SlideShare a Scribd company logo
1 of 23
Two-way repeated measures ANOVA with one-between and one-within subjects factor
SPSS demo
Mike Crowson, Ph.D.
March 2020
Link to video presentation: https://youtu.be/KZHHAuTj8GU
In a previous presentation (video: https://youtu.be/TH7FVKevAcQ, Powerpoint: https://drive.google.com/open?id=1aRgrc4-
nh3tie-pECjRnC0UkrtOmljpX), I discussed and demonstrated the use of repeated measures analysis of variance when you
measure an outcome variable repeatedly within a single group. It is oftentimes the case that repeated measures analysis is
applied to data involving repeated measurements when you are working with more than a single group. This approach, for
instance, may be used in cases where the researcher hypothesizes that the variation in means associated with repeated
measurements on the outcome varies across groups. This effectively translates into a hypothesis concerning an interaction
between the repeated factor and a grouping variable.
One might adopt this approach when testing whether differences in means observed over time are the same or different
across levels of a grouping variable. If differences are found, it is possible to examine whether the mean differences reflect
different trends over time. This can be particularly handy if one of the groups being compared is a control condition and
other conditions are experimental in nature.
This approach could also easily be used when testing whether individuals react the same or differently across levels of a
repeated factor (for example, different stimuli for which a person is exposed) and a grouping variable.
For the current example (the data below is partial), our between-subjects’ factor is “tx.group” coded 1=control,
2=treatment A, 3=treatment B. We are going to test whether there are significant mean differences in anxiety scores over
three measurement occasions, as well as whether there are group differences in terms of how the means vary over time.
The full dataset can be
downloaded here:
https://drive.google.com/op
en?id=1090t8zuEI8eSD8QW
vDXNWYMna3gE516H
Because we are introducing a between-groups
factor, we now include “Homogeneity tests”.
The selection of ‘Compare main effects’ and ‘Bonferroni’ as the confidence interval adjustment will yield
Bonferroni-adjusted pairwise comparisons across level levels of the repeated factor. [It will not yield
comparisons within groups, however. That would require simple effects tests, which I will address later.]
Plotting the mean anxiety scores by time and by time
and treatment group.
If we are interested in testing whether there are pairwise
difference in subjects’ average (i.e., averaged across time)
anxiety score, we can select Tukey’s post hoc tests.
Descriptive statistics for each group at each time point.
Here, we have the multivariate test
results for time (the within-subjects
factor) and the time X group
interaction.
Box’s test is a test of the assumption of equality variance-covariance matrices of difference scores
between groups (Weinfurt, 2000). This is an assumption of the multivariate tests below. If Box’s test is
significant, then you have evidence of a violation. [Note: Box’s test is sensitive to multivariate
nonnormality, thereby increasing rejection rate.]
Nevertheless, the multivariate test results are fairly robust when you have equal or nearly equal n’s in
your groups (e.g., largest n/smallest n < 1.5). If the ratio of the largest to smallest group size is large, then
the results of the multivariate tests can be biased. When the larger group is associated with smaller
variability, the multivariate test becomes too liberal. When the larger group is associated with the greater
variability, the test becomes too conservative (Pituch & Stevens, 2016).
The main effect of time is
statistically significant, Wilks’
lambda=.173, F(2,65)=155.687,
p<.001. This effect, however, is
qualified by a significant time X
group interaction, Wilks’ lambda =
.440, F(4,130)=16.480, p<.001.
A significant Box’s test result indicates a violation of homogeneity of covariance matrices,
whereas a non-significant result is consistent with the MANOVA assumption.
Here, we see that p<.001, where there is evidence of the violation. Nevertheless, the ratio of the
largest n to smallest n is 26/19 = 1.368 (which is less than the 1.5 threshold suggested by Pituch
& Stevens, 2016).
The interaction is indicating that the variation in the means on anxiety over the repeated measurement occasions itself
varies as a function of treatment group membership.
The sphericity assumption is required for all
univariate main effects tests and interaction tests
(O’Brien & Kaiser, 1985). Given Mauchly’s test is
impacted by non-normality and by sample size, it is
not highly recommended when evaluating whether
