Question

1)A researcher is interested in assessing the impact of a number of changes in a factory on the job satisfaction of workers. Before the changes are implemented the researcher distributes a questionnaire to a sample of workers which measures their attitudes to their work and their overall job satisfaction. The same questionnaire is distributed to the same group of workers one month after the workplace changes were implemented, and again three months later.

(a) Which parametric statistical technique could the researcher use to see if workers’ job satisfaction levels had changed across the three time periods measured? Briefly justify your answer.

(b) What non-parametric technique (if any) could be used to explore this question?

Answer 1

(a) We can assume the three time points to be treatments that influence the job satisfaction. Now the idea is that the three responses we get from a workerfor the three treatments are correlated. Thus we need to assume a worker effect.

b)

We can emply some non-parametric tests. Had the three samples been independent, then we could have ignored the worker effect and then the resultant model would have been a 1-way ANOVA. We could have very easily used the Kruskal-Wallis test for testing equality of treatment effects.

Now, since the samples are not independent (in fact they are the same in the three scenarios), we do not have that option. What we can do is compare the samples pairwise.

We can carry out the above procedure for all the 3 possible pairs.