Towards the end of this report we shall see the effectiveness of this regression standard format. This standard format will work more effectively on the interpretation of the regression analysis. For now we are going to concentrate on Descriptive statistics. We firstly start with the interpretation of our scatterplots that we have generated from the observation of the factors of absenteeism. For a good and clear interpretation these scatterplots we shall include them in the appendix. However, we will briefly interpret their meaning in the body of this report.
Literature review on the economics of absenteeism There are a lot of factors that influence absenteeism in the workplace, could it be the treatment they get from their bosses, the wages they get paid and the general atmosphere at work. In this report based on the practical that we have done we will show some of the factors that contribute to absenteeism in the workplace. Amongst many things that could make people be absent from their works the wages and treatment in the workplace are the leading factors. When employees are paid higher they actually act in two different ways.
It depends on how they are attached/love or how demanding their work is and sometimes the unions protection to employees is the other factor that contributes to the actions that employees take when paid higher. So other employees demand more working hours because an increased payment is an incentive to work harder to earn even more than they are currently. To those people who act in that manner they treat leisure as an inferior good. If leisure is an inferior good, employees consumes less of it as wages increase. The substitution and an income effect move to the same direction.
As a result, if leisure is an inferior good, a wage increase unambiguously causes the hours worked to rise. This then leads to less or decreased absenteeism in the workplace. However, if leisure is a normal good, when wages rises or when unions install more of the regulations to protect employees, employees consume more leisure. The substitution and income effect work towards the opposite directions. So as wages rises, less hours are worked and more leisure is consumed and therefore the labour supply curve is backward bending. This then mean that more employees becomes absent at work.
So when there is an increased volume of absenteeism in the workplace the outputs produces decreases and the economy contract. Absenteeism is also seasonal as well. During the times of Christmas on December the companies loses millions of rands through absenteeism because people are going away to holidays then when there are few workers, the outputs produced is also decreased. A study from the Confederation of British Industry (CBI) reckons that absence from work, for whatever reason, cost firms and public bodies ? 11 billion last year.
So as entertaining as it is to those who do it, it is no joke to profit making companies and it cost them fortunes. The higher the absenteeism for whatever reason, the lower the output and the lower the profit. This could be even dangerous to employees as well because when the profit is contracting they might be retrenched from work. Descriptive Statistics Analysis The separate XY scatterplots, in the Appendix, show that the relationship between Y and all the X’s is consistent with economic theory although all the relationships appear to have very weak linear relationships.
The average employee wage (X2) as well as the percentage of part time employees in a company (X3) shows weak negative non- linear relationship, and the percentage of unionized employees in a company (X4) shows a weak positive linear relationship. As the average employee wages (X2) increase the incentive to work harder decreases. Employees earning lower wages tend to take fewer absent days and work harder than those earning higher wages. As the percentage of part time employees (X3) increase in can be seen, in the scatterplot, that the average number of days absent per employee (Y) decreases.
Part time workers have more of an incentive to work harder than full time employees, therefore it is expected that average number of days absent per employee (Y) should decrease as the percentage of part time employees increases. As the percentage of unionized employees in a company (X4) increases it can be seen, from the scatterplot, which the average number of sick days per employee increases as well. Unionized employees tend to strike more than those not in a union, therefore it is to be expected that as the percentage of unionized workers increase the average number of days absent per employee increases as well. Frequency Table
The matrix does not show any possible signs of multi-colinearity meaning there is no high correlation between two independent variables. There is however a positive relationship between the percentage of part time employees and percentage of unionised workers in a company but it is very weak. Elasticity The elasticity which is in the Appendix shows the number of days absenteeism per employee is both income and cross price inelastic.
Or holding other things constant, as the number of days absent increases by 1 percentage point, on the average, the part time employees’ decreases by 11 percentage points. Therefore there’s a negative relationship between the average number of days absent per employee and the part time employees in a company A positive relationship prevails between unionized employees in a company and the average number of days absent per employee. Holding all other things constant, as the number of days absent per employee increases by 1 percentage point, on the average, the percentage of unionized employees’ increases by 6 percentage points. 0. 05985 (X4) measures the percentage of unionized workers in the company.
So what this figure means is that the percentage of absenteeism for unionized employees increases by 0. 05982. D5 If there is shift work available in a company then we can say that the average number of days absent per employee will increase by (10. 265+1. 562) = 11. 827 units but if there is no shift available in a company then the average number of days absent per employee will increase by 10.26 units only because 1. 562 would be multiplied by (0), holding all other things constant.
If union management relationship is good then the average number of days absent per employee will decrease by 7. 62 units but if the union management relationship is not good then the average number of days absent per employee increases by 10. 26 units because 2. 6366 is multiplied by (0), holding all other things constant. All the coefficients are statistically significant or they are different from each other as shown it the table at the ANOVA Excel output in the appendix.
The p values in summary shows this below. p = (0. 000)* (0. 000)* (0. 000)* (0. 000)* (0. 002)* (0. 000)* *p value below or statistically significant at the 5% level **p value greater than or statistically insignificant at the 5% level, not applicable in this report As indicated above that a single stared p value (*p) means that the coefficients are statistically significant and those which are double starred (**p) are statistically insignificant. To measure significance in this report we have used ? = 0. 05 or 5%. We are not interested in other percentages except that of ?
None of the coefficients are statistically insignificant as can be shown with the key above and the ANOVA Excel output in the appendix. The following are the actual values of p from ANOVA Excel output uncut. p = (8. 11681E-14) (1. 43035E-07) (0. 000471097) (5. 38289E-06) (0. 002496639) (5. 9905E-07) These are very small values even the one for D5 is still less than ? = 0. 05 or 5% as ? 0. 24% At ? = 0. 05 or 5% F stat is also significant as its F = 21. 40086746 and Significant F is 3. 08395E-14 of which is a very small number 0. 00000000000003.
This value is provided The R2 = 0. 5323484, that is the percentage of the total variation in the dependent variable Y explained by four explanatory variables X2, X3, X4 and dummy variables D5 and D6. The result of R2 is nowhere closer to 1 which indicates the poor fit of the fitted sample regression. Adjusted R2 = 0. 507473315 which came about 1 – (1 – 0. 5323484)*(99/94) this amount is lower than the original R2 as is always is. This is as a result of the added explanatory terms, but it is increasing if the absolute t value added is greater than 1 so is R2 but stay lower than R2.
This is the measure of the goodness of fit that is adjusted for (i. e. takes into account explicitly) the number the number of explanatory variables both qualitative and quantitative in the model. Collectively all the variables in the model are statistically significant since the p value of the computed F value of 21. 40 is extremely low. The null hypothesis for Ho: Bi= 0 and for i= 2,3,4, for a two tailed test, we can reject the null hypothesis since the p values of all the variables are less than 5%. Each individual variable has an extremely low p value.