You are on page 1of 6

Page 1 of 6

Repeated Measures Anova

A repeated measures One-way ANOVA was conducted on 15 participants to measure if mental health is
improved under three different conditions, at the start of gardening class, two weeks of gardening, four weeks of
gardening, and two weeks after gardening class.

Hypothesis

H0: There is no significant difference between the four conditions, μ1 = μ2 = μ3 = μ4.


H1: At least one condition is different, ≠μ1 ≠μ2 ≠ μ3 ≠ μ4.

Descriptives:

Figure 1

Descriptive Statistics
Std.
Mean Deviation N
Time 1: BDI score at 8.8667 2.26358 15
start of gardening class
Time 2: BDI score after 7.3333 2.79455 15
2 weeks of gardening
Time 3: BDI score after 5.7333 3.08143 15
4 weeks of gardening
Time 4: BDI score 2 5.6000 4.53242 15
weeks after gardening
class

Mental health was improved in ‘at start of gardening’ condition than the other two conditions
It would appear that the mean score is highest at the start of the class, followed by

Requirements:

Requirement # 1: One categorical (nominal or ordinal) independent variable (three or


more distinct conditions measured across the same group)
Page 2 of 6

Requirement # 2: One scale dependent variable (Should be interval or ratio)

Requirement # 3: The research design is within-group study.


There are four different conditions, the requirement is met.

Assumptions

Assumption # 1: dependent data must be parametric.


The scores for the 4 conditions are normally distributed, therefore, we meet this assumption.
Figure 2

Tests of Normality

Kolmogorov-Smirnova Shapiro-Wilk
Statistic df Sig. Statistic df Sig.
*
Time 1: BDI score at .175 15 .200 .946 15 .463
start of gardening class
Time 2: BDI score after .198 15 .117 .920 15 .192
2 weeks of gardening
Time 3: BDI score after .199 15 .114 .936 15 .336
4 weeks of gardening
Time 4: BDI score 2 .132 15 .200* .935 15 .327
weeks after gardening
class
*. This is a lower bound of the true significance.
a. Lilliefors Significance Correction

H0= Data is not different from the normal distribution.


H1= Data is significantly different from the normal distribution.

A Shapiro-Walk was conducted to assess the normality and it indicated that the data are normally distributed at
each condition: score at start of gardening class W(15), =.946, P=.463; score after 2 weeks of gardening W(15), N=.920,
P=.192; score after 4 weeks of gardening; W(15), .936, P=.336; score 2 weeks after gardening class W(15), .935,
P=.327. We accept the null hypothesis.
Page 3 of 6

Assumption # 2: The independent variable must be categorical.


There are four different conditions thus the assumption is met.

Assumption # 3: Homogeneity of variance

Figure 3

Mauchly's Test of Sphericitya


Measure: MEASURE_1
Epsilonb
Within Subjects Mauchly's Approx. Chi- Greenhouse-
Effect W Square Df Sig. Geisser Huynh-Feldt Lower-bound
Crochet .542 7.797 5 .169 .695 .818 .333
Tests the null hypothesis that the error covariance matrix of the orthonormalized transformed dependent variables is
proportional to an identity matrix.
a. Design: Intercept
Within Subjects Design: crochet
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.

H0=Variances are equal.


H1=Variances are not equal.

Mauchly’s test of sphericity was conducted, and we can be confident that the sphericity can be assumed as there
was a nonsignificant outcome .169>.05 Thus, we fail to reject the null hypothesis. The assumption is met.

Assumption 4: Groups must be independent.


Page 4 of 6

Running the analysis


ANOVA:
Figure 4
Tests of Within-Subjects Effects
Measure: MEASURE_1
Type III Sum Partial Eta
Source of Squares Df Mean Square F Sig. Squared
Crochet Sphericity Assumed 106.583 3 35.528 6.305 .001 .311
Greenhouse-Geisser 106.583 2.084 51.145 6.305 .005 .311
Huynh-Feldt 106.583 2.455 43.407 6.305 .003 .311
Lower-bound 106.583 1.000 106.583 6.305 .025 .311
Error(crochet) Sphericity Assumed 236.667 42 5.635
Greenhouse-Geisser 236.667 29.175 8.112
Huynh-Feldt 236.667 34.376 6.885
Lower-bound 236.667 14.000 16.905

H0= There are no significant differences between the condition’s means.


H1: At least one is different.

A repeated measures ANOVA showed a significant difference in mental health across the four conditions,
F(3,42)=6.305, P=.001.We reject the null hypothesis.

Tests of Within-Subjects Contrasts


Measure: MEASURE_1
Type III Sum Partial Eta
Source crochet of Squares df Mean Square F Sig. Squared
Crochet Linear 97.470 1 97.470 9.551 .008 .406
Quadratic 7.350 1 7.350 2.042 .175 .127
Cubic 1.763 1 1.763 .569 .463 .039
Error(crochet Linear 142.880 14 10.206
Page 5 of 6

) Quadratic 50.400 14 3.600


Cubic 43.387 14 3.099

A significant linear trend suggests that the scores change in a straight line. Mental health has improved
throughout the sessions, and depression decreased.

Post-Hoc

Pairwise Comparisons
Measure: MEASURE_1
Mean 95% Confidence Interval for
(I) Difference (I- Differenceb
crochet (J) crochet J) Std. Error Sig.b Lower Bound Upper Bound
1 2 1.533 .689 .258 -.582 3.648
*
3 3.133 .723 .004 .915 5.352
4 3.267 1.157 .081 -.283 6.816
2 1 -1.533 .689 .258 -3.648 .582
3 1.600 .660 .177 -.427 3.627
4 1.733 1.016 .661 -1.386 4.852
*
3 1 -3.133 .723 .004 -5.352 -.915
2 -1.600 .660 .177 -3.627 .427
4 .133 .839 1.000 -2.441 2.707
4 1 -3.267 1.157 .081 -6.816 .283
2 -1.733 1.016 .661 -4.852 1.386
3 -.133 .839 1.000 -2.707 2.441
Based on estimated marginal means
*. The mean difference is significant at the .05 level.
b. Adjustment for multiple comparisons: Bonferroni.

A post-hoc Bonferroni test was conducted, and it indicated there is a significant difference between time 1 and
time 3 M=5.73 SD=3.08 (P=.004). But there is no significant difference between the other times.
Page 6 of 6

The graph shows that depression levels have decreased throughout the weeks, which means mental health has
improved.

Results
A repeated-measures one-way ANOVA indicated that there was a significant difference in improvement in mental health
across the four conditions. A post-Hoc Bonferroni test indicated that there was a significant difference between time 1
and time 3 (P=.004).

You might also like