In what instances would you choose the stepwise method over other types of multiple regressions? I am trying to identify when one method is preferred over another.
Thanks a lot for making it easily understandable Dr. Grande!Can you also tell me how to interpret the impact of a binary categorical variable ? For example , if Gender value is entered as 0 and 1 for males and females respectively, how to calculate the impact on the DV?
Thank you for the great video! Just one question, you mentioned that one of the assumptions of the model is that the sample size is greater than 20. So, if you have a sample size that is less than 20 is there an alternative method that you could use which will give the same sort of output.
Thanks! Great clear lecture. Is stepwise method always the best for multiple regression analysis? Also, is it possible to compare the results of various multiple regressions with respect to the same dependent variable?
My adjusted R square is very low even though the model includes three sig predictors. My predictors are on a 5-point Likert scale, while my criterion is 0 % - 100%. Does that matter? The model ends up being beautiful when I make the criterion the Adjusted Predicted Value, but I guess that would not be allowed. Please help me. Thank you.
Very beautiful explanation. I am working with F4 and F5 generations of a crop.Can I use the yield associated traits in the F4 generation as predictors for grain yield in the F5 generation?
I am working in a model with crop yield as the dependent variable and weather parameters as the independent variable. When I use the Step-wise Regression method then only one predictor variable is selected in the equation with an adjusted R square value of 0.62. But if I change the probability of F entry from the default value of 0.05 to 0.12 and probability of F Remove from default value of 0.10 to 0.15 then i get a model with 7 predictors and an Adjusted R-square value of 1.00. So, what should I do? Which model Should I consider as more appropriate?
I agree with Ivy below; under what circumstances would you choose to use the Stepwise method? Also, what's the primary difference between a linear regression and a multiple regression?
Stepwise is used a lot for omnibus hypothesis. You always do one DV and multiple predictor variables in the model, but you can do multiple DV and IV tests... you just have to run them separately. You would want to use stepwise when you want the tools algorithm to tell you if the model is significant, by the same time removing predictors from the model. Linear Regression is taking one DV and IV and drawing a straight linear line until their paths cross or called an intercept. The intercept is where we "predict" the outcomes. The movie "The Martian" has some good scenarios to see this in action when the they test the speed of the Ship to intercept to the speed of getting Matt Damon into the ship. The intercept was the point paths would cross and in their case they were too fast and needed to slow down.
Hi Todd, Thanks, this is an amazing and helpful video, I wanted to ask some questions please: if I have a scatter plot with the points outside my maximum and minimum values, what does that imply. Also the point on my normal plot are inverse to the line, they are not close to the line as yours in the video above, what does that imply as well. Thank you.
thanks, but my problem is how to know what is going to be my dependent variable while I have many categorical variables and (very high,high,average, small, very small) represented by (5,4,3,2,1) respectively. I want to predict the impact of delay using regression model. but i don't know how I am going to formulate my dependent(predictive)variable. Please help me.
Guys very new to using SPSS I was just wondering to run a stepwise multiple regression can we use a combination of nominal and scale predictors or just scale predictors for our Independent variables?
Is this the best method regression to use when dealing with survey data that contains likert scales? All of the variables I am interested have the same scale (Strongly Agree-Strongly Disagree).
My favorite test and you look like a boss. Many arguments support and disagree with the methods. For support, you take out the researcher human element and the system adds/removes variables. Disagreements in the field is that the researcher is needed for inclusion/exclusion of criteria for statistical power to have the highest fidelity. In other words, the more variables the lower statistical power in terms of the stepwise regression adding or removing for model fit.
Thank you for this video. How many independent variable is possible to use for stepwise regression. I have 16 independent variables, can I use them all for the stepwise lineair regression?
Dear Dr. Grande, Would you please update (e.g., 2021, 2020, 2019, etc.) this tutorial video? I need your permission to add a few of your tutorial videos to my final dissertation. Best regards, Hossein Divan-Beigi, M.D., Ph. D.-Student
There is an advanced multivirant statistics book that helps. Pretty much you want to test for multicollienarity b/c with stepwise the system does the work. So, highly correlated variables are going to be in the outcomes with looking at p-values. Kinda pointless in a sense of the word. Moreover, you want to focus on data distribution by using power analysis for statistical sample size requirements for a MR study, and you'll want to focus on what Vasti below states.
Hi Todd, I simply owe every single article I have published to you. Thank you so much.
Thank you so much for the wonderful work Dr. Todd Grande
In what instances would you choose the stepwise method over other types of multiple regressions? I am trying to identify when one method is preferred over another.
Very informative Explanation. It would be nice if we had more of them.
Thanks a lot for making it easily understandable Dr. Grande!Can you also tell me how to interpret the impact of a binary categorical variable ? For example , if Gender value is entered as 0 and 1 for males and females respectively, how to calculate the impact on the DV?
Thank you for your easiest lecture and very enjoyable session too.