the sphericity condition has been met.
A Greenhouse-Geisser epsilon (ε) value < .75,
suggests using the Greenhouse-Geisser adjustment
with the univariate test of mean differences (see
table of “Tests of within-subjects effects”), whereas a
value falling between .75 and 1 suggests the use of
the Huynh-Feldt adjustment with the univariate
tests. [ε=1 is consistent with sphericity]. The
sphericity assumed test can be used if you determine
sphericity is not violated.
[FYI, the Lower-Bound test is generally overly
conservative and is not typically used]
All three test results yield the same conclusions
with respect to the main and interaction effects.
The main effect of time on anxiety scores is
statistically significant, sphericity assumed
F(2,132)=120.752, p<.001.
This effect was qualified by a significant time X
group interaction effect, sphericity assumed
F(4,132)=20.658, p<.001.
Although the test of the linear component of the trend is significant (p<.001), the higher-order quadratic component was also
significant [F(1,66)=19.373, p<.001]. This suggests that across groups, the mean level of anxiety exhibited a quadratic trend
over the three measurement occasions. This is further suggested by examining the profile plot of the means.
Assessment of trending over time (irrespective of group membership)
Testing for differential trending across groups
We see here that although the test of the interaction between the linear component of the trend and treatment group is
significant, the interaction between treatment group and the higher-order quadratic component was also significant
[F(2,66)=23.903, p<.001]. Moreover, looking at the profile plot of means, we see that the curvature of the lines is less
pronounced for the Control group and Treatment A. However, the line for Treatment B appears more substantially curved.
Since these trends are not parallel, it is no surprise the test of the time X tx.group interaction was significant.
Interpretation: The main effect of treatment group on
the average anxiety score across time is statistically
significant, F(2, 66)=28.949, p<.001.
The Levene’s test results involve tests of differences in variances at
each time point, an assumption of the univariate ANOVA (see Tests of
Between-subjects effects). It turns out that the standard Levene’s tests
(and robust tests, based on median, etc.) are significant for Time 1 and
Time 3. Nevertheless, a violation of this assumption is less of an issue
with roughly equivalent sample sizes (where largest n / smallest n <
1.5).
The Tests of Between-subjects Effects is a test of the main effect of the
grouping variable on scores on the repeated measure averaged over
time. The result presented here is simply a test of group differences on
the average of anxiety scores (i.e., those scores averaged over time for
each person).
Bonferroni-adjusted paired t-tests. Here we see all pairwise differences
on anxiety are statistically significant (p’s≤.016).
These are pairwise comparisons on the average
anxiety score (averaged over time) for each group.
All pairwise differences were significant (as all p’s were
≤ .04).
These are Bonferroni adjusted pairwise
comparisons.
These comparisons are based on Tukey’s
tests.
Group means on anxiety at each measurement
occasion.
Simple effects test as a follow-up to a significant interaction
If you find evidence of a significant interaction (as we found in our analysis) between the between-subjects and within-
subjects factors, you may wish to describe the nature of the interaction using simple-effects tests. An easy approach is to
click on Paste after specifying your options…
When you do this, a syntax editor will open up. You can obtain simple
effects tests with very minor modifications to the syntax.
By adding the Compare command
(along with Bonferroni adjustment),
your output (after highlighting
everything and pressing the green
button) will include…
Multivariate tests of mean differences in anxiety scores over time by group (see left).
You also obtain pairwise tests of mean differences in anxiety
between time points within each group (see table to the right).
References
Lomax, R.G., & Hahs-Vaughn, D.L. (2012). An introduction to statistical concepts (3rd ed). New York: Routledge.
O’Brien, R. G., & Kaiser, M. K. (1985). MANOVA method for analyzing repeated measures designs: An extensive
primer. Psychological Bulletin, 97, 316-333.
Pituch, K.A., & Stevens, J.P. (2016). Applied multivariate statistics for the social sciences (6th ed). New York:
Routledge.
Weinfurt, K.P. (2000). Repeated measures analysis: ANOVA, MANOVA, and HLM. In L.G. Grimm & P.R. Yarnold
(Eds.), Reading and understanding more multivariate statistics (pp. 317-361). Washington, DC: American
Psychological Association.

