SlideShare a Scribd company logo
1 of 30
Custom mediation models using Process macro with SPSS: An introduction
Mike Crowson, Ph.D.
August 2023
Video link: https://youtu.be/A4Sg5FjvpOA
A couple of suggestions:
1. Draw out your conceptual model. It may also help you to draw out the components of the statistical
model that are used to estimate your model parameters.
2. My recommendation is as part of your reporting ‘fill in’ the regression parameters (either unstandardized
or standardized) into the conceptual model diagram. This is along the lines of what you would see with a
path diagram when using a structural equation modeling program. This makes it easier to communicate
your results to your reader.
For more details on custom Process models, see Appendix B in Hayes (2022) book, Introduction to mediation,
moderation, and conditional process analysis (3rd Ed.).
The examples provided in this presentation are based on data from the article below. Nevertheless, the
examples are not intended as replications of what the authors did in their analyses.
Keshtidar M., Behzadnia, B. (2017). Prediction of intention to continue sport in athlete students: A self-determination
theory approach. PLoS ONE, 12(2): e0171673. doi:10.1371/journal.
pone.0171673
Link to SPSS data: https://drive.google.com/file/d/1x8IC8Iup1RQqh-nnNj2v5VydUN2eq0Br/view
Conceptual model
Task
orientation
Autonomous
motivation
Controlled
motivation
Intention to
continue
Process syntax to generate results for this model.
y=consequent variable
x=antecedent variable
m=mediator(s)
conf=confidence interval (here set at 95%)
stand=1 (requests fully standardized regression coefficients)
seed=12345 (This use important for generate the same bootstrap confidence intervals upon re-analysis of the same data with
same model specifications; the seed # you use is optional; I simply used 12345)
model=4 (model number of our mediation analysis)
The specification using syntax is the same as this via the dropdown menu
One downside of the dropdown menu is you cannot set a seed value. Thus, if you run your analysis more than once, the
lower and upper bounds of the bootstrap confidence intervals will vary somewhat over analyses. The syntax option allows
you to use a constant seed number across analyses that will result in the same lower and upper bound values.
task auton
1.5144
1.5144
task control
.2166
1.5144
.2166
The estimates I focus on throughout this Powerpoint will mainly be unstandardized regression coefficients from the
component regression models. These are what I am showing in the diagrams above (and throughout the Powerpoint).
However, you can easily use the standardized coefficients when making your own diagrams (if you would rather focus
on those).
1.5144
.2166
task
intention
auton
control
.0982
.0982
.1289
.1289
.0143
.0143
Unstandardized indirect effects of task orientation (X) on intention to
continue (Y) via each proposed mediator. These are computed using the
unstandardized regression coefficients from the previous models. The specific
indirect effect of autonomous motivation is .1952. The specific indirect effect
of controlled motivation is .0031. The total indirect (see TOTAL) is computed
as the sum of the two specific indirect effects: .1952 + .0031 = .1983
These effects are tested using the bootstrap 95% confidence intervals. If 0
(the null) falls between the lower and upper bound of an interval, that
particular indirect effect is non-significant. If 0 falls outside the lower and
upper bound for an interval, then the indirect effect is considered significant.
We see here that the specific indirect effect of task orientation on intention to
continue via autonomous motivation (IE=.1952) is significant [bootstrap 95%
CI = (.1144,.2928)], whereas the specific indirect effect (IE=.0031) of task
orientation via controlled motivation is not significant [bootstrap 95% CI = (-
.0104,.0237)]. The total indirect effect (IE=.1983) on intention to continue via
the two mediators is significant [bootstrap 95% CI = (.1140,.3002).
Completely standardized specific indirect effects of task
orientation (X) on intention to continue (Y) via each proposed
mediator. These are computed using the standardized
regression coefficients from the previous models.
To be able to generate custom mediation models, we must create a bmatrix. This essentially is a matrix that specifies which
regression parameters in your model are estimated and which are fixed to 0. My recommendation is to begin this process
by (1) drawing out your model (if you have not already done so) and then (2) creating a k X k matrix (where k=number of
variables in your analysis. Each column is headed by a variable that potentially points away from one variable in the model
to another. Think of these variables as ‘upstream’. Each row represents a variable ‘downstream’ being pointed to by an
