 Students, now we are going to again perform multiple linear regression with forward or backward method. Before performing this method, we have studied two methods, stepwise and enter method. Let's see how do we do multiple linear regression with a backward or forward method. We are not going to do multiple linear regression again, and here we will change the method stepwise and we will go on the forward method. The rest of our things will be the same, our variables are also the same, we have changed the method, the options are also the same, we will go to OK. If we go to OK, then we will see that we have five equations, and the results of the five equations are the same, and if we look at the bottom, then we have the same ANUVA, all of them are significant, because the variables are the same, and here we have five coefficients and the models are also the same. So, in this, the backward method and the stepwise method, what is the difference? This is almost the same analysis as we did in stepwise. There is a theoretical difference between the two. In the stepwise method, we opt when we are not able to decide which variable is the major predictor and which variable is the least predictor. In the forward method, we know theoretically that which is our number one variable and which is our number five variable, i.e. which is our predictor and which is your least predictor. If we go to the backward method, and then go to the regression, and then go to the backward method, then we will know that you have three equations, one, two, and three. The least predicted one has been removed from the model. There are only three that are actually predicting your dependent variable, and it has also checked the three, and it has made three blocks of coefficients in which it has shown you that these two are insignificant. So in this way, based on our theoretical stances and our requirements, we can see which method suits us in multiple linear regression. While doing multiple linear regression, I will take you again on that. We have one more thing that we can make blocks of variables. For example, these three, these five variables were related to motivation. Now I will make a next block, and in that I say that I want to see the use of our variables related to the use of Facebook, we take the intensity of the use of Facebook. In that we take the intensity of use of Facebook and Facebook relationship containers behavior. Along with this, we add the number of Facebook accounts, duration of Facebook accounts, average time spent on Facebook account. We add this, and in the third block, we add our age, GPA, we add this. And the rest, we first look at the results with the enter method. The rest will stay the same, our options are okay. Because we have added new variables, so we have the normality plot is okay, the scatter plot is also okay about your homosinacity, and if we look here, then we have a big correlation, doesn't matter, between IVs, let's move on. Now look here, model 1, 2, 3. Now what he has done is that he has told you which predictors are the first, and which predictors are the second, and which predictors are the third. If he had told us three blocks, that we have data in three blocks, then we have three equations. Our scale is 17.7%, the second is 17.8%, and the third is 17.7%. If the variables are added, but actually, the variables of our first model, the beta of that is distributed, and the variables of the third level are added. And this is theoretically correct, because NOVA is everyone's significant, it means that the third added, it was actually their poor relation with the first variables. So that's what he has explained. So look here, in model 1, we find that a lot of variables are insignificant. And in model 2, we also find that a lot of variables are insignificant. Same is the case with model 3 as well. Now because we had opted for the enter method, it doesn't exclude the insignificant variables. So that's why it will come in this way. Now if we take this, linear, which we took stepwise, then we will do it stepwise. So you have, look here, four equations. And the unknowns of these equations are also there. And here, if you look, in its coefficients, the insignificance is also there. So basically, it's up to you which design you have to use. Because this is multivariate analysis, your theory is driven by multivariate analysis. You have to decide which variable you have to use in regression, multiple linear regression. Which analysis you have to use in multiple linear regression. So for this, you have to study literature. You have to study theory, previous studies. That's why you are in this position that you should go with A priori design and do your analysis accordingly.