hypothesis testing

 A hypothesis is a tentative explanation for something.  


An important feature of a hypothesis is its testability.  As attractive as it may be, if a hypothesis cannot be tested, it is of no immediate use to science.  Hypotheses are not testable if the concepts to which they refer are not adequately defined.  To facilitate clarity in communication, operational definitions offer means of evaluating whether our hypotheses contain scientifically acceptable concepts. 

Hypotheses are also untestable if they are circular.  In this case the event itself becomes an explanation for the event.  For example if you say, "your eight-year-old son is distractable in school because he has attention deficit hyperactivity disorder," the sentence is circular.  Attention deficit hyperactivity disorder is partially defined by the inability to pay attention, so making the above statement simply says that he doesn't pay attention because he doesn't pay attention.

A hypothesis may also be untestable if it appeals to ideas or forces that are not recognized by science.  Science deals with the observable, the demonstrable, the empirical.  To suggest that people who commit horrendous acts of violence are under orders from the Devil is not testable because it invokes a principle (the Devil) that is not in the province of science.

Steps in testing a Hypothesis

1.  Develop research hypothesis  (what you hope to discover)

2.  Set up the null hypothesis (the assumption that the independent variable had no effect...the opposite of what you want to discover)

3.  Construct the sampling distribution on the assumption that Ho is true and collect data

4.  Compare the sample statistic to that distribution (What is the likelihood of obtaining the difference that we did obtain in our experiment if the null hypothesis were true?)

5.  Reject or retain Ho depending on the probability (If the likelihood is small, we reject the null hypothesis and conclude that the independent variable did have an effect.  Outcomes that lead us to reject the null hypothesis are said to be statistically significant*.)

*A statistically significant result means that the difference we have obtained in our experiment is larger than would be expected if error variation alone (i.e. chance) were responsible for the outcome.