upstream variable. Since a variable cannot predict itself, I have indicated that with a dash (-) in the matrix below for cells
M1-M1 and M2-M2. Since cell M1-M2 (row 2, col 1) is open for specification, we are not allowed to leave cell M2-M1 (row
1, col 2) open for specification.
- -
-
X
M1
M2
M1
M2
Y
Upstream variables with arrows pointing FROM them
Downstream variables
with arrows pointing
TO them
Next, we designate parameters that are freely estimated with a 1 and parameters that are not estimated with a 0.
1 - -
1 0 -
1 1 1
X
M1
M2
M1
M2
Y
In the Process syntax, the bmatrix option will collect the 1’s and 0’s row by row in this matrix:
bmatrix=1,1,0,1,1,1
The first 1 is in row 1. The 1 followed by 0 is in row 2. The 1, 1, and 1 are found in row 3.
Upstream variables with arrows pointing FROM them
Downstream variables
with arrows pointing
TO them
In the Process syntax, the bmatrix option will collect the 1’s and 0’s row by row in this matrix:
bmatrix=1,1,0,1,1,1
The first 1 is in row 1. The 1 followed by 0 is in row 2. The 1, 1, and 1 are found in row 3.
1 - -
1 0 -
1 1 1
task
autonomous
controlled
intention
controlled
autonomous
Same as previous…
The syntax in Process…
To test this model, we need to remove the direct effect of task orientation (X) to intention to continue (Y).
1 - -
1 0 -
0 1 1
task
autonomous
controlled
intention
controlled
autonomous
Going back to our matrix, we set the parameter in row 3, column 1 to 0 (bolded below).
Process syntax…
Same results as before
intention
auton
control
.1402
.0138
1.5144
.2166
.1402
.0138
In this model, the direct effects of autonomous motivation and controlled motivation are not controlling for task
orientation (as had occurred in our previous analysis).
This model includes serial mediation, where one mediator has a direct effect on another mediator. To test this model,
we need to remove the direct effect of task orientation (X) to intention to continue (Y) – as before – and add the direct
effect of autonomous motivation on controlled motivation.
1 - -
1 1 -
0 1 1
task
autonomous
controlled
intention
controlled
autonomous
Going back to our matrix, we (again) set the parameter in row 3, column 1 to 0 (bolded below). Now, however, we
change the former 0 (in row 2, column 2) to 1 (bolded below).
Process syntax…
Important: If you have mediators that follow a particular sequence (ordering), you need to set m=<first mediator in
sequence> <second mediator in sequence> within the Process syntax. This also dictates the layout of the bmatrix.
First
mediator
Second
mediator
Output for the first regression model is the same as before
The second regression model output reflects both task-
orientation and autonomous motivation as predictors
of controlled motivation.
1.5144
-.2704
.3216
1.5144
-.2704
.3216
.1402
.0138
The specific and total indirect effects here are still products of the
paths that trace from X to Y.
Ind1 = 1.5144 * .1402 = .2123
Ind2 = -.2704 * .0138 = -.0037
Ind3 = 1.5144 * .3216 * .0138 = .0067
This is a key for which tracings from X to Y are associated
with the indirect effects displayed in the table(s) above.
1.5144
-.2704
.3216
.1402
.0138
What if we want to change the specified direction of influence, where we have a direct effect of controlled motivation on
autonomous motivation? To ensure the sequencing is correct, we need to reverse the order of the mediators (in our
table below) and in the Process syntax.
1 - -
1 1 -
0 1 1
task
controlled
autonomous
intention
autonomous
controlled
Order reversed from
previous example
Unlike the previous example (see right), the
first mediator here is controlled motivation.
.2166
1.4991
.0705
.2166
1.4991
.0705
.1402
.0138
Final note:
I am sometimes asked whether it is possible to include multiple antecedent (X) variables in Process models. I have found
there is no straightforward way to do this. It is possible to include other antecedents (X’s) as covariates [using the ‘cov=‘
option] using Process syntax. However, when these variables are included in this way, their effects will be registered on all
variables ‘downstream’. In other words, these antecedents/covariates will be included as regressors in each regression
model predicting the downstream variables. To my knowledge, there is no direct way of fixing some of those effects to 0
[when you do not wish to include them in a model]. If this is a problem for you, the only thing I can suggest is to test your
model as a path model using an SEM program.
References
Hayes, A. F. (2022). Introduction to mediation, moderation, and conditional process analysis (3rd Ed.). New York: The
Guilford Press.
Keshtidar M., Behzadnia, B. (2017). Prediction of intention to continue sport in athlete students: A self-determination
theory approach. PLoS ONE, 12(2): e0171673. doi:10.1371/journal.
pone.0171673

