Categorical Predictor Variables Using SmartPLS/Dummy Variables Regression Model in SMART-PLS3

Categorical Predictor Variables Using SMART-PLS/Using Dummy Variables in SMART-PLS. This session guides on how to use Categorical Predictor Variables in SMART-PLS through Dummy Coding.
SmartPLS4 is now released. if you want to learn the new and updated PLS-SEM tool, please refer to the step by step playlist to learn SmartPLS 4.
• SmartPLS4
For Notes on the topic, please visit
researchwithfawad.com/index.p...
#SmartPLS #SMARTPLS3 #CategoricalPredictorVariables #DummyVariables #DummyCoding #SEM #PLSSEM #QuestionnaireDataAnalysis #ResearchCoach

[@zahidmasroor3703]

Exactly what i needed, thanks a bunch sir.

[@researchwithfawad]

Thanks. I am glad you liked it

[@iqbalnatakusumah6804]

Salam Good Sir Fawad, thank you so much for the video. It really helps my thesis!

[@researchwithfawad]

Thanks l. I am glad you liked it.

[@jhonyh8884]

Thankyu sir, your tutor so helpfully to understand dummy variabel

[@researchwithfawad]

You are most welcome

[@m.sswetha3165]

Awesome explanation about it sir eagerly waiting for categorical variable as a moderator in PLS

[@researchwithfawad]

Thanks. I am preparing contents. Hopefully that video will be done this week.

[@lidyachristine5214]

@@researchwithfawad hi, have you made this one yet?

[@renato_psy]

Excellent video, thank you very much for the explanation. I have the following question. What if I want to compare not only the variables to the reference variable (senior vs junior, and middle vs junior), but I am also interested on the differences between senior and middle. What should I do or look? I guess the beta to see who is more predictive, but I am not sure. In my case, I am doing a research on work modalities, such as teleworking, face to face and hybrid work in a complex model with a lot of variables to see their effects on work family conflict. I could do an anova but i want to incorporate all of them on the pls sem model, to compare the three with each other, not only two of them in relation with the reference. Thank you!

[@researchwithfawad]

Thanks for watching. You can change the reference category.

[@inayahali9691]

Thank you so much for this informative video. Helped me loads to understand in testing hypotheses using categorical IDV with continuous DV. I have a question please- I have 1 categorical Independent variable (users vs non-users) and 5 DV (continuous) so can I do moderation analysis in this using one moderator (continuous). Looking forward for your response sir! JAK heaps

[@researchwithfawad]

Yes, why not. The procedure doesn't change. You will have multiple moderating effects if the moderator is moderating multiple relationships.

[@inayahali9691]

@@researchwithfawad thank you heaps!

[@grisdymahardikana8372]

Hi sir thank you for your video, i would like to ask you about the categorical variable on your latest example. in your video the Middle level is not significant to all of the latent variabel, so i want to ask what we should do next to treat this variable? it's better to remove middle level from the model? or just keep it in the model, then if it's better to keep it in the model why should it keep on the model (in case that it doesn't seem significant at all) ? Thanks in advanced :)

[@researchwithfawad]

Thanks. I am glad you liked it. No need to remove the variables if they are not significant. You will need to discuss as to what are the possible reasons for insignificance.

[@pusamus]

Hi Fawad, I love your PLS videos. It is so much easier to understand compared to others. For dummy variables, I saw that you create a latent variable for each indicator. Instead all n-1 dummy indicators pointing towards a formative latent variable. May I know the rationale? Thanks

[@researchwithfawad]

Thank you. I am glad you liked the videos. Yes, since with categorical variable, the results are interpreted as comparison to the reference category.

[@tayyabsafder4346]

Respected sir, I am highly thankful for your explanation. I want to clear one important thing, what's happened if the beta coefficient (Sample O) showing negative. What should we need to understand because t-value is greater than threshold value of acceptance 1.96 but beta value is in negative. Is it significant relationship or not. May kindly share your knowledge. Looking forward to your valuable response. JAZAK ALLAH 👍

[@researchwithfawad]

Thanks l. I am glad you liked it. A negative beta would mean a negative relationship between the variables. Negative is sign tonshow how an exogenous variable affects the endogenous variable. Even if the relationship is positive or negative, the relationship still should be tested for significance, this is where you check t value or p value.

[@tayyabsafder4346]

@@researchwithfawad Once again thank you. If you further explain its interpretation, let we are writing responsiveness dimension on satisfaction (Responsiveness ---> Satisfaction), its beta coefficient is - 0.43, t-value is 3.82 and p-value is 0.000. How may we interpret this result of path coefficient. Your response will be highly valuable for me. Jazak Allah.

