A moderation analysis via PROCESS model 1: meaning, procedure, plotting and interpretation

[@LissaAOE]

Just wanted to say that this was so helpful! I knew that it was probably an easy thing to do in Excel but struggling to figure out a simple and accurate approach. Thanks for posting this!

[@darkostojanovic900]

So far, the best explanation of process model 1. Tnx

[@Nobody-u9n]

Thank you mam . You solved my question of continuous and categorical moderator

[@hzolcs]

thank you for this video, I could never understood the whole thing without it and I would never finish my degree

[@golditewari]

I appreciate your comment! Thanks!

[@golditewari]

Hi Zoltan. You can subscribe to my website for similar content. edusecrets.com/subscribe/

[@georgiosbaxevanis7811]

Thank you very mych for the excellent video. Do you think it is possible to plot without the main effects and just with the interaction terms? Is this a proper statistical analysis?

[@BCFreelancer228]

Thank you, saved my time.

[@gurupritkaur4459]

Very clearly explained ma'am...

[@fauziahhayati6886]

Thank you so much, it really help me to understand the moderation with process

[@golditewari]

Thanks Fauziah! You can subscribe to my website for similar content. edusecrets.com/subscribe/

[@jackieleong5564]

thank you for saving my life :')

[@golditewari]

You are welcome!

[@张小爽]

you save me, really appreciate your work!

[@surangnutkongmunwattana485]

Hi, thanks for your VDO. I have continuous moderator, interaction effect is insignificant but the direct effect for each of them is significant. Also, you mentioned about the conditional effect (-1, 0, 1), 0 and 1 are significant. How can I interpret them?

[@golditewari]

hello surangnut, As per my knowledge, 0 and 1 may mean that the conditional indirect effect of your medium and high condition of your IV is significant on your DV.

[@fintansmith209]

Really clear explanation, thanks so much (-:

[@fuckschoolize]

This was very helpful! Thank you so much!

[@madhulata5953]

Hi Goldi, thanks for this video. You explained it really very well. I have a query. I am using model 14 to test moderated mediation. My moderator is a continuous variables not categorical. The index of moderated mediation is significant. But in conditional indirect effects, effects are significant at medium and high level but insignificant at the low level. I want to know if I can plot the graph and report it in the paper. I mean do we plot graph despite having insignificant conditional indirect effects?

[@charlamalover]

Do you need to check any assumptions before doing the PROCESS analysis? Since you have to check different assumptions using multipele regression, does this account for PROCESS too or is it not necessary to check any assumptions?

[@ng5437]

is the mediator and moderator same in this case? because from your model its a moderator but when doing the Hayes model its under mediator?

[@avanthikaswamy7258]

Hello, I have a question and need an answer asap please. I have one IV (no subscales), one moderator (4 subscales) and one DV ( 2 subscales). My moderator is a organisational justice. Can i run this model for it? if not which one can i use then? Please help its urgent. thank you

[@RahmanKhan-re7fy]

Thank you for such an informative video. I have moderation results as below: coeff se t p LLCI ULCI int_1 -.2517 .0560 -4.4963 .0000 -.3620 -.1414 Interaction effect is significant as Zero does not lies between LLCI and ULCI but with Negative sign on Coeff value does it suggest that Moderation has negative impact on relationship? For example with inclusion of Leadership between Self efficacy and Team performance the relationship becomes negative? Thanking you in advance.

[@golditewari]

As per my knowledge, following should b the interpretation: 'the effect of self efficacy on team performance is influenced by the moderator of leadership'. To know the effect of X on Y under different levels of Leadership (W), you should check the table 'Conditional effect of X on Y at values of the moderator(s)'.

[@IVeraaaI]

Hello Goldi! Can you tell me if I have to use the 'Dummy Variable'? I need to do a Moderator-analysis (Model 1) with a moderator that is nominal. I always read that i have to use a dummy variable instead (if it is not metric), but i don't know if it is for PROCESS necessary or do I only have to use a dummy variable ifI do it another way and can I do the analysis with a moderator that ist nominal?! I hope that you can help me. Thanks in advance!

[@mihikaprakash7240]

How will this process change when we have multiple independent variables and we want to test the moderator's relationship with the dependent variable?

