21:09 I'm assuming these are unstandardized b-values, correct? What's the best way to see the standardized betas for each of these values? Also, what's the best way to depict the changing path coefficients in a path diagram, considering the moderator variable is in play? Thanks!
Indeed, the coefficients derived from the PROCESS are unstandardized, meaning they are presented in their original metric and may vary widely depending on the scale of the variables involved. The PROCESS produces mainly unstandardized coefficients. In any version of PROCESS, you can standardize your variables first prior to the use of the PROCESS, and this will generate standardized coefficients. Standardizing variables prior to utilizing the PROCESS tool can be accomplished seamlessly within software packages such as SPSS. Good luck!
Dear Dr Gurbuz, Thank you for the clear and detailed explanation. May I ask one question? The conceptual framework model of the thesis I chose has 5 independent variables (with 5-point Likert scale items), 2 moderating variables( items with Yes or No question types), and 1 dependent variable. What is the better way to analyze this? regards, Samuel G.
Hi Samuel, To assess your conceptual model effectively, I recommend employing the simple moderation model, designated as number 1. You need to test your hypotheses one by one. E.g., first predictor, first moderator, and the outcome variable. The procedural steps are largely similar and straightforward.
Thank you for your vedio, i amy ask please if the direct relationship between independent and dependent variable is negative and significant, and moderating variable is negatively moderate what does this mean please? Is this mean when when the moderator is high the negative relationship between independent and dependent variable is weakness? Or how please. Many thanks for your help and your cooperation
Thank you for your response. Understanding the nuances of regression analysis, particularly when it involves moderating effects, can indeed be challenging. To facilitate interpretation, it's helpful to delve into the Process outputs, specifically under the section titled "Conditional effect of Focal predictor at value of the moderator." This section allows you to see how the association between the independent and dependent variables varies based on different levels of the moderator. By examining these conditional effects, you can gain insights into how the relationship between the independent and dependent variables changes as the moderator variable fluctuates. Good Luck!
Hellou. I am currently working on data that explores the link between negative childhood experiences, the Dark Triad and alcohol consumption. In a linear regression, I found that both negative childhood experiences and the Dark Triad predict alcohol consumption. However, in the analysis using PROCESS, I found that the direct relationships between these variables were no longer significant. Instead, I observed a moderating effect, with negative childhood experiences acting as a moderator between the Dark Triad and alcohol consumption, supporting my hypothesis. I would be interested in your views on these differences in results and perhaps some advice on how to better understand and interpret these results. Could you please explain to me why these differences between the results of the linear regression and PROCESS analysis might occur? Could it be related to the moderation effect that I have observed? Thank you for your time and help.
Thank you for sharing your intriguing findings regarding the link between negative childhood experiences, the Dark Triad, and alcohol consumption. It's fascinating to see how different analytical approaches can yield varied results, as you've described. The differences you've observed between the results of the linear regression and the PROCESS analysis are not uncommon, especially when there are interaction effects among predictor variables. These differences could be attributed to several factors, including suppression, multicollinearity, and moderation, as you mentioned. Multicollinearity, in particular, can distort results in linear regression by inflating standard errors and making it difficult to assess the individual contributions of predictors. To mitigate this issue, it's advisable to ensure that predictors are not highly correlated and consider mean centering or standardizing them before analysis. To better understand and interpret these results, I recommend a two-step approach. First, thoroughly examine the direct effects in linear regression to assess the individual contributions of predictors. Once you've established these direct effects, focus on testing your moderation hypothesis using the PROCESS macro. In this later analysis, pay particular attention to the moderating effect (X.M), as you've already addressed the direct effects through linear regression without the PROCESS macro. Good Luck!
@@saitgurbuz Thank you for your reply. I have already carried out the analysis in two stages. First I checked the predictions in linear regression, the multicollinearity is good, the VIF is also below 4. Then the predictors were standardised. So you are suggesting that in the PROCESS analysis I should focus only on the moderation effect data, but how can I defend the difference in the data when I am presenting my master's thesis, can you advise me?
@@ritabogusiene7445 Dear, in addressing your concern about defending the differences in data presentation for your master's thesis, I believe there's a strong argument to be made based on the specificity of your hypothesis regarding the direct effect. Given that your hypothesis regarding the direct effect doesn't explicitly incorporate all predictors and interaction effects, it's reasonable to conduct separate analyses to test the main effects using linear regression and the moderating effect using the Process macro. This approach allows for a focused examination of each hypothesis, ensuring clarity and precision in your analysis. Good Luck!
@@saitgurbuz Thank you for your reply. But does the use of different methods explain why the direct effects disappear in the moderation analysis, whereas they did in the linear analysis?
You should report that there is no evidence of a moderating effect in your model. No need to report the slope test or further details. For the discussion section, you should provide some potential reasons for the lack of moderation.
![[TF2] The Russian Bear Trap](http://i.ytimg.com/vi/YFNSwz2_rOA/mqdefault.jpg)
This is great explanation. Thank you, saved me hours of work.
