Hello again Leila,

I will make an assumption here that each of the dimensions are measuring something respective to their variable, I will call them a construct. If so then in essence you have 3 multi-item constructs. That is, you have 5 items measuring the moderator and 3 items for each of the 2 other constructs.

The first test that needs to be done is a reliability test for each of the items on their respective construct (refer to Nunnally and Bernstein, 1994). This can be achieved through SPSS. 

The next stage is to run validity test to make sure that each item is measuring what it purports to measure (See Pedhazur and Schmelkin, 1991). There are 2 major tests to run here: convergent validity (are they measuring what the should be measuring) and discriminant validity (how well they not measuring one of the other constructs). There are two ways you can do this. One is to use SPSS factor analysis or secondly use AMOS following the first step in the two stage process advocated by  Anderson and Gerbing (1988) to find the average variance extracted (AVE) where the AVE must be greater than .5 (See Fornell and Larker, 1981). The latter method has gained high credence in the academic world.

When you have achieved validity then there are there are two methods to test your theory. One, aggregate and average your valid items for each respective construct and then run the regressions as identified in previous communication using SPSS; or, two, use the second stage of Anderson and Gerbing methodology using AMOS but remember in this case you will need a large sample N>200.

So in summary you will need to run these amounts of tests: 3 reliability tests using SPSS (1 for each construct), 1 validity check if using AMOS to test for both convergent and discriminant validity or 1 factor analysis if using SPSS, then once the validated items have been aggregated and averaged,  3 regressions using SPSS (Using the Baron and Kenny approach) or 1 path model approach using AMOS.

What exact method you use I will leave up to you, but make sure that the method used is accepted by the journal you are aiming the article for.

Hope this helps.

Hi Leila,

Perhaps it doesn't suit your research question, but would it not be easier to use the total score of your likert scales? I would suggest using the PROCESS macro for SPSS (www.afhayes.com).

bbnbn

Hypotheses are a keystone to causal theory. However, wording hypotheses is clearly a struggle for many researchers (just select at random any article from a good academic journal, and count the wording issues!). In this section I offer examples of how you might word different types of hypotheses. These examples are not exhaustive, but they are safe.

Direct effects

"Diet has a positive effect on weight loss"

"An increase in hours spent watching television will negatively effect weight loss"

Mediated effects

For mediated effects, be sure to indicate the direction of the mediation (positive or negative), the degree of the mediation (partial, full, or simply indirect), and the direction of the mediated relationship (positive or negative).

"Exercise positively and partially mediates the positive relationship between diet and weight loss"

"Television time positively and fully mediates the positive relationship between diet and weight loss"

"Diet affects weight loss positively and indirectly through exercise"

Interaction effects

"Exercise positively moderates the positive relationship between diet and weight loss"

"Exercise amplifies the positive relationship between diet and weight loss"

"TV time negatively moderates (dampens) the positive relationship between diet and weight loss"

Multi-group effects

"Body Mass Index (BMI) moderates the relationship between exercise and weight loss, such that for those with a low BMI, the effect is negative (i.e., you gain weight - muscle mass), and for those with a high BMI, the effect is positive (i.e., exercising leads to weight loss)"

"Age moderates the relationship between exercise and weight loss, such that for age < 40, the positive effect is stronger than for age > 40"

"Diet moderates the relationship between exercise and weight loss, such that for western diets the effect is positive and weak, for eastern (asia) diets, the effect is positive and strong"

Mediated Moderation

An example of a mediated moderation hypothesis would be something like:

“Ethical concerns strengthen the negative indirect effect (through burnout) between customer rejection and job satisfaction.”

In this case, the IV is customer rejection, the DV is job satisfaction, burnout is the mediator, and the moderator is ethical concerns. The moderation is conducted through an interaction. However, if you have a categorical moderator, it would be something more like this (using gender as the moderator):

“The negative indirect effect between customer rejection and job satisfaction (through burnout) is stronger for men than for women.”