[@tayyabsafder4346]

Please share your publication link or journal name wherein you published your articles. Jazak Allah 👍

[@researchwithfawad]

This actually shows responsiveness has a negative relationship with satisfaction. Although, theoretically this should not have been the case.

[@researchwithfawad]

Here is the list of my publications researchwithfawad.com/index.php/publications/

[@yamadoudiop1293]

Hello thank you very much for your many very instructive videos. Please I have a question do we need to check the reliability and validity of a Dummy variable ( continious categorical) eg small, medium, big company represented by number of employees? or is it not even possible?

[@researchwithfawad]

Thanks. I am glad you liked it. Reliability and Validity is only assessed for latent constructs.

[@yamadoudiop1293]

@@researchwithfawad ok thanks !!!

[@muhammadasyrafmohdkassim924]

Good morning Sir.. I would to get an explanation on how to run the measurement model for this dummy variables? this video shows the analysis for structural model.

[@researchwithfawad]

Thanks for your comment. Dummy variables are not assessed for quality criteria.

[@nursyahirah3200]

Thank you for the video sir. I have a question, i want to measure their knowledge/awareness which the better measurement is a Yes/No question instead of likert scale. Can I include Yes/No type of question in pls-sem? Thank you in advance

[@researchwithfawad]

You can have Yes and No questions but as Predictors, if you do it for dependent variables, this will require you to run Binary Logistic Regression which is not possible in Smart-PLS

[@nursyahirah3200]

@@researchwithfawad thank you so much for the reply.

[@zafaralam3406]

@@nursyahirah3200 Hi Ms Nur, Warm Greetings. Did you measure knowledge/awareness being independent variable on Yes/No question based scale and successfully apply PLS - SEM in your research article? Please reply as it is crucial for my research.

[@chefberrypassionateresearcher]

Sir, can i use this method in place of one way ANOVA? I feel it will solve my purpose. I want to do it over formative constructs, so can i do it?

[@researchwithfawad]

It depends on your study objectives.

[@trishalachauhan1615]

Very useful video..thank you so much sir..i would like to ask you one thing..if i want to check impact of demographic variables (age, income, gender, occupation etc .) On customer brand engagement being independent variable(likert scale)...can smart pls be used??

[@researchwithfawad]

Yes you can use these variables, if they are categorical, treat them as dummy variables, if they are numerical (interval or ratio) use them as indicators.

[@trishalachauhan1615]

Thank you so much sir

[@trishalachauhan1615]

Sir one more thing i want to know, if i am using three dimensions of customer engagement ie content, contribution, consumption and items for them, so my latent dependent variables would be the 3 dimensions or the customer engagement?

[@researchwithfawad]

If you just taking the CE at higher level then you have CE at DV otherwise you can take the three dimensions as DV at Lower Level.

[@trishalachauhan1615]

@@researchwithfawad that's really helpful...thank you so much sir

[@KaranBerry-ol2hg]

Prof. Can I use this technique to find significant differences, if my construct is FORMATIVE in nature? What should i do to analyze difference in individual items of a formative construct between three groups? Each of my items are measured on a 5 point Likert scale.

[@researchwithfawad]

Yes, you can. Same procedure for individual items.

[@KaranBerry-ol2hg]

@@researchwithfawad Thanks for the reply. Q1. I don't understand if i want to find significant difference between the groups on individual items CSR1 to CSR 6 (incase they are formative items), then how to do it...? Please suggest.. Q2. And also do i need to do MICOM (measurement invariance) before doing this, as during MICOM second step, if some of the items may be insignificant in my constructs, which i will have to remove them...

[@researchwithfawad]

You need to read about MICOM and how it is different from categorical variables used in regression. You don't need MICOM if you are using categorical variables in regression

[@chefberrypassionateresearcher]

Sir, Can i do the same process, if my dependent variable is a formative construct, and not a reflective construct. Please answer.... What process i should use incase of a formative construct

[@researchwithfawad]

Yes you can.

[@user-kz8dq5hu8m]

Hello, it’s quite useful for me. I would like to ask one question. If one control variable(1 year/2year/3year /4year and above), I also create dummy variables in spss and bootstrap in smartpls, but 1year is significant, 2 year/3 year is insignificant. But how do I report this control variable from whole? Thank you.

