EXPLAINING - REGRESSION MODEL

 

The Explaining - Regression Model procedure shows the degree to which the values of a variable called "dependent variable" might be explained by a given set of variables called "independent variables". This procedure is based on the multiple regression model of the statistic theory. If theory and experience indicate cause-effect relationships between certain variables, the regression model can give a measure of those relationships. Moreover, the variables having more explanatory power could be used as policy-based directions of interventions.

Example: Explaining the variance of the indicator %MWRA_02_MODERN (percentage of married women of reproductive age using modern contraceptive methods). As explaining factors  5 indicators were chosen, as listed below:

Variable Description
 %FAM_POV_INCIDENCE  %Families with annual PC income under per capita poverty threshold
 ESTIM_MIDWIVES/10,000WRA  Estimated midwives for 10,000 WRA ratio
 %MWRA_GAIN_EMPL_PRV  %MWRA gainfully employed in private of tot WRA
 %FAM_EXP_HEALTH_1997  % of family expenses for health in 1997
 POP_DENSITY_2000  Density of population - inhabitants per square km

 

·         The bar in the Explain window shows the percentage of variation in the dependent variable that is explained by the variation in each independent variable (the R squares in the statistical theory), as well as the percentage of variation explained jointly by all independent variables. For example:

  1. %FAM_POV_INCIDENCE explains 27.16%, 

  2. ESTIM_MIDWIVES/10,000WRA explains 7.58%,  

  3. %MWRA_GAIN_EMPL_PRV explains 34.68% (meaning the most explanatory variable)

  4. ... 

Together, the five variables explain 68.35% of the variance of %MWRA_02_MODERN (percentage of married women of reproductive age using modern contraceptive methods). 

·         The multiple regression equation is displayed in brown at the bottom of the Chart.

 For practical purposes, the chart uses color-coding:

    - Between 0 and 12.5 % of variation explained the bar colors in red.

    - Between 12.5% and 25%, of variation explained the bar colors in yellow.

    - Over 25% of variation explained the bar colors in green.

 The Diagram shows a more suggestive picture of the linkage.

The thicker the line, the greater the level of explanation of the independent variable.

Close Window