[@belkiss17]

How can I use a multicategorical variable as a moderator? What model should I choose? And what should I report exactly in the output?

[@golditewari]

you can simply follow the same steps I mentioned in the video.

[@badawysaied7069]

Thank you. how to interpret the moderated moderation, please?

[@zyroulhisyam5829]

Hi, your video is very helpful. Thank you. I have 3 questions: 1) Can we just look at the p-value rather than looking at the LLCI & ULCI? 2) My moderator is a continuous variable, in the "model summary", the p-value = 0.00, and so do the p-values under the "conditional effect of X on Y...." section, however, the p-value under the "model" section = 0.9793. So are the interactions significant? 3) Is it possible to form the "Y=aX + bM = cXM +E" equation from the table?

[@golditewari]

Hello Zyroul, I hope following info helps: 1) you can, but normally for reporting moderation analysis, confidence intervals are used. You can refer some papers on that. 2) you should look for confidence intervals (LLCI & ULCI) in the int_1. 0 should be outside the confidence interval 3) I think you can form the equation, but I don't know the exact way of doing that. You can refer the book of Hayes on that. All the best!

[@StatisticalCat]

Actually, the correct answer is that whenever you are using bootstrapping then you have to look at LLCI and ULCI as the standard errors are different while performing a bootstrap. So your P Value and Bootstrap Value may differ. You may relate this to Type 1 or Type 2 error if you about them

[@caffelatte636]

Thank you for the video! It's very easy to understand for beginners like me :D

[@golditewari]

Thanks for the kind words Seo. You can subscribe to my website for similar content. edusecrets.com/subscribe/

[@mariaangelicapenacardenas5767]

I also have my independent and moderator coded. 0 and 1..Nonetheless When I looked at the conditional effects I dont get 0 and 1 as you do, I get random values. I am doing something wrong? thank you!!1

[@sorinaneamtu5191]

Thank you for the video! I still have a question tho. If my interaction doesn't present a significant value but on the conditional effect table I do have significant values can I say my hypothesis it s confirmed or not?!

[@golditewari]

No, the moderation effect is not significant if intr_1 is not significant.

[@eleanorehoo5687]

thank you for your video! i have a dummy variable too 1 (yes) and 2 (no) when assessing use of coping stategies (: however, when i ran the PROCESS analysis i did not manage to get the conditional effects as ".000" and "1.000". Instead they split the information into numbers, which i think represent low average and high. did you enounter this problem? and can you advise me on how to continue with this? thank you :)

[@golditewari]

yes, you are right. Model 1 splits your moderator into low medium and high categories. As far as I know, even if your moderator is a categorical variable, you can still use the same interpretation i.e for low medium and high conditions of your moderator. But you have the advantage that you can also just choose to use the results of low and high conditions of moderator. I hope this info helps. Good luck with you analysis!

[@GanasoundariJoshua]

Your explanation was so simple that I think even a novice can understand. Thanks again. I have question. If the responses were coded in likert scale, how do we study the interaction.

[@golditewari]

The scales used in the study mentioned in the video were also in LIkert scale. You can use the same interpretation as mentioned in the video. Normally, int1 is the value for interaction in Model 1

[@OMGimsoASIAN]

Hi! I have a question! I used the process macro 2018 (not sure if it makes a difference) and did what you said, however, my moderator is an ordinal. I cannot see the extra "data list" to plot a graph... (my X/independant variable has 2 conditions as well...) How can I make sure that the "data list" also runs?

[@golditewari]

hi! I think PROCESS MACRO 2018 is more advanced and has many new features. I still need to learn that.

[@dr.mayankpant1571]

Great video. Can you please explain what do you mean by when zero does not lie between your confidence interval that means your effect is significant

[@golditewari]

For example in my intr_1, Lower confidence interval (LICI)=.0936; Upper confidence interval UICI=.9861.Zero doesn't exist between these two confidence intervals (CI)

[@davidgleaves5175]

@@golditewari that's not two confidence intervals; it's one. The distance between the lower and upper limit is one interval.

[@tatianaray1324]

Thank you very much for your helpful video!

[@cheehyunyou950]

Thank you very much!! It helped me a lot!!

[@TheHeboabdi]

Thank you. Your video is amazing.

[@jun2bicos]

Thank you! Very informative!

