Starts with a hypothesis or educated guess formulated so that they are testable. Research design: a method to test hypotheses (e. g. in certain case we could use fmri!, dependant variable: the behaviour influenced by the independent variable, the one. Independent variable: cause or influences behaviour, we manipulate it! we are actually measuring: null hypothesis: rejected if there is a significant relationship between groups. Ways to increase internal validity by minimizing confounds variable (e. g. i am testing a new drugs. We know that some people metabolized in a better way than other, lot faster. And maybe i have this kind of people in my group. I have to minimize this kind of confounded variable): so we use control groups (they receive a placebo for example), random assignment, use analogue models (for example pain) Statistical methods: protect against biases in evaluating data. Statistical versus clinical significance: statistical significance: the results are beyond chance, clinical significance: the results are clinically meaningful.