More Related Content

Similar to Custom mediation models using Process macro An introduction 2023.pptx

[M2A2] Data Analysis and Interpretation Specialization
[M2A2] Data Analysis and Interpretation Specialization [M2A2] Data Analysis and Interpretation Specialization
[M2A2] Data Analysis and Interpretation Specialization Andrea Rubio
 
Predicting Employee Churn: A Data-Driven Approach Project Presentation
Predicting Employee Churn: A Data-Driven Approach Project PresentationPredicting Employee Churn: A Data-Driven Approach Project Presentation
Predicting Employee Churn: A Data-Driven Approach Project PresentationBoston Institute of Analytics
 
Sarcia idoese08
Sarcia idoese08Sarcia idoese08
Sarcia idoese08asarcia
 
Software quality management question bank
Software quality management question bankSoftware quality management question bank
Software quality management question bankselinasimpson3001
 
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...gerogepatton
 
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...ijaia
 
Churn Analysis in Telecom Industry
Churn Analysis in Telecom IndustryChurn Analysis in Telecom Industry
Churn Analysis in Telecom IndustrySatyam Barsaiyan
 
Course Title: Introduction to Machine Learning, Chapter 2- Supervised Learning
Course Title: Introduction to Machine Learning,  Chapter 2- Supervised LearningCourse Title: Introduction to Machine Learning,  Chapter 2- Supervised Learning
Course Title: Introduction to Machine Learning, Chapter 2- Supervised LearningShumet Tadesse
 
Machine Learning.pdf
Machine Learning.pdfMachine Learning.pdf
Machine Learning.pdfBeyaNasr1
 
New Directions in Mahout's Recommenders
New Directions in Mahout's RecommendersNew Directions in Mahout's Recommenders
New Directions in Mahout's Recommenderssscdotopen
 
5.0 -Chapter Introduction
5.0 -Chapter Introduction5.0 -Chapter Introduction
5.0 -Chapter IntroductionSabrina Baloi
 
NPTL Machine Learning Week 2.docx
NPTL Machine Learning Week 2.docxNPTL Machine Learning Week 2.docx
NPTL Machine Learning Week 2.docxMr. Moms
 
Lecture - 8 MLR.pptx
Lecture - 8 MLR.pptxLecture - 8 MLR.pptx
Lecture - 8 MLR.pptxiris765749
 
Multiple Regression.ppt
Multiple Regression.pptMultiple Regression.ppt
Multiple Regression.pptTanyaWadhwani4
 
BOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACRO
BOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACROBOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACRO
BOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACROAnthony Kilili
 

Similar to Custom mediation models using Process macro An introduction 2023.pptx (20)

[M2A2] Data Analysis and Interpretation Specialization
[M2A2] Data Analysis and Interpretation Specialization [M2A2] Data Analysis and Interpretation Specialization
[M2A2] Data Analysis and Interpretation Specialization
 
Predicting Employee Churn: A Data-Driven Approach Project Presentation
Predicting Employee Churn: A Data-Driven Approach Project PresentationPredicting Employee Churn: A Data-Driven Approach Project Presentation
Predicting Employee Churn: A Data-Driven Approach Project Presentation
 
Sarcia idoese08
Sarcia idoese08Sarcia idoese08
Sarcia idoese08
 
Software quality management question bank
Software quality management question bankSoftware quality management question bank
Software quality management question bank
 
Differential evolution optimization technique
Differential evolution optimization techniqueDifferential evolution optimization technique
Differential evolution optimization technique
 
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
 
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
A BI-OBJECTIVE MODEL FOR SVM WITH AN INTERACTIVE PROCEDURE TO IDENTIFY THE BE...
 