[@sarahhshake]

Hi Goldi! I'm working with Model 2 in PROCESS on SPSS and I'm wondering in which boxes to move my moderator variables into? I tried putting one of the moderators in Proposed Moderator W and the other in Proposed Moderator Z, but the dialog box won't let me run the analysis when I do that. Does this mean that one of the two moderators has to go under "M Variable(s)" - and if so, why? Thank you, Sarah

[@golditewari]

Hi Sarah! try putting both variables (proposed moderators) together in the M variable(s).

[@sarahhshake]

Okay! Thanks!

[@sarahhshake]

Didn't work to put them together but put one in M Variable(s) and the other in Proposed Moderator W and it worked

[@golditewari]

thanks for the info. It's useful for me too!

[@ASADKHAN-fd5lw]

Goldi Tewari hi mam how to enterprit the moderator variable

[@naynarawat117]

Ma'am as you said moderation analysis is a kind of regression analysis. So that mean process macro also comes under parametric test. RIGHT?

[@golditewari]

yes!

[@naynarawat117]

@@golditewari thankyou so much ma'am for taking out your time to reply. I have one more question that if the population is 190 so what is the minimum sample size required for performing process macro?

[@golditewari]

@@naynarawat117 no problem. You can use a sample side of at least 30.

[@golditewari]

@@naynarawat117 Thanks for the kind words Nayna. You can subscribe to my website for similar content. edusecrets.com/subscribe/

[@naynarawat117]

@@golditewari sure ma'am. Thankyou once again

[@niroshakodithuwakku]

Hi can you explain how to do when 1 factor act as an moderating variable for mode than one independent variables

[@muqaddaskhalid8596]

I have a question. On the x-axis, you have mentioned the message frame. Shouldn't it be the moderator, i.e. crisis? because the message frame is shown by the lines in the graph.

[@eoinhassett6447]

great video thank you

[@francescagasparini4762]

Hi! Thank you for the video! I am currently having a problem with my process extension. When performing the moderation analysis, i do not have the "conditional effect" table in my output. How can i obtain it? I am using process v3.3 by Hayes on my Mac. Thank you to everyone willing to help!

[@golditewari]

Hello Francesca. May I know which model you are using.

[@shaziahasan1722]

Hi Dr. Tiwari, I want to ask that my direct effect and interaction effect both are insignificant. But the conditional effects of focal predictors at the values of moderator is significant. Is justifiable. How to interpret this result

[@deborahvanharten2014]

Hello! Thanks for this video. Do you know how to plot a graph for model 10? I know there is an effect, but I dont know under which condition of my independent variable.

[@golditewari]

hello Deborah, I am sorry, I am not sure about the syntax of model 10. all the best for your research!

[@deborahvanharten2014]

Ok thanks!

[@KhurramAziz]

Hi Goldi, Very good job. Any video on Process Model 59? Please share.

[@golditewari]

Thanks Aziz. I still need to work upon that. You will get the update when I upload a video related to that.

[@beenanair2786]

Good explanation. Please can you tell how to plot for model 2, with 2 continuous variables as moderator

[@alisonlobo3355]

hi, how to interpret negative scores obtained in the interaction effect?

[@alawiahalzawawi2786]

hello my moderator is type of franchisees i coded parallel as 0 and novice as 1 however i don't get o and 1 in Conditional effect of X on Y at values of the moderator(s)i get -.3089 and .6911 plus another question if my be was significant however LLCI was - and ULCI was + what should i do

[@golditewari]

hello Alwiah, You don't have significant results with this analysis. the LLCI and ULCI both should have the same sign for 0 to be outside the CI. I don't understand "I don't get o and 1 in Conditional effect of X on Y at values of the moderator(s)i get -.3089 and .6911". The effect need not to be in 0 or 1 digits.

[@garry6882]

how to set up the model in process if i have 4 IV and 2 moderating variable moderating those 4 IV?? thanks b4

[@inspiredliving4279]

Its well described

[@workid3556]

i am also trying to find an academic article where model 5 is used.i couldnt find it.can u please help me in this regard?

