Testing For Multiple Regression Part 2

Image result for multiple regression analysis"

RECODE Q46A (99=0) (ELSE=1) INTO Democracy10.

VARIABLE LABELS  Democracy10 ‘Satisfaction with democracy’.

EXECUTE.

REGRESSION

  /MISSING LISTWISE

  /STATISTICS COEFF OUTS R ANOVA COLLIN TOL

  /CRITERIA=PIN(.05) POUT(.10)

  /NOORIGIN

  /DEPENDENT ADULT_CT

  /METHOD=ENTER Q1 Democracy10

  /SCATTERPLOT=(*ZRESID ,*ZPRED)

  /RESIDUALS DURBIN HISTOGRAM(ZRESID)

  /SAVE COOK ZRESID. 

Variables Entered/Removeda
Model Variables Entered Variables Removed Method
1 Satisfaction with democracy, Q1. Ageb . Enter
a. Dependent Variable: ADULTCT: Number of adults in household
b. All requested variables entered.
Model Summaryb
Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson
1 .057a .003 .003 2.465 1.391
a. Predictors: (Constant), Satisfaction with democracy, Q1. Age
b. Dependent Variable: ADULTCT: Number of adults in household
ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression 999.038 2 499.519 82.228 .000b
Residual 309929.057 51019 6.075    
Total 310928.095 51021      
a. Dependent Variable: ADULTCT: Number of adults in household
b. Predictors: (Constant), Satisfaction with democracy, Q1. Age
Coefficientsa
Model Unstandardized Coefficients Standardized Coefficients t Sig. Collinearity Statistics
B Std. Error Beta Tolerance VIF
1 (Constant) 3.198 .048   66.850 .000    
Q1. Age 4.055E-005 .001 .000 .054 .957 .995 1.005
Satisfaction with democracy .496 .039 .057 12.794 .000 .995 1.005
a. Dependent Variable: ADULTCT: Number of adults in household
Collinearity Diagnosticsa
Model Dimension Eigenvalue Condition Index Variance Proportions
(Constant) Q1. Age Satisfaction with democracy
1 1 2.846 1.000 .01 .02 .01
2 .121 4.854 .01 .65 .28
3 .034 9.183 .99 .34 .71
a. Dependent Variable: ADULTCT: Number of adults in household
Residuals Statisticsa
  Minimum Maximum Mean Std. Deviation N
Predicted Value 3.20 3.70 3.65 .140 51022
Std. Predicted Value -3.239 .329 .000 1.000 51022
Standard Error of Predicted Value .011 .057 .017 .008 51022
Adjusted Predicted Value 3.19 3.70 3.65 .140 51022
Residual -2.698 50.304 .000 2.465 51022
Std. Residual -1.095 20.410 .000 1.000 51022
Stud. Residual -1.095 20.410 .000 1.000 51022
Deleted Residual -2.700 50.306 .000 2.465 51022
Stud. Deleted Residual -1.095 20.494 .000 1.000 51022
Mahal. Distance .096 26.551 2.000 3.371 51022
Cook’s Distance .000 .008 .000 .000 51022
Centered Leverage Value .000 .001 .000 .000 51022
a. Dependent Variable: ADULTCT: Number of adults in household

What is your research question? Answer: how can we determine whether real assumptions exist between number of adults in the household, their age and their satisfaction with the level of democracy today?

Using Afrobarometer data set (IMB SPSS Statistics 21, n.d.), I have created a dummy variable using interval/ratio variable, labelled as level of democracy:  today. The categorical dummy variable tells us two things: you are either satisfy by your level of democracy today or not satisfy by your level of democracy today. If you elect 1, you are satisfied, and if you elect 0, you are not satisfied.

The coefficient table tells us more information about individual independent variables. Another important consideration to look into is the variance inflation factor (VIF). VIF is the number that shows the level of severity of multicollinearity in an ordinary least-squares regression analysis (Warner, 2012)— values within 10 and above 10 indicate serious multicollinearity or high probability of correlation in the model (Wagner, 2016).  However, 1.005 for both the predictors indicate normal level of correspondence or assumption. In my Model Summary table, the Durbin-Watson statistic, which tells us about the independence of errors (Laureate Education, 2016j), is showing a value of 1.391. This value is an example of an absolute absent of correlation between the residuals (Laureate Education, 2016j).

The Anova table shows the overall statistical significant of the calculated variables. In this case, we have a statistical significant of 0.000, indicating the rejection of the null hypothesis when conventional P-value as set to P<0.05.Our Cook’s distance shows an unnecessary relationship on the model ranging from 0.0- 0.008, with value of 1.0 or greater showing possible influence of correlation.  Our scatter plot provides strange and slightly uniform display of homoscedasticity, which give some linearity slip details in the relationship of the variables. However, the histogram indicates how the distribution of correlation or no errors exists (Wagner, 2016). Looking at the histogram, the distribution display of the frequency and regression standardized residual shows an insignificant deviation from normalcy.

 In terms of the positive implication for social change and after analyzing and reviewing all tables and data of the applicable variables, there seem to exhibit little possible violations on the assumptions of the resulted data. Therefore, majority of assumptions were possibly made. I have used the assumption, under the SPSS value label of interval ratio (i.e. the level of democracy: today) to conclude the notion that if you don’t know or never heard about democracy, we will assume that you are not satisfy, or have not specifically participated in the democratic process. While all others who refused or kept quit means you are satisfy. This tells us the basic concept of creating dummy variable; that is the provision or creation of predictor variable (categorical variable) that answers the question of yes or no about group participation (Warner, 2012). The assumptions of my resulted data, tell us how implication for positive change relies on how community involvement can increase and strengthen transparency in democratic process. Last year, I attended a 3-day workshop in the city of DeKalb, Iowa State. The mission of the workshop was to increase public awareness and community involvement in politics and democracy. The speeches were given by famous politicians on how community participation on democratic process could yield positive outcome in electoral process or electing apparent leader.

Reference

      IMB SPSS Statistics 21. (n.d.). Afrobarometer [Data file]. Retrieved from Walden University my student Account.

        Laureate Education (Producer). (2016j). Regression diagnostics, model evaluation, and dummy variables [Video file]. Baltimore, MD: Author.

       Wagner, W. E. (2016). Using IBM® SPSS® statistics for research methods and social science statistics (6th ed.). Thousand Oaks, CA: Sage Publications

      Warner, R. M. (2012). Applied Statistics from bivariate through multivariate techniques (2nd ed.). Thousand Oaks, CA: Sage Publications.