Hierarchical Multiple Regression in SPSS with Assumption Testing

hello this is dr. grande welcome to my video conducting a hierarchical multiple regression in SPSS hierarchical multiple regression is similar to a multiple linear regression in that we have two or more independent variables or predictor variables and we want to determine how much variance they explain in a single continuous dependent variable the added component of a hierarchical multiple regression as compared to a multiple linear regression is that in H M R you can control for a predictor variable it's similar to an an Cova analysis of covariance where you have the independent variables of interest and then you have a covariant and your concern that covary is explaining the differences on a dependent variable and you want it to determine what the differences are on your independent variables while controlling for that covariant well in h mr is using the same concept except in regression so looking at these fictitious data have loaded in the data view you can see I have a depression score variable an anxiety score variable and these are going to be my two predictor variables and let's assume this family support variable this indicates obstacles to family support so a higher value here would indicate lower family support because we indicate more obstacles and then we have a general symptom scale and again a higher value here we represent more mental health symptoms so we want to see how much variance in the symptoms variable is accounted for by depression and anxiety scores but we want to control for the family support obstacles variable because we know that when there are obstacles to family support that's probably going to cause an increase in symptoms so for this analysis an HMO our would be appropriate we have two predictor variables we have one variable that we want to control for and we have one dependent variable the assumptions for hierarchical multiple regression are the same as they are for a multiple linear regression you want to have 20 cases for each predictor variable and the sample have 100 you want your dependent variable to be normally distributed so check for that I'll go to analyze descriptive statistics Explorer and the dependent variable here is symptom so I'm going to move symptoms over to the dependent list under plots I'm going to check off histogram and normality plots with tests click continue and then click OK I'm looking for a non statistically significant p-value for the Shapiro Wilk and I have that point 5 1 6 so we're going to assume that the dependent variable is normally distributed based on this result we also want to make sure we don't have outliers and we'll check for that as part of the multiple linear regression procedure now in this case course running a hierarchical multiple regression but other than one step where we load blocks of different variables the procedure is identical to a multiple linear regression we're going to check for a linear relationship between the predictor variables and the dependent variable and we're going to check to make sure that the predictors are not multi collinear so to begin the HMR we'll go to analyze regression linear and as you can see this is the same linear regression dialog that you would use for a simple linear regression that's one predictor variable a multiple linear regression which is two or more predictor variables or a hierarchical multiple regression so we'll put symptoms in the dependent variable box and we're going to use two blocks you can see here's as block 1 of 1 we're going to create two blocks here and in the first block and block 1 of 1 as its labeled now we're going to want to put the predictor that we want to control for so in this case it was the family support obstacles and you can see when I drop that in the next button becomes available I click Next and now I have a block 2 of 2 option and in this block I'm going to put the predictor variables of interest which are depression and anxiety so we have now block 1 of 2 that has the family support obstacles and then block 2 of 2 which has the depression and anxiety scores under statistics I'm going to check off r-squared change descriptives part and partial correlations and collinearity Diagnostics as well as case wise Diagnostics estimates is checked off by default and i'm going to leave it checked off could continue under plots the z re s ID goes in the y axis and the z PR IDI goes in the x axis we're also going to check off normal probability plot and click continue under save the only change I'm going to make here to the default is to check off cooks distance and click continue under options I'm making no changes so at this point we're ready to conduct the hierarchical multiple regression click OK and you can see it starts with the descriptive statistics table you can see the mean and standard deviation values for all four variables using the correlations tab we can check for multicollinearity between the predictors so we're looking for any value here that's greater than 0.7 between the predictors that wouldn't be the symptoms variable that would be the other three and we can see that none of these values are above 0.7 we also want to make sure that the predictors correlate with the outcome greater than 0.3 looking at symptoms here we can see family support that's above 0.3 depressions above 0.3 and anxiety is above 0.3 then I'm going to move down I'll come back to model summary I'm going to move down to residual statistics and at cooks distance at this row we want to make sure that the maximum is not above 1 this is at point 1 6 8 so we're good there and we want to make sure the minimum and maximum values for the standard residual or within at negative 3 to 3 so here we're good we're negative 2.9 and 1.9 moving down to the charts for the normal probability probability plot we want these points to be as close to the line as possible and we do have some deviations here but generally these points do follow the line and moving down to the scatter plot we want all these values to be both on the x axis and the y axis between negative 3 and 3 and all the points are so we're good there moving back up the output page and taking a look at the collinearity statistics the tolerance we want the tolerance to be greater than 0.1 and it is and we want the variance inflation factor this value here to be less than 10 and all those values are less than 10 so then moving up the output back to the model summary in the model summary we can interpret the results of the hierarchical multiple regression and we can see that we're given two models models 1 & 2 and we have an R and R square value so if we look at model 2 if we look at this R square value this is saying that 47.7% of the variance in the dependent variable is explained by our predictor variables now that's saying it's explained by all the predictor variables including the family support obstacles variable including the variable that we want to control for the value of R squared change that takes into account our two variables with the family support obstacles variable controlled is here under change statistics R squared change so you can see that the R squared change from model two is 36% so when controlling for the family support obstacles variable 36% of the variance in the dependent variable is explained by the depression and anxiety variables also note here that we have a significant value for both model one and two and looking at mile 1/2 in the ANOVA we have a statistically significant p-value for both of these models this ANOVA test the null hypothesis that the slope of the line is 0 so we want these values to be statistically significant so again to review the output of this hierarchical multiple regression using all the variables 47.7% of the variance the dependent variable is explained by the independent variables and when controlling for the family support obstacles variable 36 and the variance and dependent variable symptoms is explained by the depression and anxiety scores I hope you found this video on conducting a hierarchical multiple regression to be useful as always if you have any questions or concerns feel free to contact me and I'll be happy to assist you

Channel: Dr. Todd Grande

Views: 57,895

Rating: 4.9198542 out of 5

Keywords: SPSS, hierarchical, hierarchical multiple regression, hierarchical regression, multiple linear regression, multiple regression, linear regression, control for variable, covariate, ANCOVA, dependent variable, independent variable, predictor variable, outcome variable, model fit, ANOVA, residuals, r-squared, variance, normality, linearity, outliers, continuous variable, cook’s distance, scatterplot, counseling, Grande, Regression Analysis, Statistics (Field Of Study)

Id: W2HreOD0AQk

Length: 11min 32sec (692 seconds)

Published: Wed Oct 28 2015