More Related Content

Similar to Repeated measures ANOVA, with one-within and one-between factors.pptx

Discussion Please discuss, elaborate and give example on the topi
Discussion Please discuss, elaborate and give example on the topiDiscussion Please discuss, elaborate and give example on the topi
Discussion Please discuss, elaborate and give example on the topi
widdowsonerica
 
Discussion Please discuss, elaborate and give example on the topi.docx
Discussion Please discuss, elaborate and give example on the topi.docxDiscussion Please discuss, elaborate and give example on the topi.docx
Discussion Please discuss, elaborate and give example on the topi.docx
duketjoy27252
 
Discussion Discuss, elaborate and give example on the topic below
Discussion Discuss, elaborate and give example on the topic belowDiscussion Discuss, elaborate and give example on the topic below
Discussion Discuss, elaborate and give example on the topic below
widdowsonerica
 
Assessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docxAssessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docx
festockton
 
Assessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docxAssessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docx
galerussel59292
 
Calculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docx
Calculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docxCalculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docx
Calculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docx
aman341480
 
Inferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing ResearchInferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing Research
LaticiaGrissomzz
 
Inferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing ResearchInferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing Research
LizbethQuinonez813
 
#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N
#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N
#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N
troutmanboris
 
Discussion Please use the Referencemodule I provided. Professor
Discussion Please use the Referencemodule I provided. Professor Discussion Please use the Referencemodule I provided. Professor
Discussion Please use the Referencemodule I provided. Professor
widdowsonerica
 

Similar to Repeated measures ANOVA, with one-within and one-between factors.pptx (20)

Statistical Tools
Statistical ToolsStatistical Tools
Statistical Tools
 
12-5-03.ppt
12-5-03.ppt12-5-03.ppt
12-5-03.ppt
 
12-5-03.ppt
12-5-03.ppt12-5-03.ppt
12-5-03.ppt
 
Factorial Design
Factorial DesignFactorial Design
Factorial Design
 
Discussion Please discuss, elaborate and give example on the topi
Discussion Please discuss, elaborate and give example on the topiDiscussion Please discuss, elaborate and give example on the topi
Discussion Please discuss, elaborate and give example on the topi
 
Discussion Please discuss, elaborate and give example on the topi.docx
Discussion Please discuss, elaborate and give example on the topi.docxDiscussion Please discuss, elaborate and give example on the topi.docx
Discussion Please discuss, elaborate and give example on the topi.docx
 
One way repeated measure anova
One way repeated measure anovaOne way repeated measure anova
One way repeated measure anova
 
Discussion Discuss, elaborate and give example on the topic below
Discussion Discuss, elaborate and give example on the topic belowDiscussion Discuss, elaborate and give example on the topic below
Discussion Discuss, elaborate and give example on the topic below
 
Commonly used Statistics in Medical Research Handout
Commonly used Statistics in Medical Research HandoutCommonly used Statistics in Medical Research Handout
Commonly used Statistics in Medical Research Handout
 
Assessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docxAssessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docx
 
Assessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docxAssessment 4 ContextRecall that null hypothesis tests are of.docx
Assessment 4 ContextRecall that null hypothesis tests are of.docx
 
Calculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docx
Calculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docxCalculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docx
Calculating Analysis of Variance (ANOVA) and Post Hoc Analyses Follo.docx
 
Lec1_Methods-for-Dummies-T-tests-anovas-and-regression.pptx
Lec1_Methods-for-Dummies-T-tests-anovas-and-regression.pptxLec1_Methods-for-Dummies-T-tests-anovas-and-regression.pptx
Lec1_Methods-for-Dummies-T-tests-anovas-and-regression.pptx
 
Analysis of Variance
Analysis of VarianceAnalysis of Variance
Analysis of Variance
 
Aca 22-407
Aca 22-407Aca 22-407
Aca 22-407
 
STATISTICS : Changing the way we do: Hypothesis testing, effect size, power, ...
STATISTICS : Changing the way we do: Hypothesis testing, effect size, power, ...STATISTICS : Changing the way we do: Hypothesis testing, effect size, power, ...
STATISTICS : Changing the way we do: Hypothesis testing, effect size, power, ...
 
Inferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing ResearchInferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing Research
 
Inferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing ResearchInferential AnalysisChapter 20NUR 6812Nursing Research
Inferential AnalysisChapter 20NUR 6812Nursing Research
 
#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N
#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N
#35921 Topic Discussion2Number of Pages 1 (Double Spaced)N
 
Discussion Please use the Referencemodule I provided. Professor
Discussion Please use the Referencemodule I provided. Professor Discussion Please use the Referencemodule I provided. Professor
Discussion Please use the Referencemodule I provided. Professor
 

Recently uploaded

Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan 081901222272 Obat Penggugur Kandu...
Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan  081901222272 Obat Penggugur Kandu...Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan  081901222272 Obat Penggugur Kandu...
Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan 081901222272 Obat Penggugur Kandu...
Halo Docter
 
Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...
Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...
Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...
Cara Menggugurkan Kandungan 087776558899
 
Physiologic Anatomy of Heart_AntiCopy.pdf
Physiologic Anatomy of Heart_AntiCopy.pdfPhysiologic Anatomy of Heart_AntiCopy.pdf
Physiologic Anatomy of Heart_AntiCopy.pdf
MedicoseAcademics
 
Difference Between Skeletal Smooth and Cardiac Muscles
Difference Between Skeletal Smooth and Cardiac MusclesDifference Between Skeletal Smooth and Cardiac Muscles
Difference Between Skeletal Smooth and Cardiac Muscles
MedicoseAcademics
 

Recently uploaded (20)

Top 10 Most Beautiful Russian Pornstars List 2024
Top 10 Most Beautiful Russian Pornstars List 2024Top 10 Most Beautiful Russian Pornstars List 2024
Top 10 Most Beautiful Russian Pornstars List 2024
 
Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan 081901222272 Obat Penggugur Kandu...
Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan  081901222272 Obat Penggugur Kandu...Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan  081901222272 Obat Penggugur Kandu...
Obat Aborsi Ampuh Usia 1,2,3,4,5,6,7 Bulan 081901222272 Obat Penggugur Kandu...
 
Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...
Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...
Cara Menggugurkan Kandungan Dengan Cepat Selesai Dalam 24 Jam Secara Alami Bu...
 
TEST BANK For Guyton and Hall Textbook of Medical Physiology, 14th Edition by...
TEST BANK For Guyton and Hall Textbook of Medical Physiology, 14th Edition by...TEST BANK For Guyton and Hall Textbook of Medical Physiology, 14th Edition by...
TEST BANK For Guyton and Hall Textbook of Medical Physiology, 14th Edition by...
 
VIP ℂall Girls Thane West Mumbai 9930245274 WhatsApp: Me All Time Serviℂe Ava...
VIP ℂall Girls Thane West Mumbai 9930245274 WhatsApp: Me All Time Serviℂe Ava...VIP ℂall Girls Thane West Mumbai 9930245274 WhatsApp: Me All Time Serviℂe Ava...
VIP ℂall Girls Thane West Mumbai 9930245274 WhatsApp: Me All Time Serviℂe Ava...
 
