Pearson’s correlation coefficient measures the linear relationship between two or more variables in statistical analysis. The test is concerned with the degree of the relationship between the two or more variables and the direction of the relationship. It does not show the dependent variable and the independent variable in the test, it purely deals with the strength of the relationship between variables under study. The Pearson’s correlation coefficient denoted (r) takes the values in the range of (-1<r<1) (Beck, 1995, p. 325). An evaluation conducted showed the linear relationship between employees’ ages and employee engagement scale score using the Pearson correlation coefficient r test to be -0.66. This indicates the relationship between the employees’ ages and their engagement scale are strongly correlated. The relationship between the two variables has a negative correlation meaning the slope of the curve/line has a negative gradient. This implies an increase in one variable leads to a decrease in the value of the other variable. The result of the test means that an Increase in the employee age results in a corresponding decrease in employee engagement scale score, or a decrease in employee age results in a corresponding increase in the employee engagement scale score.
A lot of care is required when computing the significance of Pearson’s r, when dealing with a small sample the correlation may be misleading not to reach significance, and when dealing with a large sample result may be misleading for being significant for a small correlation. It has been established there is a strong relationship between the employees’ age and the employee engagement scale score, and also there exists a statistical significance between gender and employee engagement. This enables a regression test to be carried out to use gender and age to predict the score of employee engagement. Multiple regression tests try to establish the relationship between two or more predictors and a dependent variable. Multiple regressions allow the researcher to establish the best predictor variable of the test being carried out hence it is a subjective test. Generally, multiple regression problems are solved through a computational approach where the best fit line is plotted of the data set involved. In our case, the linear relationship regression equation line can be generally be represented as y=b0 +b1*x1 +b2*x2; Where y > represents the dependent variable (employee engagement), b0 > represents a regression weight called the constant of linearity (y-intercepts, point at which the line cuts the y axis)
b (1, 2, 3,)> represent the regression weights which are computed in a manner to minimize squared deviation sum.
X (1, 2, 3…)> represents independent variables (age, gender)
Using the above linear relationship equation employment engagement can be determined for a given age and gender. Value (a) has to be computed for the relationship by plotting the best fit line of the data set provided, the point at which the line cuts the y – axis gives the constant of linearity of the equation. The equation can be used to predict any value of employee engagement for a given age and their corresponding gender.
Statistical significance is usually cited with a p-value of 0.05, which helps to assess how to distinguish an observed difference in statistical analysis from random chance. If the computed value of significance level is less than 0.05 we conclude there is a statistical correlation difference between the variables (Beck, 1995, p. 321). This calls for acceptance of the alternative hypothesis at expense of the null hypothesis. Statistical significance does not tell whether the effect magnitude is something to worry about. Substantive significance not only distinguishes the observed values from chance but also tells the effect of the value magnitude on the phenomenon under study.
Reference List
Beck, M. (1995). Data Analysis: An Introduction. New York. Wadsworth Publishing.