Dealing with Control Variables in PLS Path Model using SmartPLS: Path Analysis and Interpretation

Поделиться
HTML-код
  • Опубликовано: 8 фев 2025

Комментарии • 50

  • @asarwaribit
    @asarwaribit 3 года назад +1

    Very good explanation of Smart-PLS, reasonable speed.

  • @anugrewal9790
    @anugrewal9790 5 месяцев назад

    Greetings sir
    Can we take demographic variables such as age,gender,education as control variables in pls sem?
    Also is there a need to perform mga as these are control variables?

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  4 месяца назад

      @@anugrewal9790 Inclusion of CVs is already explained in this lecture. Please note that structural techniques don't provide detailed CV analysis as in HLR. MGA should only be performed if moderation is hypothesized.

    • @anugrewal9790
      @anugrewal9790 4 месяца назад

      @@ChMahmoodAnwar thank you sir
      In case a control variables comes significant
      Do I need to perform mga ? Because it’s not my hypothesis part to check the impact of demographic variables?
      I will be very thankful

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  4 месяца назад

      @@anugrewal9790 You can simply report these significant results. MGA can also be run as an auxiliary test.

    • @anugrewal9790
      @anugrewal9790 4 месяца назад

      @@ChMahmoodAnwar ok thank you sir

  • @MM-mt2jy
    @MM-mt2jy 3 года назад

    Thanks for your video. I’d like to ask 1) if I have more control variables such as gender, age, type of job, how do we include in the model? 2) If I have more than 2 types of job, how do I record it as control variable? Should it be like 1=doctor, 2=accountant, 3=teacher ?

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад +1

      Hi. You may include other control variables in similar way as briefed in this lecture. Code controls as you mentioned. Best.

  • @lisarei6193
    @lisarei6193 2 года назад

    Thank you for the video. Can I use consistent PLS algorithms if my IVs are reflective but I include gender, age, income etc as control variables? or do I need to consider the model as including formative elements when I include control variables?

  • @azadehart
    @azadehart 3 года назад

    Thanks for the video. I was wondering did you extract control variable from spss? If yes how did you do it? you have latent variable here. I need to do the same. I appreciate your help.
    Thanks

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад +1

      Nominal or categorical controls are not considered as latent variables. Watch my lecture "How to prepare and upload Data file for PLS Path Modeling in SmartPLS" to get an idea how to prepare data file for SmartPLS.

    • @azadehart
      @azadehart 3 года назад

      @@ChMahmoodAnwar thanks i will check it

  • @NgocLe-mi7ys
    @NgocLe-mi7ys 11 месяцев назад

    Hi Sir, thank you for your email. I would like to ask something. What is the difference between testing control variables and ANOVA? If I am not familiar with testing control variables in your way, may I use Smart PLS for assessing measurement and structural model and use ANOVA to test the impact of control variables, such as income level ? Thanks.

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  11 месяцев назад

      Control variables are analyzed using Stepwise Regression Analysis, whereas ANOVA is used to assess mean differences between 3 or more sample means. For control variables analysis in SPSS, please search and watch my lecture on it. Thanks

  • @furongjia896
    @furongjia896 2 года назад

    Hi Ch. Mahmood Anwar
    your video helps me a lot but may I ask how do you code gender. for example there are male and femal and how do you make them in one variable? thx

  • @asarwaribit
    @asarwaribit 3 года назад

    if we are using SMart-PLS, do we need to find out the edogeneity? (2) can we calculate ENDOGENEITY USING SMART-PLS? WHICH SOFTWARE IS THE BEST FRO CALCULATION OF ENDOGENEITY? HOPE YOU WILL ANSWER IN DETAIL

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад +1

      Endogeneity is a critical CLRM assumption, regardless of analysis technique you are using i.e., OLS, ML, PLS etc. It should be tested, especially, in causal models. It can easily be detected with Durbin-Wu-Hausman test (augmented regression test) in STATA. Once you get an evidence of Endogeneity, you need to use 2SLS to get unbiased and consistent results asymptotically. Hope this will help!

    • @asarwaribit
      @asarwaribit 3 года назад

      @@ChMahmoodAnwar Thanks for response

  • @arshiyasaba2259
    @arshiyasaba2259 2 года назад

    Sub sample 500 is the sample size? I have 251 as my sample size, so should I need to change the sub sample size into 251? Please explain

  • @yasirrasool2594
    @yasirrasool2594 3 года назад

    Salam, I am just a learner, but you are an expert. As per my little knowledge, gender as control variable has significant effect on loyalty because both BCLL and BCUL values are negative. What you say?