You are welcome - thanks for watching -
Thank you for the great video! Just one question, you mentioned that one of the assumptions of the model is that the sample size is greater than 20. So, if you have a sample size that is less than 20 is there an alternative method that you could use which will give the same sort of output.
Thanks! Great clear lecture. Is stepwise method always the best for multiple regression analysis? Also, is it possible to compare the results of various multiple regressions with respect to the same dependent variable?
My adjusted R square is very low even though the model includes three sig predictors. My predictors are on a 5-point Likert scale, while my criterion is 0 % - 100%. Does that matter? The model ends up being beautiful when I make the criterion the Adjusted Predicted Value, but I guess that would not be allowed. Please help me. Thank you.
Very beautiful explanation. I am working with F4 and F5 generations of a crop.Can I use the yield associated traits in the F4 generation as predictors for grain yield in the F5 generation?
I am working in a model with crop yield as the dependent variable and weather parameters as the independent variable. When I use the Step-wise Regression method then only one predictor variable is selected in the equation with an adjusted R square value of 0.62. But if I change the probability of F entry from the default value of 0.05 to 0.12 and probability of F Remove from default value of 0.10 to 0.15 then i get a model with 7 predictors and an Adjusted R-square value of 1.00. So, what should I do? Which model Should I consider as more appropriate?
I agree with Ivy below; under what circumstances would you choose to use the Stepwise method? Also, what's the primary difference between a linear regression and a multiple regression?
Stepwise is used a lot for omnibus hypothesis. You always do one DV and multiple predictor variables in the model, but you can do multiple DV and IV tests... you just have to run them separately. You would want to use stepwise when you want the tools algorithm to tell you if the model is significant, by the same time removing predictors from the model. Linear Regression is taking one DV and IV and drawing a straight linear line until their paths cross or called an intercept. The intercept is where we "predict" the outcomes. The movie "The Martian" has some good scenarios to see this in action when the they test the speed of the Ship to intercept to the speed of getting Matt Damon into the ship. The intercept was the point paths would cross and in their case they were too fast and needed to slow down.
Hi Todd,
Thanks, this is an amazing and helpful video, I wanted to ask some questions please:
if I have a scatter plot with the points outside my maximum and minimum values, what does that imply.
Also the point on my normal plot are inverse to the line, they are not close to the line as yours in the video above, what does that imply as well.
Thank you.
Such a great video. This helped me doing my statistics :)
I am glad you found it helpful, thanks for watching -
thanks, but my problem is how to know what is going to be my dependent variable while I have many categorical variables and (very high,high,average, small, very small) represented by (5,4,3,2,1) respectively. I want to predict the impact of delay using regression model. but i don't know how I am going to formulate my dependent(predictive)variable. Please help me.
Guys very new to using SPSS I was just wondering to run a stepwise multiple regression can we use a combination of nominal and scale predictors or just scale predictors for our Independent variables?
Can I kindly request the data file you are using here for me to practice what I have learned on this particular video? Thank you very much.
Is this the best method regression to use when dealing with survey data that contains likert scales? All of the variables I am interested have the same scale (Strongly Agree-Strongly Disagree).
The Stepwise component seems like a really nice feature in SPSS.
My favorite test and you look like a boss. Many arguments support and disagree with the methods. For support, you take out the researcher human element and the system adds/removes variables. Disagreements in the field is that the researcher is needed for inclusion/exclusion of criteria for statistical power to have the highest fidelity. In other words, the more variables the lower statistical power in terms of the stepwise regression adding or removing for model fit.
the only big big drawback of the stepwise is that stepwise is introducing with each IV 5% chances of Type I error in the model.
how do I take the resulting coefficient and apply it to the raw score prediction formula???
how cal I CONTROL effect of linearity in stepwise regression please ?
Thank you for this video. How many independent variable is possible to use for stepwise regression. I have 16 independent variables, can I use them all for the stepwise lineair regression?
hi, did you use 16 variables for your regression analysis? i know this was 5 years ago but still
is there any drawback in model when we use many independet variables?@@efefmichelle
hello,how can i use 2sls in SPSS? please help!!
and i have the same problem as luvemel
Dear Dr. Grande, Would you please update (e.g., 2021, 2020, 2019, etc.) this tutorial video? I need your permission to add a few of your tutorial videos to my final dissertation. Best regards, Hossein Divan-Beigi, M.D., Ph. D.-Student
Thank you very much it's excellent but some views were hazy
Thanks for the video :))
The volume is pretty low :/
good explanation ! but sorry, your sound is to low. i can't hear well. Please CHECK before you upload. thx
Again having a harder time in these videos determining the exact way to test for assumptions. Seems less clearly defined at this point.
There is an advanced multivirant statistics book that helps. Pretty much you want to test for multicollienarity b/c with stepwise the system does the work. So, highly correlated variables are going to be in the outcomes with looking at p-values. Kinda pointless in a sense of the word. Moreover, you want to focus on data distribution by using power analysis for statistical sample size requirements for a MR study, and you'll want to focus on what Vasti below states.