducted and also across time,
the effect size is large effect, the partial-eta squared = .207, which represents 20.7%. There was a
substantial main effect of time, the effect size is 75.2 (partial-eta squared = .752). The effect size
of the between-subject effect is partial eta squared = .043, which represents 4.3%, is small effect
size. Therefore it did not reach statistical significance.
6. Using the SURVEY. SAV file, select a set of variables (YOUR CHOICE!!) suitable to conduct a one-
way MANOVA.
Null hypothesis:
Ho1: There is no significant between with child group and without child group in terms of negative
affect.
Ho2: There is no significant between with child group and without child group in terms of positive
affect.
Ho3: There is no significant between with child group and without child group in terms of perceived
stress.
Alternative hypothesis:
Ha1: There is significant between with child group and without child group in terms of negative
affect.
Ha2: There is significant between with child group and without child group in terms of positive
affect.
Ha3: There is significant between with child group and without child group in terms of perceived
stress.
b. Make your decision and conclusion and justify your answer.
Residuals Statisticsa
Between-Subjects Factors
Value Label N
2 NO 249
Descriptive Statistics
Box's M 8.391
F 1.387
df1 6
df2 1047961.923
Sig. .215
Multivariate Testsa
Partial Eta
Effect Value F Hypothesis df Error df Sig. Squared
Corrected Model Total positive affect 298.730a 1 298.730 5.748 .017 .013
b. Exact statistic Total negative affect 91.275b 1 91.275 1.820 .178 .004
Tests the null hypothesis that the error variance of the dependent variable is equal
across groups.
a. Design: Intercept + child
child
95% Confidence Interval
Dependent Variable child Mean Std. Error Lower Bound Upper Bound
There was a statistically significant difference between with child group and without child group
on the combined variables, F (3, 427) = 2.089, p = .101. Wilks Lambda = .986; partial eta
squared = .014. Results analyzed separately indicated that there were not significant differences
between with child group and without child group to total negative affect and to total perceived
stress. However, there was a significant difference between with child group and without child
group to total positive effect, F (1, 429) = 5.748, p = .017, partial eta squared = .013. An
inspection of the mean scores indicated that the with child group have higher total positive affect
(M = 36.64, SD = 7.713) than without child group (M = 32.95, SD = 7.236).