Forecasting (1)
Forecasting (1)Forecasting (1)
Forecasting (1)
 
SEM
SEMSEM
SEM
 
Churn Analysis in Telecom Industry
Churn Analysis in Telecom IndustryChurn Analysis in Telecom Industry
Churn Analysis in Telecom Industry
 
Course Title: Introduction to Machine Learning, Chapter 2- Supervised Learning
Course Title: Introduction to Machine Learning,  Chapter 2- Supervised LearningCourse Title: Introduction to Machine Learning,  Chapter 2- Supervised Learning
Course Title: Introduction to Machine Learning, Chapter 2- Supervised Learning
 
Machine Learning.pdf
Machine Learning.pdfMachine Learning.pdf
Machine Learning.pdf
 
New Directions in Mahout's Recommenders
New Directions in Mahout's RecommendersNew Directions in Mahout's Recommenders
New Directions in Mahout's Recommenders
 
Comparison and evaluation of alternative designs
Comparison and evaluation of alternative designsComparison and evaluation of alternative designs
Comparison and evaluation of alternative designs
 
5.0 -Chapter Introduction
5.0 -Chapter Introduction5.0 -Chapter Introduction
5.0 -Chapter Introduction
 
report
reportreport
report
 
NPTL Machine Learning Week 2.docx
NPTL Machine Learning Week 2.docxNPTL Machine Learning Week 2.docx
NPTL Machine Learning Week 2.docx
 
Lecture - 8 MLR.pptx
Lecture - 8 MLR.pptxLecture - 8 MLR.pptx
Lecture - 8 MLR.pptx
 
Multiple Regression.ppt
Multiple Regression.pptMultiple Regression.ppt
Multiple Regression.ppt
 
BOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACRO
BOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACROBOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACRO
BOOTSTRAPPING TO EVALUATE RESPONSE MODELS: A SAS® MACRO
 

Recently uploaded

Formulas dax para power bI de microsoft.pdf
Formulas dax para power bI de microsoft.pdfFormulas dax para power bI de microsoft.pdf
Formulas dax para power bI de microsoft.pdfRobertoOcampo24
 
Exploratory Data Analysis - Dilip S.pptx
Exploratory Data Analysis - Dilip S.pptxExploratory Data Analysis - Dilip S.pptx
Exploratory Data Analysis - Dilip S.pptxDilipVasan
 
AI Imagen for data-storytelling Infographics.pdf
AI Imagen for data-storytelling Infographics.pdfAI Imagen for data-storytelling Infographics.pdf
AI Imagen for data-storytelling Infographics.pdfMichaelSenkow
 
How I opened a fake bank account and didn't go to prison
How I opened a fake bank account and didn't go to prisonHow I opened a fake bank account and didn't go to prison
How I opened a fake bank account and didn't go to prisonPayment Village
 
The Significance of Transliteration Enhancing
The Significance of Transliteration EnhancingThe Significance of Transliteration Enhancing
The Significance of Transliteration Enhancingmohamed Elzalabany
 
Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...
Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...
Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...Valters Lauzums
 
一比一原版纽卡斯尔大学毕业证成绩单如何办理
一比一原版纽卡斯尔大学毕业证成绩单如何办理一比一原版纽卡斯尔大学毕业证成绩单如何办理
一比一原版纽卡斯尔大学毕业证成绩单如何办理cyebo
 
Pre-ProductionImproveddsfjgndflghtgg.pptx
Pre-ProductionImproveddsfjgndflghtgg.pptxPre-ProductionImproveddsfjgndflghtgg.pptx
Pre-ProductionImproveddsfjgndflghtgg.pptxStephen266013
 
社内勉強会資料  Mamba - A new era or ephemeral
社内勉強会資料   Mamba - A new era or ephemeral社内勉強会資料   Mamba - A new era or ephemeral
社内勉強会資料  Mamba - A new era or ephemeralNABLAS株式会社
 
basics of data science with application areas.pdf
basics of data science with application areas.pdfbasics of data science with application areas.pdf
basics of data science with application areas.pdfvyankatesh1
 
一比一原版西悉尼大学毕业证成绩单如何办理
一比一原版西悉尼大学毕业证成绩单如何办理一比一原版西悉尼大学毕业证成绩单如何办理
一比一原版西悉尼大学毕业证成绩单如何办理pyhepag
 
2024 Q1 Tableau User Group Leader Quarterly Call
2024 Q1 Tableau User Group Leader Quarterly Call2024 Q1 Tableau User Group Leader Quarterly Call
2024 Q1 Tableau User Group Leader Quarterly Calllward7
 
一比一原版麦考瑞大学毕业证成绩单如何办理
一比一原版麦考瑞大学毕业证成绩单如何办理一比一原版麦考瑞大学毕业证成绩单如何办理
一比一原版麦考瑞大学毕业证成绩单如何办理cyebo
 
