How to Interpret Interaction Graph/Slope with Moderation Analysis?

[@ssakuindy]

Thank you so much, I had no idea on how to interpret this graphic and your explanation helped me a lot.

[@researchwithfawad]

You're very welcome!

[@drmanishasingh5606]

Amazing analysis! Simplified for a new learner...always dependable!Thanks a lot for all the efforts you make!

[@researchwithfawad]

Thanks. I am glad you liked it.

[@JeanYeli]

Thank you so much for this video..Please how do you go about 3 way interaction effects?

[@researchwithfawad]

Thanks for watching. I havent yet got anything on it. But hopefully soon.

[@ainunfadilah313]

​@@researchwithfawadUpppp

[@fatimasheikh8514]

Thank you so much! It helped me a lot in my exam

[@researchwithfawad]

Glad to hear that!

[@lindsaycy7518]

Hi Fawad. Thank you for another great video. Please could you advise how you were able to adapt the tool kit for Model 2? The one in your link only seems to have space for one moderator?

[@researchwithfawad]

Thanks for watching. You will need to create two separate graphs.

[@covieducation9998]

Very informative Dr. Thank you for the video. I have received a comment from the reviewer stating that I should draw interaction chart conataining 7 points since I have used 7 point Likert scale. The excel sheets are by default on 5 points. What could be its solution how can I drag it to 7 points instead of 5 ?.

[@researchwithfawad]

If you have access to the excel file, you can consult and excel expert to make modifications.

[@covieducation9998]

@@researchwithfawad Dr. Sb, I have received a second revision from the Journal I am stuck with one comment from the reviewer, and I do not feel satisfied with the comments from different sources. I request that you help me address this issue. The comment is given below: "My question on email vs. social media sampling was not fully answered. I appreciate the authors’ response here, but more to the point, I would like to see the authors at least statistically compare the email group to the social media group to ensure that there are no meaningful differences between these sub-groups within the sample. Otherwise, if the authors can support their response with recent research or sources and clearly explain how the two sources of participants should not have created sub-group differences within the sample, then I would potentially entertain such a response (though the use of two different sampling frames within the same study is still potentially problematic and at the very least would call for some analysis to rule that out)." I want you to suggest: 1. What statistical tool can be used to compare the email group and social media networks? 2. Provide me with a couple of the latest references to address this problem. 3. Is there any other suggestion or comment that you think is more appropriate? Thank you in anticipation. Looking forward to hearing from you soon. Regards,

[@ouedraogoliliane7118]

Hi Dr Fawad Thank you so much for your amazing videos! I've an urgent question :) I was wondering if is possible to take the coefficient of a SEM, done in SAS to do the slope analysis. If not, is there a way to adjust xlim and ylim of the slop analysis plot ( I just want to see positive values). Thank you so much for your help !

[@researchwithfawad]

I havent used SAS, but i do not see any issue in using the path coefficient.

[@shontenegeri4495]

Dear Dr. Fawad thank you very much and be blessed for sharing such important knowledge with us. I found it is so helpful for my study. Having said this, can I use this particular table for an ordinal variable? Or this is only used for categorical variables?

[@researchwithfawad]

Yes more or less it will remain the same.

[@shontenegeri4495]

@@researchwithfawad Ok, many blessings to you🙏🙏🙏🙏🙏

[@Hotdog_Connoisseur]

Hi Dr. Fawad thank you for the videos, I am wondering how to interpret data where the moderator coefficient and interaction coefficient is not significant but for the simple slope the low moderator is significant. I am required to report regardless of significance, just wondering how I would interpret that. Thank you Sir

[@researchwithfawad]

This means overall the moderator is insignificant. However, at only one level it is significant.

[@chiaradadamo7050]

Dr Fawad Thank you so so much for your incredibly helpful videos!! I have an urgent question. I am proof reading the research report that I need to ubmit tomorrow and I am only now detecting a potential error. My main effects (simple linear regressions) are negative. and the B of the interaction effect in my multiple regression analysis is positive. I am interpreting this as my moderator strengthens the negative relationship between my IV and DV. However, I used the excel stats tool that you used in another one of your videos, and that claimed that my moderator 'dampened the negative relationship between my IV and DV'. Please, could you clarify the difference between strengthen and dampen? Do 'dampen' and 'weaken' mean the same things? I don't know how I should be interpreting my results. In addition, my scatterplots show that the line for LOW moderator are steeper compared to the line of the HIGH moderator. However the B for the interaction is positive. How is this possible?