7 steps How to prevent Thalassemia : Dr Sharda Jain & Vandana Gupta
7 steps How to prevent Thalassemia : Dr Sharda Jain & Vandana Gupta7 steps How to prevent Thalassemia : Dr Sharda Jain & Vandana Gupta
7 steps How to prevent Thalassemia : Dr Sharda Jain & Vandana Gupta
 
Physiologic Anatomy of Heart_AntiCopy.pdf
Physiologic Anatomy of Heart_AntiCopy.pdfPhysiologic Anatomy of Heart_AntiCopy.pdf
Physiologic Anatomy of Heart_AntiCopy.pdf
 
Top 10 Most Beautiful Chinese Pornstars List 2024
Top 10 Most Beautiful Chinese Pornstars List 2024Top 10 Most Beautiful Chinese Pornstars List 2024
Top 10 Most Beautiful Chinese Pornstars List 2024
 
Creeping Stroke - Venous thrombosis presenting with pc-stroke.pptx
Creeping Stroke - Venous thrombosis presenting with pc-stroke.pptxCreeping Stroke - Venous thrombosis presenting with pc-stroke.pptx
Creeping Stroke - Venous thrombosis presenting with pc-stroke.pptx
 
TEST BANK For Porth's Essentials of Pathophysiology, 5th Edition by Tommie L ...
TEST BANK For Porth's Essentials of Pathophysiology, 5th Edition by Tommie L ...TEST BANK For Porth's Essentials of Pathophysiology, 5th Edition by Tommie L ...
TEST BANK For Porth's Essentials of Pathophysiology, 5th Edition by Tommie L ...
 
Part I - Anticipatory Grief: Experiencing grief before the loss has happened
Part I - Anticipatory Grief: Experiencing grief before the loss has happenedPart I - Anticipatory Grief: Experiencing grief before the loss has happened
Part I - Anticipatory Grief: Experiencing grief before the loss has happened
 
Physicochemical properties (descriptors) in QSAR.pdf
Physicochemical properties (descriptors) in QSAR.pdfPhysicochemical properties (descriptors) in QSAR.pdf
Physicochemical properties (descriptors) in QSAR.pdf
 
ABO Blood grouping in-compatibility in pregnancy
ABO Blood grouping in-compatibility in pregnancyABO Blood grouping in-compatibility in pregnancy
ABO Blood grouping in-compatibility in pregnancy
 
Difference Between Skeletal Smooth and Cardiac Muscles
Difference Between Skeletal Smooth and Cardiac MusclesDifference Between Skeletal Smooth and Cardiac Muscles
Difference Between Skeletal Smooth and Cardiac Muscles
 
Intro to disinformation and public health
Intro to disinformation and public healthIntro to disinformation and public health
Intro to disinformation and public health
 
Face and Muscles of facial expression.pptx
Face and Muscles of facial expression.pptxFace and Muscles of facial expression.pptx
Face and Muscles of facial expression.pptx
 
Test bank for critical care nursing a holistic approach 11th edition morton f...
Test bank for critical care nursing a holistic approach 11th edition morton f...Test bank for critical care nursing a holistic approach 11th edition morton f...
Test bank for critical care nursing a holistic approach 11th edition morton f...
 
spinal cord disorders and paraplegia .
spinal cord disorders  and  paraplegia .spinal cord disorders  and  paraplegia .
spinal cord disorders and paraplegia .
 
SEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdf
SEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdfSEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdf
SEMESTER-V CHILD HEALTH NURSING-UNIT-1-INTRODUCTION.pdf
 
VIP ℂall Girls Arekere Bangalore 6378878445 WhatsApp: Me All Time Serviℂe Ava...
VIP ℂall Girls Arekere Bangalore 6378878445 WhatsApp: Me All Time Serviℂe Ava...VIP ℂall Girls Arekere Bangalore 6378878445 WhatsApp: Me All Time Serviℂe Ava...
VIP ℂall Girls Arekere Bangalore 6378878445 WhatsApp: Me All Time Serviℂe Ava...
 