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад +1

      W.Salam. For control variables, don't look at CIs, simply observe t-statistic and its significance (in this example it is less than 1.96, insignificant at 0.059-Red colour). Hint: Run the PLS Algo for model with control variables than run Bootstrapping for same model. You will find that outcome is similar for control variables.

  • @sadafpervez4188
    @sadafpervez4188 2 года назад

    sir plz guide threshold of the control variable, and their interpretation.

  • @asarwaribit
    @asarwaribit 3 года назад

    Please explain how to handle 2 or more than 2 Dependent variable while writing equation in Eviews and interpret the results so obtained or share any link which explains this issue. Thanks

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      You can write a loop in Eviews like this and interpret results in standard way:
      for %dep dep1 dep2 dep3
      equation eq{%dep}.ls {%dep} c x1 x2 x3
      next
      Hope it'll work for you.

    • @asarwaribit
      @asarwaribit 3 года назад

      @@ChMahmoodAnwar the mentioned above did not work. assume I have 3 independent variable and multiple independent variables, how to write equation. please suggest

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      @@asarwaribit It really works. May be you are not entering the code properly. You should read the programming chapter of the EViews Command and Programming Reference. Thanks.

  • @aumtantrajin
    @aumtantrajin 3 года назад

    What if I have more than one control variables? should I separate them?

  • @reynreyn9900
    @reynreyn9900 3 года назад

    what if the control variable is a latent construct (like trust in your paper)?

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      Theoretically, SEM and PM were not designed to analyze control variables. Use regression if you want to analyze or work with controls.

    • @reynreyn9900
      @reynreyn9900 3 года назад

      @@ChMahmoodAnwar thank you so much for the answer Professor. I have one more question, are control variable and moderator variable the same?

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад +1

      @@reynreyn9900 Controls and moderators lie at same level. Controls need to be controlled in the model, whereas moderators regulate IV-DV signal strength and/or relationship direction.

    • @reynreyn9900
      @reynreyn9900 3 года назад

      @@ChMahmoodAnwar thank you so much Professor! I understand better now

    • @ben20007
      @ben20007 3 года назад

      Wonderful. Sir how can I Contact you.

  • @jerryhsiehbig
    @jerryhsiehbig 3 года назад

    Salam, so SmartPLS could just basically tell you whether significant or not, but can it tell you whether a male or a female has a higher mean value? Or this sort of test should be done in ANOVA using SPSS.
    BTW If the gender or other control variables are significant in smartpls , is it possible to be insignificant in SPSS since they were actually using different methods?
    English ain't my native tongue, hope you could actually understand what I'm talking about

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      W. Salam. Actually Path Models and SEM were not developed to analyze controls. Regression is the classic model that can handle controls and other confounding variables. As far as SmartPLS is concerned, you can run a Multi Group Analysis (MGA), but will only tell you whether a group significantly influence the outcome or not?

    • @Austresi
      @Austresi 3 года назад

      Hey, I am doing analysis similar to yours. Can you help how you did that?

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      @@Austresi Hi. You can find my email on About section and send me details.

  • @wipaloengbudnark6800
    @wipaloengbudnark6800 3 года назад

    Do we need to convert gender into dummy variables?

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      Gender is not a dummy variable although it looks like. You should code gender as 1-M, 2--F. For more details watch "How to Code, Report and Interpret Control Variables in SPSS: Example of Gender and Age" on ruclips.net/video/EQq1hTbu-6I/видео.html

  • @corinacrn7684
    @corinacrn7684 3 года назад

    To be frank, I do not understand that well. I need help

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      What you didn't understand? Please mention.

    • @corinacrn7684
      @corinacrn7684 3 года назад

      ​@@ChMahmoodAnwar First of all, thanks for replying. Sometimes a control variable (gender, age, education) is added as dummy, but I do not understand why it is done. Which one is correct? I am confused about it. By the way, could you send an example of control variable reported on the paper, especially with amos or smart? I do not know how to report. Thanks for your time.

    • @ChMahmoodAnwar
      @ChMahmoodAnwar  3 года назад

      @@corinacrn7684 To understand this you need to know why controls are added to the model? Basically, Hierarchical Regression is best to analyze controls, AMOS and SmartPLS were not developed to analyze controls. You can consult Jabbour et al. (2014) to have an idea.