[@researchwithfawad]

Thanks for watching. For categorical predictor on ordinal scale you do not need to create dummy variables.

[@user-kz8dq5hu8m]

@@researchwithfawad really? So I can directly draw this variable and point it to Y in smartpls?

[@researchwithfawad]

Yes, you can

[@user-kz8dq5hu8m]

@@researchwithfawad Great！ thanks so much! 👍

[@norainejamil1494]

Good day sir. I have created dummies variable for AGE and ran bootstrapping but my sample mean and p value turned out to be n/a. Kindly advise. Thank you sir

[@researchwithfawad]

If age is a Categorical variable, you create dummy variables and add all except the reference category for age.

[@samadkakar6933]

Salam Fawad Latif, I am using variables in my study. The variables are age groups, gender, experience. How do I code age groups? For instance, I have age groups ranged between 21-5, 26-30, 31-35, and so on. Please guide me I shall be very thankful to you.

[@researchwithfawad]

If there are multiple categories like Age, you will have similar number of dummy variables - 1 (n-1). Each will behave as separate variable and will be added separately in SMART-PLS. If it is gender, you do not need any dummy variables, Gender basically has two categories, so no need for further variables. See this video as well for more clarification on Dummy Variables kzbin.info/www/bejne/sJKyd4ukjahrj5I

[@ingyselim2760]

Would You plz clarify more, I have 4 control variables, 2 of them have 5 categories (e.g academic rank) I already created dummy variables for this control variable which created for me 5 categories. When it comes to smartpls if I added the 5 categories at once I feel the results are invalid, what should I do if I need to know the effect of each category on the DV. I would be really grateful if you replay

[@talhamansoor7108]

SIr, can we use categorical variable as a moderator in PLS>

[@researchwithfawad]

Yes we can. I am in process of making a video on that.

[@talhamansoor7108]

@@researchwithfawad right Sir

[@zahidmasroor3703]

@@researchwithfawad Sir, I'm waiting for it.

[@farahfarhana4014]

Salam Dr, let say I got two groups (0 - 40) and (41- 60+), can I assign them as 1 = 0-40 and 2 = 41-60+. Then I just add in Age groups put them as ordinal. Connect them via moderating effect on SmartPLS4?

[@researchwithfawad]

WAS. Yes.

[@Alhamzah_F_Abbas]

Thank you, prof. actually, I tried to run it on SPSS but I couldn't find it, I don't know why

[@researchwithfawad]

But this is SmartPLS

[@Alhamzah_F_Abbas]

@@researchwithfawad I meant the creat dummy variables icon in SPSS

[@researchwithfawad]

Which version are you using, try searching Google for Solution.

[@Alhamzah_F_Abbas]

@@researchwithfawad I will Prof. Thank you

[@chefberrypassionateresearcher]

IS MEASUREMENT INVARIANCE REQUIRED TO BE DONE FOR THE THREE GROUPS BEFORE THIS PROCEDURE?

[@researchwithfawad]

Yes for MGA

[@chefberrypassionateresearcher]

@@researchwithfawad Thanks for your reply. Sir, pls guide how to do measurement Invariance testing in smart pls3, on a formative construct.

[@fizzahnaveed455]

Kindly tell where are the values of reference category variable i-e junior category

[@researchwithfawad]

Reference category values are not there. The comparison is made to the reference category in terms of whether the effect is high or low. For example, Lets assume you have variable Gender (Female = 0, Male = 1) and you have linked it to a variable called Job Commitment. A Positive beta would mean that Male have higher Commitment in comparison to female. I hope this helps.

[@fizzahnaveed455]

@@researchwithfawad JaZak Allah Khair Sir... You are very helping and kind . May Allah SWT bless you with more knowledge and devotion ameen

[@fizzahnaveed455]

@@researchwithfawad assalam-o-alaikum Sir Kindly tell that can we use 3,4 demographic factors as dummy variables (the demographic factors are used as independent variables )in model

[@fizzahnaveed455]

Kindly guide

[@researchwithfawad]

Yes, you can use as many dummies as you like.

[@shivamazad2712]

But what if I also want to compare between Pakistan and China, should I've to change my dummies and re-run the model?

[@researchwithfawad]

Yes you can do that

[@shivamazad8466]

Thank you for the information. If I've taken Income (upto 3 lakhs, 3-5 lakhs, 5+) as a Moderator, and I'm checking the relationship between investment intention and financial benefits regarding a product. Can I take Income as Ordinal variable and interpret that higher the income group have more/less tendency towards investment?