一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理
一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理
一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理pyhepag
 
Supply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflictSupply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflictJack Cole
 
NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...
NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...
NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...Amil baba
 
Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...
Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...
Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...ssuserf63bd7
 
一比一原版阿德莱德大学毕业证成绩单如何办理
一比一原版阿德莱德大学毕业证成绩单如何办理一比一原版阿德莱德大学毕业证成绩单如何办理
一比一原版阿德莱德大学毕业证成绩单如何办理pyhepag
 

Recently uploaded (20)

Formulas dax para power bI de microsoft.pdf
Formulas dax para power bI de microsoft.pdfFormulas dax para power bI de microsoft.pdf
Formulas dax para power bI de microsoft.pdf
 
Exploratory Data Analysis - Dilip S.pptx
Exploratory Data Analysis - Dilip S.pptxExploratory Data Analysis - Dilip S.pptx
Exploratory Data Analysis - Dilip S.pptx
 
AI Imagen for data-storytelling Infographics.pdf
AI Imagen for data-storytelling Infographics.pdfAI Imagen for data-storytelling Infographics.pdf
AI Imagen for data-storytelling Infographics.pdf
 
How I opened a fake bank account and didn't go to prison
How I opened a fake bank account and didn't go to prisonHow I opened a fake bank account and didn't go to prison
How I opened a fake bank account and didn't go to prison
 
The Significance of Transliteration Enhancing
The Significance of Transliteration EnhancingThe Significance of Transliteration Enhancing
The Significance of Transliteration Enhancing
 
Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...
Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...
Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...
 
一比一原版纽卡斯尔大学毕业证成绩单如何办理
一比一原版纽卡斯尔大学毕业证成绩单如何办理一比一原版纽卡斯尔大学毕业证成绩单如何办理
一比一原版纽卡斯尔大学毕业证成绩单如何办理
 
Pre-ProductionImproveddsfjgndflghtgg.pptx
Pre-ProductionImproveddsfjgndflghtgg.pptxPre-ProductionImproveddsfjgndflghtgg.pptx
Pre-ProductionImproveddsfjgndflghtgg.pptx
 
社内勉強会資料  Mamba - A new era or ephemeral
社内勉強会資料   Mamba - A new era or ephemeral社内勉強会資料   Mamba - A new era or ephemeral
社内勉強会資料  Mamba - A new era or ephemeral
 
Machine Learning for Accident Severity Prediction
Machine Learning for Accident Severity PredictionMachine Learning for Accident Severity Prediction
Machine Learning for Accident Severity Prediction
 
Slip-and-fall Injuries: Top Workers' Comp Claims
Slip-and-fall Injuries: Top Workers' Comp ClaimsSlip-and-fall Injuries: Top Workers' Comp Claims
Slip-and-fall Injuries: Top Workers' Comp Claims
 
basics of data science with application areas.pdf
basics of data science with application areas.pdfbasics of data science with application areas.pdf
basics of data science with application areas.pdf
 
一比一原版西悉尼大学毕业证成绩单如何办理
一比一原版西悉尼大学毕业证成绩单如何办理一比一原版西悉尼大学毕业证成绩单如何办理
一比一原版西悉尼大学毕业证成绩单如何办理
 
2024 Q1 Tableau User Group Leader Quarterly Call
2024 Q1 Tableau User Group Leader Quarterly Call2024 Q1 Tableau User Group Leader Quarterly Call
2024 Q1 Tableau User Group Leader Quarterly Call
 
一比一原版麦考瑞大学毕业证成绩单如何办理
一比一原版麦考瑞大学毕业证成绩单如何办理一比一原版麦考瑞大学毕业证成绩单如何办理
一比一原版麦考瑞大学毕业证成绩单如何办理
 
一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理
一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理
一比一原版加利福尼亚大学尔湾分校毕业证成绩单如何办理
 
Supply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflictSupply chain analytics to combat the effects of Ukraine-Russia-conflict
Supply chain analytics to combat the effects of Ukraine-Russia-conflict
 
NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...
NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...
NO1 Best Kala Jadu Expert Specialist In Germany Kala Jadu Expert Specialist I...
 
Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...
Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...
Data Visualization Exploring and Explaining with Data 1st Edition by Camm sol...
 