Very clear and lucid explanation 👌
Great video! Thank you for the clear explanation
21:09 I'm assuming these are unstandardized b-values, correct? What's the best way to see the standardized betas for each of these values? Also, what's the best way to depict the changing path coefficients in a path diagram, considering the moderator variable is in play? Thanks!
Indeed, the coefficients derived from the PROCESS are unstandardized, meaning they are presented in their original metric and may vary widely depending on the scale of the variables involved. The PROCESS produces mainly unstandardized coefficients. In any version of PROCESS, you can standardize your variables first prior to the use of the PROCESS, and this will generate standardized coefficients. Standardizing variables prior to utilizing the PROCESS tool can be accomplished seamlessly within software packages such as SPSS. Good luck!
Dear Dr Gurbuz,
Thank you for the clear and detailed explanation. May I ask one question? The conceptual framework model of the thesis I chose has 5 independent variables (with 5-point Likert scale items), 2 moderating variables( items with Yes or No question types), and 1 dependent variable. What is the better way to analyze this?
regards,
Samuel G.
Hi Samuel,
To assess your conceptual model effectively, I recommend employing the simple moderation model, designated as number 1. You need to test your hypotheses one by one. E.g., first predictor, first moderator, and the outcome variable. The procedural steps are largely similar and straightforward.
Thank you for your vedio, i amy ask please if the direct relationship between independent and dependent variable is negative and significant, and moderating variable is negatively moderate what does this mean please? Is this mean when when the moderator is high the negative relationship between independent and dependent variable is weakness? Or how please.
Many thanks for your help and your cooperation
Thank you for your response. Understanding the nuances of regression analysis, particularly when it involves moderating effects, can indeed be challenging. To facilitate interpretation, it's helpful to delve into the Process outputs, specifically under the section titled "Conditional effect of Focal predictor at value of the moderator." This section allows you to see how the association between the independent and dependent variables varies based on different levels of the moderator. By examining these conditional effects, you can gain insights into how the relationship between the independent and dependent variables changes as the moderator variable fluctuates. Good Luck!
Hellou. I am currently working on data that explores the link between negative childhood experiences, the Dark Triad and alcohol consumption. In a linear regression, I found that both negative childhood experiences and the Dark Triad predict alcohol consumption. However, in the analysis using PROCESS, I found that the direct relationships between these variables were no longer significant. Instead, I observed a moderating effect, with negative childhood experiences acting as a moderator between the Dark Triad and alcohol consumption, supporting my hypothesis.
I would be interested in your views on these differences in results and perhaps some advice on how to better understand and interpret these results. Could you please explain to me why these differences between the results of the linear regression and PROCESS analysis might occur? Could it be related to the moderation effect that I have observed?
Thank you for your time and help.
Thank you for sharing your intriguing findings regarding the link between negative childhood experiences, the Dark Triad, and alcohol consumption. It's fascinating to see how different analytical approaches can yield varied results, as you've described.
The differences you've observed between the results of the linear regression and the PROCESS analysis are not uncommon, especially when there are interaction effects among predictor variables. These differences could be attributed to several factors, including suppression, multicollinearity, and moderation, as you mentioned. Multicollinearity, in particular, can distort results in linear regression by inflating standard errors and making it difficult to assess the individual contributions of predictors. To mitigate this issue, it's advisable to ensure that predictors are not highly correlated and consider mean centering or standardizing them before analysis.
To better understand and interpret these results, I recommend a two-step approach. First, thoroughly examine the direct effects in linear regression to assess the individual contributions of predictors. Once you've established these direct effects, focus on testing your moderation hypothesis using the PROCESS macro. In this later analysis, pay particular attention to the moderating effect (X.M), as you've already addressed the direct effects through linear regression without the PROCESS macro. Good Luck!
@@saitgurbuz Thank you for your reply. I have already carried out the analysis in two stages. First I checked the predictions in linear regression, the multicollinearity is good, the VIF is also below 4. Then the predictors were standardised. So you are suggesting that in the PROCESS analysis I should focus only on the moderation effect data, but how can I defend the difference in the data when I am presenting my master's thesis, can you advise me?
@@ritabogusiene7445 Dear, in addressing your concern about defending the differences in data presentation for your master's thesis, I believe there's a strong argument to be made based on the specificity of your hypothesis regarding the direct effect.
Given that your hypothesis regarding the direct effect doesn't explicitly incorporate all predictors and interaction effects, it's reasonable to conduct separate analyses to test the main effects using linear regression and the moderating effect using the Process macro. This approach allows for a focused examination of each hypothesis, ensuring clarity and precision in your analysis. Good Luck!
@@saitgurbuz Thank you for your reply. But does the use of different methods explain why the direct effects disappear in the moderation analysis, whereas they did in the linear analysis?
@@ritabogusiene7445 Certainly, since you have more predictors in testing moderation using Process, sometimes direct effects emerge as insignificant.
Is it possible to use process macro if my IV is measured on a likert scale, my DV is continuous and my moderator is categorical?
Yes
if my p value not sig and there is no role for moderator in my model what can i do or what can i right in the result report?
You should report that there is no evidence of a moderating effect in your model. No need to report the slope test or further details. For the discussion section, you should provide some potential reasons for the lack of moderation.