Handling controls

When including controls in hypotheses (yes, you should include them), simply add at the end of any hypothesis, "when controlling for...[list control variables here]" For example:

"Exercise positively moderates the positive relationship between diet and weight loss when controlling for TV time and diet"

"Diet has a positive effect on weight loss when controlling for TV time and diet"

Another approach is to state somewhere above your hypotheses (while you're setting up your theory) that all your hypotheses take into account the effects of the following controls: A, B, and C. And then make sure to explain why.

Supporting Hypotheses

Getting the wording right is only part of the battle, and is mostly useless if you cannot support your reasoning for WHY you think the relationships proposed in the hypotheses should exist. Simply saying X has a positive effect on Y is not sufficient to make a causal statement. You must then go an explain the various reasons behind your hypothesized relationship. Take Diet and Weight loss for example. The hypothesis is, "Diet has a positive effect on weight loss". The supporting logic would then be something like:

Weight is gained as we consume calories. Diet reduces the number of calories consumed. Therefore, the more we diet, the more weight we should lose (or the less weight we should gain).

I guess mr. Hamidi copied/pasted his contribution, without referencing, from: http://statwiki.kolobkreations.com/wiki/Structural_Equation_Modeling

Thanks to you all

But i still waiting for response to my question :(

Hello Leila,

A question that many have asked. There are many way to approach the question but first you must ask yourself does theory state that "C" moderates "A" and "B". If so then you can use structural equation modeling as you have theory supporting your model (you will also need a large sample: N > 200). If you are testing this theory for the first time then there are 2 ways that this can be achieved. Please find below an exert from a journal article that provides a stringent method to test for both mediation and moderation using a 2 stage method. There are also references that may help with your question.

"First, based upon the work of Baron and Kenny (1986) simple regression analyses were applied to test for mediation and moderation effects to provide an initial set of preliminary results. However, when conducting tests for moderation using regression, interaction effects are also entered into the equation. To eliminate multicollinearity the predictor variable and the moderators were "centred" before testing the interaction (Aiken and West, 1991; Baron and Kenny, 1986). To achieve this, scores were “zeroed” by simply subtracting the sample mean from all individuals' scores on that variable, producing a revised sample mean of zero. This transformation has no impact on the level of significance of the interaction terms (Holmbeck, 1997).

In the second stage structural equation modelling (SEM) was applied as, from a statistical viewpoint, the simultaneous fitting of paths is both more parsimonious and will yield better results. That is, it will give more precise estimates (smaller standard errors of predictors) and less bias, as each effect is estimated while partialling out the other effects (Iacobucci, 2008). A second advantage of SEM is the ability to deal with measurement error. Regression models do not explain measurement error separately from the error attributable to a model's lack of fit, where in path analysis part of the model is distinguishable from the factor analysis part of the model. The importance of this being the reduction of problems with multicollinearity, that is, the multiple indicators are represented by a factor, the measurement is properly represented, and the problem is averted (Iacobucci, 2008)."

So in essence if you are testing for moderation for the first time then  would use the regression approach. If you have a sample of over 200 then you can also use SEM to validate your finding.

I hope this helps.

Cheers

Hello Leila,

Sorry but I forgot to mention to use SPSS for the regression and AMOS for the SEM

Hello again Leila,

I will make an assumption here that each of the dimensions are measuring something respective to their variable, I will call them a construct. If so then in essence you have 3 multi-item constructs. That is, you have 5 items measuring the moderator and 3 items for each of the 2 other constructs.

The first test that needs to be done is a reliability test for each of the items on their respective construct (refer to Nunnally and Bernstein, 1994). This can be achieved through SPSS. 