[@DasgesundetierDe]

Hi Goldi! :) very food explanation! thank you so much for this video and the effort you put into it. Would you be willing to answer me one question? How would the procedure look like if I had three instead of two categories for the IV? More specifically, I have four groups (three different conditions and one control group with no treatment). I would be more than happy if you would answer. Best, Sophie

[@golditewari]

It depends on what kind of IV you have. Categorical or continuous. I see that you actually have two IV here. Control and treatment and your treatment has 3 conditions.

[@mojojojoe22]

Thank you for this video. I'd like to know that there are any publication which use this method to explain moderator? I can't find it.

[@golditewari]

These are some of the books: Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. Guilford Press. Field, A. (2013). Discovering statistics using IBM SPSS statistics. Sage.

[@mojojojoe22]

Thank you for your suggestion, this really helps me. One more question, Are there any academic article which use this method?

[@golditewari]

There are ample number of studies available, you can just type 'moderation analysis' in google and you will find a lot of research articles. For instance: Peng, A. C., Lin, H. E., Schaubroeck, J., McDonough III, E. F., Hu, B., & Zhang, A. (2016). CEO intellectual stimulation and employee work meaningfulness: the moderating role of organizational context. Group & Organization Management, 41(2), 203-231. BarNir, A., Watson, W. E., & Hutchins, H. M. (2011). Mediation and moderated mediation in the relationship among role models, self‐efficacy, entrepreneurial career intention, and gender. Journal of Applied Social Psychology, 41(2), 270-297.

[@elifa.3047]

Hello, I am very confused about why we use process macro instead of performing ANOVA with two independent variables. What would be the difference if I use ANOVA with two IVs and check the simple effects?

[@golditewari]

Hello Elif, if you want to check the simple effects, you can use ANOVA. It's the same. But, if you want to use the moderation effect and also how the different conditions of your moderator affect the DV, you should use Model 1 of PROCESS Macro.

[@javeda]

I wanted to ask that can moderation analysis with PROCESS be possible with a binary dichotomous variable as the dependent variable? Please guide and help

[@golditewari]

As per my knowledge, it is only for dependent variables which are continuous in nature.

[@madiharazzam1098]

hey Goldi :) what if lower and upper limits include zero? does this mean that my moderating model is not significant?

[@golditewari]

Hello Mariah, zero should be outside the confidence interval for the model to be significant.

[@muqaddaskhalid8596]

One more question, please. Don't we have to look at the signs of the effects that have been shown in the PROCESS output? I'll be grateful if you respond to both queries of mine.

[@Cjstistajh]

Hi, I am a psychological wellbeing student and I am working on my dissertation which needs to use the moderation analysis. But I don't know which model I should choose. Can I please contact with you and ask some questions?

[@dr.mayankpant1571]

tewari ji my question is that the two variable iv and mod were both categorical?

[@golditewari]

yes, in my study.

[@helleveeg2428]

Hey! I need to analyze some data using the PROCESS macro. For this, I need to test a model with three moderators (so like model 2, but with three moderators instead of two). Apperantly , this should be possible, but I have no idea how I should do this. Do you, by any chance, know how to do this?

[@golditewari]

Hello helleveeg, for three moderators, you can use these models: Model 23 to 27 and also model 30 to 44. Just check the conceptual diagrams of these models first, you will then know which pathway is exactly being moderated and by which moderator. Can use this link for the conceptual diagrams: afhayes.com/public/templates.pdf

[@kanhiyatradershsr]

can same variable will be moderator or mediatior. means org innovation works as both moderator or mediators

[@golditewari]

Hi Atul, It is okay if your variable 'org innovation' works both as a moderator (in PROCESS model 1) and mediator (in PROCESS model 4 or 6). You can choose one model that fits more to your logic and expectations from the study or both if it's okay with the story of your study.

[@gurmeetmatharu]

Hi Goldi, Thanks for your video, its really informative, as you replied in your comments that your moderator has Likert scales but still you are using 0 and 1. so my question is have you changed your Likert scales into dichotomous variables, pls. In my case, my results as per your steps showing no significance. I tried it both ways, by applying moderator as high or low and in Likert scales as well. Below is when I used Likert scales int_1 .0432 .0722 .5982 .5500 -.0987 .1851 OR when I changed it into low or high int_1 .0129 .0814 .1586 .8741 -.1474 .1732 FYI, I'm checking Power distance effect on the relationship between Organic food purchase intention and behvaiour. Waiting for your reply.

