Testing Second Stage Moderated Mediation Models

Loading...

来自 University of Virginia 的课程

Understanding Your Data: Analytical Tools

24 个评分

University of Virginia

Understanding Your Data: Analytical Tools

24 个评分

Are you trying to understand data from your research? Learn how and when to conduct mediation, moderation, and conditional indirect effects analyses? Or, perhaps, how to theorize and test your theoretical models? If so, this is the course for you! We will walk you through the steps of conducting multilevel analyses using a real dataset and provide articles and templates designed to facilitate your learning. You'll leave with the tools you need to analyze and interpret the results of the datasets you collect as a researcher. By the end of this course, you will understand the differences between mediation and moderation and between moderated mediation and mediated moderation models (conditional indirect effects), and the importance of multilevel analysis. Most important, you will be able to run mediation, moderation, conditional indirect effect and multilevel models and interpret the results. This course is supported by the BRAD Lab at the Darden School of Business, which studies organizational behavior, marketing, business ethics, judgment and decision-making, behavioral operations, and entrepreneurship, among other areas. More: http://www.darden.virginia.edu/brad-lab/

从本节课中

Conditional Indirect Effects

Now that you know more about mediation and moderation, let's take a look at conditional indirect effects models, which are a combination of mediation and moderation models. First we will get to the heart of the differences between moderated mediation and mediated moderation models.This will allow you to fully understand the relationships between independent variables and dependent variables. By the end of the module, you will be able to theorize about conditional indirect effect models on SPSS and to test which path of the mediation model is affected by the moderator. Then you'll dive into SPSS, run different models, and learn how to interpret the results.

Conditional Indirect Effects: Mediated Moderation Models8:17

Testing Mediated Moderation Models: Steps 1-25:48

Testing Mediated Moderation Models: Steps 3-56:50

Moderated Mediation Models Overview7:30

Testing First Stage Moderated Mediation Models8:55

Second Stage Moderated Mediation Models Overview5:11

Testing Second Stage Moderated Mediation Models9:57

Dual Stage Moderated Mediation Models Overview5:18

Testing Dual Stage Moderated Mediation Models11:27

与讲师见面

Cristiano Guarana
Post-Doctoral Researcher
Darden School of Business

Let's run this model.

Let's conduct this analysis.

Into the model that we'll be running here,

we have job meaningfulness having an indirect effect on job performance via engagement,

and the relationship between engagement and performance is moderated by task complexity.

That's the model that we are running.

The first step is to look for the mediation and test for the mediation model.

Go to process, and how do you get there?

Go to analyze, and then once you are in the analyze option,

go to regression and then click on process.

Once you did that,

click and select the correct model.

It's model number four because we are running the mediation model,

and then you enter your variables.

You have test performance or job performance as your dependent variable,

you have meaningfulness as your independent variable and engagement as your mediator.

Go to options, ask for the total effect model and once you do that,

just click on OK.

I'm not showing to you the output for

a mediation model because I've already done that in a different session.

Please go back to the mediation session if

you have questions on how to interpret the output.

And now, we run a moderation model and

we look at the interaction or the interaction term,

and the interaction term is the multiplication of

the engagement variable and the task complexity variable.

You look at that term and if that term predicts performance.

And how do we do that?

Again, go to process.

Don't forget to select model number one because now it's a moderation model

and instead of having job engagement as your mediator,

you'll have job engagement as your independent variable and you

have task complexity as your moderator.

Don't forget to mean center,

and don't forget to get the matrix or the covariance matrix.

The covariance matrix is important because you need to

plot the interaction term and you get

the coefficients for plotting and conducting a simple slope analysis from this matrix.

Now, it's time to click on OK and you will get the output.

Once more, I'm not covering

all the output part here because we did already in a different session.

Again, go back and watch that session on

moderation if you have questions about how to interpret the output.

Now, we have to run model 15.

And model 15 will give us the output for interpreting or for

testing this conditional indirect effect

especially for second stage moderated mediation models.

Don't forget to select model 15,

and in one of these slides that I covered,

I entered model 14 or model 15.

Why model 15 and not model 14?

Both models will give you the interaction term of

the mediator and the moderator and the effects on the dependent variable.

But what is different from model 14 and model 15 is that,

with model 15 you control your partial O to the variance of

the interaction term between our independent

variable and our moderator on the dependent variable.

That's important. That's a more conservative test.

You should be conducting model 15 if you are conducting second stage models.

Another important piece of information here is that for a model 15,

your moderator goes to moderator V place.

Remember, for model eight when you are conducting first stage models,

our moderator goes to the moderator W space.

Here is V. Don't forget that,

if you'll make a mistake you'll get an error term.

And now, again, don't forget to mean center the products because we

will have an interaction term and we want it to avoid multi-collinearity issues.

And once you do that, the next step is to click on OK. And now we have the output.

So, the first thing that you do when you run a process model,

you should know now is to actually check for

the model number and this is model number 15.

We do have our dependent variable, independent variable,

mediator as engagement and our moderator is task complexity.

If we scroll down the output file,

we'll see basically those steps again.

Our first step, we'll look at the relationship between our independent variable,

job meaningfulness and job engagement.

And here we do find a significant relationship because P is less than 0.5.

The second step, we look at the interaction term and

the interaction term that effects on the dependent variable.

So, here we'll see that we do have two interaction terms because we

chose model number 15 and we are controlling

for the effects of an interaction term on the dependent variable.

Interaction term number one is job engagement and task complexity.

And interaction term number two is meaningfulness and task complexity,

the multiplication of those variables.

What we want to see is that the interaction term one has

a significant relationship with our independent variable, task performance here.

And we do find a significant relationship because P is less than 0.5.

Our interaction term number two is the control interaction term and it's not significant.

P is not less than 0.5.

The next step, we needed to take a look

at the bootstrapping analysis and the outcomes of that analysis.

We look at this part of the output here in which we have the mediator.

The first part of the output,

I can also explain to you.

This is actually that interaction term number two.

We are looking at the interaction between our independent variable

and the task complexity on performance.

Here, we can see that for low levels of task complexity,

we do have a significant mean effect,

if we are just considering that term.

But we are not interested in this part of the output.

We are interested in the second part of

the output because we wanted to see the conditional indirect effect.

Here we can see that when job engagement is low or the indirect effects

of meaningfulness on job performance via

engagement is significant only when task complexity is low.

If task complexity is, I mean,

high we can't actually perform well

regardless of our engagement level or how much meaningful the job is.

That's the theory behind it.

I don't have a stronger theoretical rationale here because the purpose of this session,

this workshop is to teach you how to conduct a conditional indirect effect analysis.

But vaguely, this is the reasoning behind.

Meaningfulness has an indirect effect on job performance via

engagement if task complexity is low or when task complexity is low.

The next step is to look at the index of moderated mediation,

and we do find that there is no zero in the bootstrapping analysis or in

the confidence interval of

the bootstrapping analysis for this index of moderated mediation,

more information or more support for our conditional indirect effect.

In this session, we covered second stage moderated mediation models.

We described how we theorized about these models and then

we analyzed our data set to show how to conduct this particular analysis.

If you have questions, go back and watch this session again.

探索我们的目录

免费加入并获得个性化推荐、更新和优惠。

Coursera 致力于普及全世界最好的教育，它与全球一流大学和机构合作提供在线课程。

© 2018 Coursera Inc. 保留所有权利。

通过 App Store 下载

通过 Google Play 获取