Variables relationship

Linear regression is statistical technique for describing and analyzing relationships between a dependent variable and one or, with multiple regression, two or more independent variables. To use regression analysis the variables must be interval - or ratio-scale, meaning they must naturally take the form of numbers (such as income or age). An exception to this is any variable that takes the form of a dichotomy (gender) or a multi categorize variable such as (education) that is collapsed to two categories as ‘less than university’ and ‘some university or more’. Regression analysis is best illustrated in a figure, where a scatter plot shows the relationship between two variances: the percentage of a population that is literate X and the population’s life expectancy Y, each representative country and the location (or coordinates) of each dot is determined by its level of life expectancy (on the vertical axis) and the percentage of the population that is literate (on X). there is a tendency for countries high on X to also be high on Y (there are no dots in the upper left); there is then, a positive relationship between literary and life expectancy. The higher the literacy level is , the longer people tend to live on the average.

Allan Johnson, The Blackwell Dictionary of Sociology - 2000