[@golditewari]

hi Gurmeet. Actually, my moderator was categorical. 0=during the crisis, 1=after the crisis

[@jameswilkinson4632]

How do I specify what characters represent two groups for the graph ? Although I've specified 1 and 0 in the spss variable it's coming up with weird numbers and I can't tell which group is which! ._.

[@oleksiikoval3486]

How do you explain a low increment in R square?

[@eurwinchandra4058]

I have 3 independent variables, 1 moderating and 1 dependent. So does it mean that I have to run it 3 times each with different independent variable?

[@amnarasheed2859]

I have same problem, did you get to know the answer?

[@eurwinchandra4058]

@@amnarasheed2859 i asked the question in the research gate and some say that you can use PROCESS by calculating each independent variable with the moderator by using model 1

[@amnarasheed2859]

@@eurwinchandra4058 Ok got it. Thank you

[@jaimemorenovillegas1326]

In Process v.3, you must put the moderator in the W or Z box; does not function in the mediator field as you showed.

[@golditewari]

Hello Jaime, you need to put the moderator in the Z box for model 1

[@nickiehimuro-fitzgerald6116]

What if the model summary p value is insignificant, but the interaction is significant?

[@golditewari]

Hi Nicole, we generally check and report the confidence interval (CI) for the results of Model 1. 0 should be outside the CI for the results to be significant.

[@hooishanlim8765]

Dear Goldi, I have a significant interaction effect for model 1, my moderator is dichotomous, which is gender. I have a simple slope where both values at the moderator are significant. How do I interpret that? I also have another predictor for the model 1 that has a significant interaction effect. However, the simple slope for both values at the moderator are not significant. What does that mean? Thanks a lot.

[@golditewari]

Dear Shan, the graph in model 1 shows Conditional effect of X on Y at values of the moderator (M). So, you can interpret is as: X has a significant effect on Y for both, male (p

[@ikhhiomalik901]

can u plz tell me how to do "moderated hierarchical regression"?

[@golditewari]

I am sorry I am not very well-versed in "moderated hierarchical regression". But I think it's something that can be done using SEM.

[@amnarasheed2859]

Hi, I have a question please I have 3 independent variables, 1 dependent and 1 moderator. Do i need to run analysis 3 times? Or is there any option where i can add other independent variables?? I also have one controlled variable (age)

[@amnarasheed2859]

@@zahrakadiri1180no not yet. But i'll figure it out and if i get to know the answer i'll let you know too

[@zahrakadiri1180]

@@amnarasheed2859 Sure :)

[@amnarasheed2859]

@@zahrakadiri1180 someone asked this question on research gate and got the reply that you can use PORCESS MACRO by calculating each independent variable with the moderator by using model 1.

[@zahrakadiri1180]

@@amnarasheed2859 thank you 😊

[@alimertcankose4295]

if we have more independent variable than 1, how can we make a procress? and it s only OLS/ML estimation method(it s not benefical for ordinal data )

[@golditewari]

You can use two separate analysis for separate variables. Although, a moderator is also a kind of independent variable so Model 1 actually deals with two independent variables.

[@poonamkaushal6516]

Hii mam... Thank u for video Please help me out with a question i.e. do we have to centralize IV and moderator variable for Model 1 of Process Macro

[@golditewari]

Hi Poonam, you don't need to do that for Model 1. PROCESS already does that for you. For model 6 you need to centralize the variables.

[@drafsheenfatima1410]

Hello Dear Goldi. Thanks for the wonderful video. I have a confusion in drawing a graph. My all 3 variables are 5 point likert type scale. In my visualization conditional effect table my values are not like 0 and 1. It has 2 columns and 9 rows. In first two colums 3 values are repeating... Can you please guide me. I shall be grateful.

[@golditewari]

hello Fatima, Can you use the mean of all your variables for the analysis. I am sorry I couldn't understand your 'conditional effect table'. May be if you can paste it here I can have a look.