The next stage is to run validity test to make sure that each item is measuring what it purports to measure (See Pedhazur and Schmelkin, 1991). There are 2 major tests to run here: convergent validity (are they measuring what the should be measuring) and discriminant validity (how well they not measuring one of the other constructs). There are two ways you can do this. One is to use SPSS factor analysis or secondly use AMOS following the first step in the two stage process advocated by  Anderson and Gerbing (1988) to find the average variance extracted (AVE) where the AVE must be greater than .5 (See Fornell and Larker, 1981). The latter method has gained high credence in the academic world.

When you have achieved validity then there are there are two methods to test your theory. One, aggregate and average your valid items for each respective construct and then run the regressions as identified in previous communication using SPSS; or, two, use the second stage of Anderson and Gerbing methodology using AMOS but remember in this case you will need a large sample N>200.

So in summary you will need to run these amounts of tests: 3 reliability tests using SPSS (1 for each construct), 1 validity check if using AMOS to test for both convergent and discriminant validity or 1 factor analysis if using SPSS, then once the validated items have been aggregated and averaged,  3 regressions using SPSS (Using the Baron and Kenny approach) or 1 path model approach using AMOS.

What exact method you use I will leave up to you, but make sure that the method used is accepted by the journal you are aiming the article for.

Hope this helps.

Hello Dr.Robin D. Pentecost 

Thank you for your very useful information.

I really appreciate your help!

Hi Leila

Adding to Robin's wonderful answer to analysing data:

Moderating effect = interaction effect

1. Run preliminary data analysis (PDA). This includes necessary analyses to prepare your dataset for the actual analysis. Steps are provided in a pre-edit document (Introduction to SPSS) in the link below.

2. Create composite score for each latent variable.

3. In SPSS: Run ANOVA (twice): (1) with full factorial model , and (2) with custom model (interaction). [Analyze > General Linear Model > Univariate...]

4. Looking at your question: You may have created a product (interaction) by multiplying items for predictor variable with items for moderating variables. This way you can run the regression.

5. However, if you are looking to test your research model/theory and/or looking for predictive efficacy of your theory, I would suggest go for PLS-SEM rather than covariance-based SEM. AMOS provides covariance-based SEM. A good start is that you're already familiar with regression. Thus, PLS-SEM should be very easy for you.

You can try have a look at a simple document on PLS-SEM in the second link below. Note that I use WarpPLS in the example. You can download a full-working 3-month trial version of the software at http://www.scriptwarp.com/warppls/#Free_trial_version. 

Best wishes,

Saiyidi

Article Introduction to SPSS

Chapter Partial least square in a nutshell | Saiyidi MAT RONI 2 0 1 4

Hi,

i strongly recommend that you check Andrew Hayes' PROCESS macro to test mediation, moderations and combinations. His book is on guilford press. Easy, convenient and efficient. 

http://www.afhayes.com/introduction-to-mediation-moderation-and-conditional-process-analysis.html

Hi Leila,

Robin and Saiyidi's suggestions are good.  I am also confused with your question and not sure about you research objective. If you really want to measure moderating effects of 5 dimensional moderator with multi-item measure in each dimension where the antecedent and criterion variables are also multi-dimensional (3 dimensions in each) and all are measured by Likert scale, then you have real challenge of very complex model to test.  To do that you need very large sample to maintain moderate level of k to n ratio for a valid and reliable result.  You may consult David Kenny's suggestion in the following link.

http://davidakenny.net/cm/moderation.htm

However, if Robin's assumption is right, follow his suggestion of testing scale reliability and use arithmetic average of each variables to develop construct. Then use Moderated Regression Analysis to test the extent of interaction effect of the moderator variable (M) on the causal relation between antecedent (X) and criterion variable (Y). Run 3 Regression models - first, to measure effect of X on Y, second, effect of M Y to test if the effect is significant (check the R2-change) and finally add the product of X and M in the regression model to test moderating effect (check the R2-change and significance level to see if the interaction is actually meaningful).

Hope this will help.

Yunus

Plz go through Baron & kaenny & Aiken & West, 1991 and also take care of the problem of Multi-collinearity.