Experimental Design and Analysis

You are a training manager for a manufacturing firm and have three new training programs to assess.  You assign, at random, equal numbers of employees to each of the training programs in order to determine which program is most efficient.  How might you go about analyzing the results?

I am trying to decide which test to use?  Here is the information I have below.

First, the type (quantitative or qualitative) and level (nominal, ordinal, interval, or ratio) has much to do with the type of test we select to compare a sample to another sample or a sample to a population.

The hallmark of hypothesis testing is that each test is based on the testing of the null hypothesis.  This is the hypothesis of no difference at all - between the sample and other samples, or between the sample and the population.  If there truly is no difference any difference in means would be attributed to chance instead of treatment.  We accept this hypothesis when the means are similar.  However, the farther they are apart, the less likely this difference occurred because of chance error.  At a certain level of confidence, we feel comfortable in rejecting the null hypothesis and accepting that the difference is a real one.

I suppose I am trying to concentrate on the uses of ANOVA and chi square. ANOVA requires interval or ratio level data.  There is a mathematical relationship between ANOVA and t-tests; in fact if you square t, you end up with F.  

Chi Square can be used when assumptions of normality are violated (highly skewed data) or with nominal or ordinal data.  It's a comparison between what we see and what we expect to see