Alright, so I just ran all of the tests for my upcoming non parametrics exam. HOLY COW, that was a lot of work. The exam assignment was about factors that influence people’s happiness (a sample research). It appears the logistic regression test can predict with a succes rate of 64.7% whether someone will be happy or not in general.

Now look at the tables below that relates to one of the hypotheses. The nominal variable “Happy now?” is the dependent variable, and whether or not you have a handicap or chronic disease the independent variable (also nominal). How do you choose the test? I think it’s by the dependent variable. If it’s nominal and you’re looking to predict one variable from another the test is logistic regresion. In this case, if you work with logistic regression the dependent variable is the “criterium variable” (Happy Now?) and the independent variable is the “predictor variable” (Handicap?). You are looking for how much % more or less the addition of the predictor variable can predict that someone will be happy or not. SPSS pukes up some numbers:

According to this the addition of a chronic disease or handicap results in a 67.2% of succesfully predicting whether someone is happy or not. So, the addition of the predictor results in a 2.5% more chance of succesfully predicting against the 64.7% in general. NAO, the result of this test is significant because in “Variables in the Equation” the column “Sig” says .034, which is 3.4%. Anything below 5% is a significant result. BUT! You have to check the B-value. B = -1.139. That means: a low value on one is equal to a high value on the other. Negative B-value is opposite, positive B-value high/high or low/low. In other words: If you’re high on happiness, you’re low on handicap. If you’re low on happiness, you’re high on handicap. In other words: HYPOTHESIS CONFIRMED.

A guy named “Wald” also has something to do with results being significant or not, but somehow never got why if “Sig” can tell me everything?

Who said I don’t do numbers? WIKIPEDIA? Y U MAKE IT SO HARD TO UNDERSTAND? This:

Hell will freeze over before I understand it. I’m IT man, Big Calculator should do what it’s good at: CALCULATIN’.

Kolmogorov-Smirnov is another one of those tests…. Screw it, at around this time it’s “Da comrad, some vodka“. Whomever said a BSc / MSc Psychology is “fun stuff” is badly mistaking.

p.s. Professor-kuns. Do know I can chart these rules in one A4 so any idiot can understand. Y U SPEND 50 pages on it?