[@drafsheenfatima1410]

Thanks for the kind reply.following is the table. My framewark is like this = IV= organizational commitment, moderator = Hope, DV = Innovative performance Data for visualizing conditional effect of X on Y Paste text below into a SPSS syntax window and execute to produce plot. DATA LIST FREE/OC_IV HOP_MOD IP_DV. BEGIN DATA. 2.8572 2.7292 3.0127 3.6383 2.7292 3.1237 4.4195 2.7292 3.2347 2.8572 3.6037 3.3676 3.6383 3.6037 3.5358 4.4195 3.6037 3.7039 2.8572 4.4781 3.7225 3.6383 4.4781 3.9478 4.4195 4.4781 4.1732 END DATA. GRAPH/SCATTERPLOT=OC_IV WITH KS_DV BY HOP_MOD. ******************** ANALYSIS NOTES AND WARNINGS ************************* Level of confidence for all confidence intervals in output: 95.00 ------ END MATRIX -----

[@golditewari]

The second column shows low medium and high level of your moderator (hope). Is your intr_1 significant? You can only see this table when your moderation is significant.

[@shaimac4210]

@@golditewari Hi, I am getting same kind of table during my analysis,but my intr_1 is not significant.How to interpret this?Please help

[@golditewari]

@@shaimac4210 Hello! it means your moderator doesn't influence the relationship between your independent and dependent variable.

[@idealstwo3396]

Hi，I want to know why you did not choose "Mean center for products" when you set the options？

[@golditewari]

As per my knowledge, mean centre is only required while doing the serial mediation analysis (Model 6)

[@idealstwo3396]

but i saw some books and videos says when you do moderation analysis ,if your X or W variable is categorical variable ,you needn't mean center ,but if they both are interval variables ,it's necessary to mean center .So ,i don't know which one is correct.

[@golditewari]

might be possible for interval variable. My X and W were categorical so I didn't use mean centre for that. Can you please share the reference of the literature you referred. I can update my knowledge as well. Thanks!

[@idealstwo3396]

that is in Chinese , is OK?

[@golditewari]

ok!

[@Lucas-we2rj]

very helpful

[@hasanarslan7496]

can anyone help me? I do not have same options ( at 7:23). my menu is different. How can it be possibble?

[@golditewari]

Normally you should get it when you use model 1

[@shabrinaherawati7777]

How to get Process Spd ?

[@SaloneDutchess]

How do I add controle variables that are dummies

[@nadeemawan3244]

I am using PROCESS for moderation. I need to know that how to apply ANOVA first for control variables and then how to use or not those control variables in moderation analysis using hayes PROCESS - via YTPak(.com)

[@golditewari]

you need not to apply ANNOVA for control variables. You can just put the control variable as 'covariate' in the PROCESS analysis.

[@yasirahmed3914]

Please explain the model no 2

[@rashidaparveen7199]

it creates a sound like suzuki in Pakistan

[@workid3556]

Hi mam what are the basic assumptions of process model?

[@golditewari]

Can read this: www.itl.nist.gov/div898/handbook/pmd/section2/pmd21.htm

[@rajabilal4651]

thanks

[@hafijulshaikh1305]

Mam can I talk to you about analysis in my study?

[@khuzaimashafique9869]

HI! how to download process plugin files

[@golditewari]

here: processmacro.org/download.html

[@aishajaved2388]

Can you explain model 16

[@latoundjifolahanmedard3872]

je m'appelle Medard LATOUNDJI du BENIN. Bonjour, je voudrais la version française de"moderation analysis via PROCESS : meaning, procedure, plotting and interpretation".

[@golditewari]

Dear Folahan, I am really sorry. I don't know French so can't provide you with the translation. All the best! -Goldi

[@workid3556]

hi mam.can you please guide how to interpret process model 5

[@golditewari]

I am still working on that. Will let you know when I am done.

[@rutwik31]

which is the link to install plugin

[@golditewari]

here: processmacro.org/download.html

[@Vatika-jg7es]

ma'am can you please share your email, I some queries regarding spss in Mac

[@teknowlogychannel370]

See this video also; kzbin.info/www/bejne/ZoXWiqyAiJeBnq8

[@mytj6729]

Your graph is confusing.

[@rashidaparveen7199]

ap ny chorian pehni hwi hain next time please na pehnain it is distracting

[@Cjstistajh]

Hi, I am a psychological wellbeing student and I am working on my dissertation which needs to use the moderation analysis. But I don't know which model I should choose. Can I please contact with you and ask some questions?

[@StatisticalCat]

you can contact me benny at [e-mail]