Read the summary and the most important questions on Statistics 2 for IBM | University of Groningen

• 2 Multi regression analysis

  This is a preview. There are 13 more flashcards available for chapter 2
  Show more cards here

• Which assumptions should be  checked for Multivariate Regression Analysis

  Normality, the variables involved should be normally distributed. 

  
  

  Linearity, relationship between DV and each of the IVs should be liner, regression equation should be linear.

  
  

  Homoscedasticity, variance of the dependent variable should not depend on any of the independent variables.

  
  

  Multcollinearity, no linear relationship among independent variables

• regression equation (forward model)

  * look at the coefficients tabel (B)

  * DV = Constant(B) + IV(b) * IV + IV2(b) * IV2 + IV3(b) * IV3 + ....... + e

• Comparing regression models

  * All independent variables at once (Enter method):

  * Forward addition of independent variables (Forward method): 

  
  

• Indicate which independent variable has the highest influence on the dependent variable in the final model (Model 3) of the regression analysis.

  * The standardized coefficient column (Beta column ) is used. 

  * The variable  with the highest standardised coefficient B=0.381.

•  which independent variables have significant contribution to the prediction of the dependent variable

  * Give hypothesis
      H0:bi = 0
      H1: bi ≠ 0

      α = 0.05 

  
  

  for all variables:

  
  

  (Last) Model : 

  IV, statistics (t=4.204), p-value (p=0.000), reject? + intrepretation

  
  

• Explain which model you would select for predicting the dependent variable. Explain which criterion you use.

  * Adjusted R-squared can be used for the comparison of the three models.

  * Choose the model with the highest adjusted R-squared value.

• When use adjusted R-squared?

  Can be used for the comparision of three models.

• Indicate how good the regression model is

  R2 = 0.345, which measn that 34,5% of variance of dependent variable is explained

• Explain which independent variables have a significant contribution to theexplanation of the dependent variable y

  * bi=0

     bi =/0

  * test with different significance levels

  
  

  * look at the coefficients table

  * t-statistics =.... and p-value = .....

• Does multicollinearity provide problems in the found regression model?

  * (look at the coefficients table )

  * all tolerances need to be > 0.10

  * all VIFs < 10