Repeated measures ANOVA, with one-within and one-between factors.pptx

  • 1. Two-way repeated measures ANOVA with one-between and one-within subjects factor SPSS demo Mike Crowson, Ph.D. March 2020 Link to video presentation: https://youtu.be/KZHHAuTj8GU
  • 2. In a previous presentation (video: https://youtu.be/TH7FVKevAcQ, Powerpoint: https://drive.google.com/open?id=1aRgrc4- nh3tie-pECjRnC0UkrtOmljpX), I discussed and demonstrated the use of repeated measures analysis of variance when you measure an outcome variable repeatedly within a single group. It is oftentimes the case that repeated measures analysis is applied to data involving repeated measurements when you are working with more than a single group. This approach, for instance, may be used in cases where the researcher hypothesizes that the variation in means associated with repeated measurements on the outcome varies across groups. This effectively translates into a hypothesis concerning an interaction between the repeated factor and a grouping variable. One might adopt this approach when testing whether differences in means observed over time are the same or different across levels of a grouping variable. If differences are found, it is possible to examine whether the mean differences reflect different trends over time. This can be particularly handy if one of the groups being compared is a control condition and other conditions are experimental in nature. This approach could also easily be used when testing whether individuals react the same or differently across levels of a repeated factor (for example, different stimuli for which a person is exposed) and a grouping variable.
  • 3. For the current example (the data below is partial), our between-subjects’ factor is “tx.group” coded 1=control, 2=treatment A, 3=treatment B. We are going to test whether there are significant mean differences in anxiety scores over three measurement occasions, as well as whether there are group differences in terms of how the means vary over time. The full dataset can be downloaded here: https://drive.google.com/op en?id=1090t8zuEI8eSD8QW vDXNWYMna3gE516H
  • 4. Because we are introducing a between-groups factor, we now include “Homogeneity tests”.
  • 5. The selection of ‘Compare main effects’ and ‘Bonferroni’ as the confidence interval adjustment will yield Bonferroni-adjusted pairwise comparisons across level levels of the repeated factor. [It will not yield comparisons within groups, however. That would require simple effects tests, which I will address later.]
  • 6. Plotting the mean anxiety scores by time and by time and treatment group.
  • 7. If we are interested in testing whether there are pairwise difference in subjects’ average (i.e., averaged across time) anxiety score, we can select Tukey’s post hoc tests.
  • 8. Descriptive statistics for each group at each time point.
  • 9. Here, we have the multivariate test results for time (the within-subjects factor) and the time X group interaction. Box’s test is a test of the assumption of equality variance-covariance matrices of difference scores between groups (Weinfurt, 2000). This is an assumption of the multivariate tests below. If Box’s test is significant, then you have evidence of a violation. [Note: Box’s test is sensitive to multivariate nonnormality, thereby increasing rejection rate.] Nevertheless, the multivariate test results are fairly robust when you have equal or nearly equal n’s in your groups (e.g., largest n/smallest n < 1.5). If the ratio of the largest to smallest group size is large, then the results of the multivariate tests can be biased. When the larger group is associated with smaller variability, the multivariate test becomes too liberal. When the larger group is associated with the greater variability, the test becomes too conservative (Pituch & Stevens, 2016).
  • 10. The main effect of time is statistically significant, Wilks’ lambda=.173, F(2,65)=155.687, p<.001. This effect, however, is qualified by a significant time X group interaction, Wilks’ lambda = .440, F(4,130)=16.480, p<.001. A significant Box’s test result indicates a violation of homogeneity of covariance matrices, whereas a non-significant result is consistent with the MANOVA assumption. Here, we see that p<.001, where there is evidence of the violation. Nevertheless, the ratio of the largest n to smallest n is 26/19 = 1.368 (which is less than the 1.5 threshold suggested by Pituch & Stevens, 2016). The interaction is indicating that the variation in the means on anxiety over the repeated measurement occasions itself varies as a function of treatment group membership.