一比一原版阿德莱德大学毕业证成绩单如何办理
一比一原版阿德莱德大学毕业证成绩单如何办理一比一原版阿德莱德大学毕业证成绩单如何办理
一比一原版阿德莱德大学毕业证成绩单如何办理
 

Custom mediation models using Process macro An introduction 2023.pptx

  • 1. Custom mediation models using Process macro with SPSS: An introduction Mike Crowson, Ph.D. August 2023 Video link: https://youtu.be/A4Sg5FjvpOA
  • 2. A couple of suggestions: 1. Draw out your conceptual model. It may also help you to draw out the components of the statistical model that are used to estimate your model parameters. 2. My recommendation is as part of your reporting ‘fill in’ the regression parameters (either unstandardized or standardized) into the conceptual model diagram. This is along the lines of what you would see with a path diagram when using a structural equation modeling program. This makes it easier to communicate your results to your reader. For more details on custom Process models, see Appendix B in Hayes (2022) book, Introduction to mediation, moderation, and conditional process analysis (3rd Ed.).
  • 3. The examples provided in this presentation are based on data from the article below. Nevertheless, the examples are not intended as replications of what the authors did in their analyses. Keshtidar M., Behzadnia, B. (2017). Prediction of intention to continue sport in athlete students: A self-determination theory approach. PLoS ONE, 12(2): e0171673. doi:10.1371/journal. pone.0171673 Link to SPSS data: https://drive.google.com/file/d/1x8IC8Iup1RQqh-nnNj2v5VydUN2eq0Br/view
  • 5. Task orientation Autonomous motivation Controlled motivation Intention to continue Process syntax to generate results for this model. y=consequent variable x=antecedent variable m=mediator(s) conf=confidence interval (here set at 95%) stand=1 (requests fully standardized regression coefficients) seed=12345 (This use important for generate the same bootstrap confidence intervals upon re-analysis of the same data with same model specifications; the seed # you use is optional; I simply used 12345) model=4 (model number of our mediation analysis)
  • 6. The specification using syntax is the same as this via the dropdown menu One downside of the dropdown menu is you cannot set a seed value. Thus, if you run your analysis more than once, the lower and upper bounds of the bootstrap confidence intervals will vary somewhat over analyses. The syntax option allows you to use a constant seed number across analyses that will result in the same lower and upper bound values.
  • 8. task control .2166 1.5144 .2166 The estimates I focus on throughout this Powerpoint will mainly be unstandardized regression coefficients from the component regression models. These are what I am showing in the diagrams above (and throughout the Powerpoint). However, you can easily use the standardized coefficients when making your own diagrams (if you would rather focus on those).
  • 10. Unstandardized indirect effects of task orientation (X) on intention to continue (Y) via each proposed mediator. These are computed using the unstandardized regression coefficients from the previous models. The specific indirect effect of autonomous motivation is .1952. The specific indirect effect of controlled motivation is .0031. The total indirect (see TOTAL) is computed as the sum of the two specific indirect effects: .1952 + .0031 = .1983 These effects are tested using the bootstrap 95% confidence intervals. If 0 (the null) falls between the lower and upper bound of an interval, that particular indirect effect is non-significant. If 0 falls outside the lower and upper bound for an interval, then the indirect effect is considered significant. We see here that the specific indirect effect of task orientation on intention to continue via autonomous motivation (IE=.1952) is significant [bootstrap 95% CI = (.1144,.2928)], whereas the specific indirect effect (IE=.0031) of task orientation via controlled motivation is not significant [bootstrap 95% CI = (- .0104,.0237)]. The total indirect effect (IE=.1983) on intention to continue via the two mediators is significant [bootstrap 95% CI = (.1140,.3002).
  • 11. Completely standardized specific indirect effects of task orientation (X) on intention to continue (Y) via each proposed mediator. These are computed using the standardized regression coefficients from the previous models.
  • 12. To be able to generate custom mediation models, we must create a bmatrix. This essentially is a matrix that specifies which regression parameters in your model are estimated and which are fixed to 0. My recommendation is to begin this process by (1) drawing out your model (if you have not already done so) and then (2) creating a k X k matrix (where k=number of variables in your analysis. Each column is headed by a variable that potentially points away from one variable in the model to another. Think of these variables as ‘upstream’. Each row represents a variable ‘downstream’ being pointed to by an upstream variable. Since a variable cannot predict itself, I have indicated that with a dash (-) in the matrix below for cells M1-M1 and M2-M2. Since cell M1-M2 (row 2, col 1) is open for specification, we are not allowed to leave cell M2-M1 (row 1, col 2) open for specification. - - - X M1 M2 M1 M2 Y Upstream variables with arrows pointing FROM them Downstream variables with arrows pointing TO them