[@researchwithfawad]

Thanks. I am glad you liked it. Dampens means that moderator weakens the relationship. Whereas positive interaction means that moderator strengthens the negative relationship.

[@chiaradadamo7050]

@@researchwithfawad Thank you so much for replying! Could you please clarify whether this interpretation of my findings would be correct? I would be immensely appreciative!!! The main effect of my IV (perceived belonging) on my DV (Psychotic symptoms) is negative. In other words, Lower perceived belonging predicts more psychotic symptoms. In the moderation analysis, where the interaction effect of my IV (perceived belonging) and moderator (individualism) is negative. Can I interpret this as 'Individualism weakened the relationship between perceived belonging and 'psychotic symptoms', therefore, at higher levels of individualism, the lower perceived belonging poses a lower risk for psychotic symptoms, compared to high levels of individualism? Thank you very much in advance. You're amazing.

[@researchwithfawad]

If the moderation effect is negative, this means the the moderator weaken the relationship between the IV and DV. So the moderator (IND) weakens the direct relationship between IV and DV. You will have to see when the relationship between the IV and DV is positive or negative so to know how it is affected negatively by the moderator.

[@AzamKhan-pq3vq]

The path coeffiecnt unstandarised regression values should be from bootstrapping or pls algorithym please explain

[@researchwithfawad]

Bootstrapping

[@AzamKhan-pq3vq]

@researchwithfawad where in bootstrapping

[@shontenegeri4495]

Another question: Is there any difference between a simple slope analysis and a Slope plot during reporting moderation results?

[@researchwithfawad]

No. Both are same as both are moderation slopes.

[@shontenegeri4495]

@@researchwithfawad Thank you so much and God bless you at large🙏🙏🙏🙏🙏

[@Mskhosravi5506]

Hello Dr, For such a moderation analysis that you have run , is it necessary to report goodness of fit indexes in any thesis or article?????Because my goodness of fit in my moderation analysis is not good at all, what should I do?? Many thanks

[@researchwithfawad]

For path analysis with no latent variable and indicators, the fits are not good. Why not perform moderation with full structural model.

[@Loislee_]

Hi with this plot, please can I use the sem path coefficient figures instead of R square.

[@researchwithfawad]

Yes, path coefficients are used. R square values are not used in plotting

[@annafienz]

Dr. Fawad, what about parallel slope (both line not touching one another) ? How do i interpret it?

[@researchwithfawad]

Please refer to the gradient of the slope. The interpretation is based on the steepness of the gradient. Is it steeper at low, mean, or high level. Interpretation is later based on the which line is steeper.

[@Mskhosravi5506]

Hello Dr, excuse me do you have slope analysis for standardized regression weight for moderating effect and to share with me?? I have run moderating interaction in amos, and I need slope analysis based on standardized estimate, of course I have searched Jeremydawson website, but I have not found it. many thanks in advance.

[@researchwithfawad]

You can use James Gaskin Stats Tool Package.

[@priyashreeroy2280]

Can the Jeremy Dawson excel sheet be used for output from SPSS using Hayes process macro?

[@researchwithfawad]

Yes it can be

[@carriefang2113]

Hi Dr Fawad, where to get the R-square value?

[@researchwithfawad]

Check Square Multiple Correlation from Analysis Properties in AMOS.

[@carriefang2113]

@@researchwithfawad Thanks, Sir. I have other questions. 1. When making the graph, we only entered one group of data of unstandardized regression coefficients for the independent variable, moderator, and interaction, how can the graph present two lines (one for low moderator, one for high moderator)? 2. after confirming the significant moderating effect, the next step is to probe the interaction where the high level and low level of the moderator effects will be calculated. Any connection between the low and high moderators presented in the graph and the low and high level of moderator effect on the relationship in the next step?

[@muhammadaamirkhan8182]

What about if one line goes upward and another goes downward? What does the graph show in that situation?

[@researchwithfawad]

Is the moderation significant? Which one is steeper? If the graph is opposite to what you hypothesized, the hypothesis is rejected.

[@muhammadaamirkhan8182]

@@researchwithfawad Yes, moderations are significant. In the first case, the higher one is steeper; however, in another case, the lower one is steeper.

[@hananeid4254]

May you send the excel sheet used in explanation?

[@researchwithfawad]

Thanks. You can Google James Gaskin Stats Tool Package.