  • 11. The sphericity assumption is required for all univariate main effects tests and interaction tests (O’Brien & Kaiser, 1985). Given Mauchly’s test is impacted by non-normality and by sample size, it is not highly recommended when evaluating whether the sphericity condition has been met. A Greenhouse-Geisser epsilon (ε) value < .75, suggests using the Greenhouse-Geisser adjustment with the univariate test of mean differences (see table of “Tests of within-subjects effects”), whereas a value falling between .75 and 1 suggests the use of the Huynh-Feldt adjustment with the univariate tests. [ε=1 is consistent with sphericity]. The sphericity assumed test can be used if you determine sphericity is not violated. [FYI, the Lower-Bound test is generally overly conservative and is not typically used]
  • 12. All three test results yield the same conclusions with respect to the main and interaction effects. The main effect of time on anxiety scores is statistically significant, sphericity assumed F(2,132)=120.752, p<.001. This effect was qualified by a significant time X group interaction effect, sphericity assumed F(4,132)=20.658, p<.001.
  • 13. Although the test of the linear component of the trend is significant (p<.001), the higher-order quadratic component was also significant [F(1,66)=19.373, p<.001]. This suggests that across groups, the mean level of anxiety exhibited a quadratic trend over the three measurement occasions. This is further suggested by examining the profile plot of the means. Assessment of trending over time (irrespective of group membership)
  • 14. Testing for differential trending across groups We see here that although the test of the interaction between the linear component of the trend and treatment group is significant, the interaction between treatment group and the higher-order quadratic component was also significant [F(2,66)=23.903, p<.001]. Moreover, looking at the profile plot of means, we see that the curvature of the lines is less pronounced for the Control group and Treatment A. However, the line for Treatment B appears more substantially curved. Since these trends are not parallel, it is no surprise the test of the time X tx.group interaction was significant.
  • 15. Interpretation: The main effect of treatment group on the average anxiety score across time is statistically significant, F(2, 66)=28.949, p<.001. The Levene’s test results involve tests of differences in variances at each time point, an assumption of the univariate ANOVA (see Tests of Between-subjects effects). It turns out that the standard Levene’s tests (and robust tests, based on median, etc.) are significant for Time 1 and Time 3. Nevertheless, a violation of this assumption is less of an issue with roughly equivalent sample sizes (where largest n / smallest n < 1.5). The Tests of Between-subjects Effects is a test of the main effect of the grouping variable on scores on the repeated measure averaged over time. The result presented here is simply a test of group differences on the average of anxiety scores (i.e., those scores averaged over time for each person).
  • 16. Bonferroni-adjusted paired t-tests. Here we see all pairwise differences on anxiety are statistically significant (p’s≤.016).
  • 17. These are pairwise comparisons on the average anxiety score (averaged over time) for each group. All pairwise differences were significant (as all p’s were ≤ .04). These are Bonferroni adjusted pairwise comparisons.
  • 18. These comparisons are based on Tukey’s tests.
  • 19. Group means on anxiety at each measurement occasion.
  • 20. Simple effects test as a follow-up to a significant interaction If you find evidence of a significant interaction (as we found in our analysis) between the between-subjects and within- subjects factors, you may wish to describe the nature of the interaction using simple-effects tests. An easy approach is to click on Paste after specifying your options… When you do this, a syntax editor will open up. You can obtain simple effects tests with very minor modifications to the syntax.
  • 21. By adding the Compare command (along with Bonferroni adjustment), your output (after highlighting everything and pressing the green button) will include…
  • 22. Multivariate tests of mean differences in anxiety scores over time by group (see left). You also obtain pairwise tests of mean differences in anxiety between time points within each group (see table to the right).
  • 23. References Lomax, R.G., & Hahs-Vaughn, D.L. (2012). An introduction to statistical concepts (3rd ed). New York: Routledge. O’Brien, R. G., & Kaiser, M. K. (1985). MANOVA method for analyzing repeated measures designs: An extensive primer. Psychological Bulletin, 97, 316-333. Pituch, K.A., & Stevens, J.P. (2016). Applied multivariate statistics for the social sciences (6th ed). New York: Routledge. Weinfurt, K.P. (2000). Repeated measures analysis: ANOVA, MANOVA, and HLM. In L.G. Grimm & P.R. Yarnold (Eds.), Reading and understanding more multivariate statistics (pp. 317-361). Washington, DC: American Psychological Association.