  • 13. Next, we designate parameters that are freely estimated with a 1 and parameters that are not estimated with a 0. 1 - - 1 0 - 1 1 1 X M1 M2 M1 M2 Y In the Process syntax, the bmatrix option will collect the 1’s and 0’s row by row in this matrix: bmatrix=1,1,0,1,1,1 The first 1 is in row 1. The 1 followed by 0 is in row 2. The 1, 1, and 1 are found in row 3. Upstream variables with arrows pointing FROM them Downstream variables with arrows pointing TO them
  • 14. In the Process syntax, the bmatrix option will collect the 1’s and 0’s row by row in this matrix: bmatrix=1,1,0,1,1,1 The first 1 is in row 1. The 1 followed by 0 is in row 2. The 1, 1, and 1 are found in row 3. 1 - - 1 0 - 1 1 1 task autonomous controlled intention controlled autonomous Same as previous… The syntax in Process…
  • 15. To test this model, we need to remove the direct effect of task orientation (X) to intention to continue (Y). 1 - - 1 0 - 0 1 1 task autonomous controlled intention controlled autonomous Going back to our matrix, we set the parameter in row 3, column 1 to 0 (bolded below). Process syntax…
  • 16. Same results as before
  • 17. intention auton control .1402 .0138 1.5144 .2166 .1402 .0138 In this model, the direct effects of autonomous motivation and controlled motivation are not controlling for task orientation (as had occurred in our previous analysis).
  • 18.
  • 19. This model includes serial mediation, where one mediator has a direct effect on another mediator. To test this model, we need to remove the direct effect of task orientation (X) to intention to continue (Y) – as before – and add the direct effect of autonomous motivation on controlled motivation. 1 - - 1 1 - 0 1 1 task autonomous controlled intention controlled autonomous Going back to our matrix, we (again) set the parameter in row 3, column 1 to 0 (bolded below). Now, however, we change the former 0 (in row 2, column 2) to 1 (bolded below). Process syntax…
  • 20. Important: If you have mediators that follow a particular sequence (ordering), you need to set m=<first mediator in sequence> <second mediator in sequence> within the Process syntax. This also dictates the layout of the bmatrix. First mediator Second mediator
  • 21. Output for the first regression model is the same as before The second regression model output reflects both task- orientation and autonomous motivation as predictors of controlled motivation. 1.5144 -.2704 .3216
  • 23. The specific and total indirect effects here are still products of the paths that trace from X to Y. Ind1 = 1.5144 * .1402 = .2123 Ind2 = -.2704 * .0138 = -.0037 Ind3 = 1.5144 * .3216 * .0138 = .0067 This is a key for which tracings from X to Y are associated with the indirect effects displayed in the table(s) above. 1.5144 -.2704 .3216 .1402 .0138
  • 24. What if we want to change the specified direction of influence, where we have a direct effect of controlled motivation on autonomous motivation? To ensure the sequencing is correct, we need to reverse the order of the mediators (in our table below) and in the Process syntax. 1 - - 1 1 - 0 1 1 task controlled autonomous intention autonomous controlled Order reversed from previous example
  • 25. Unlike the previous example (see right), the first mediator here is controlled motivation.
  • 28.
  • 29. Final note: I am sometimes asked whether it is possible to include multiple antecedent (X) variables in Process models. I have found there is no straightforward way to do this. It is possible to include other antecedents (X’s) as covariates [using the ‘cov=‘ option] using Process syntax. However, when these variables are included in this way, their effects will be registered on all variables ‘downstream’. In other words, these antecedents/covariates will be included as regressors in each regression model predicting the downstream variables. To my knowledge, there is no direct way of fixing some of those effects to 0 [when you do not wish to include them in a model]. If this is a problem for you, the only thing I can suggest is to test your model as a path model using an SEM program.
  • 30. References Hayes, A. F. (2022). Introduction to mediation, moderation, and conditional process analysis (3rd Ed.). New York: The Guilford Press. Keshtidar M., Behzadnia, B. (2017). Prediction of intention to continue sport in athlete students: A self-determination theory approach. PLoS ONE, 12(2): e0171673. doi:10.1371/journal. pone.0171673