Formulating the Research Hypotheses

Chapter 7. Hypothesis Testing: Introduction to Hypothesis Testing (p-value Approach)

Formulating the Research Hypotheses

The first step of any statistical test is to formulate the research hypotheses.
#\phantom{0}#

Hypothesis

In the context of statistical testing, a hypothesis is an expectation or prediction about a population characteristic.

#\phantom{0}#
Every statistical test has two hypotheses: a null hypothesis and an alternative hypothesis.
#\phantom{0}#

Null Hypothesis

The null hypothesis of a test, denoted #H_0#, represents a zero-effect. That is, any observed pattern in the sample data is purely due to chance.

Generally speaking, the null hypothesis is the hypothesis that you are trying to disprove.

The null hypothesis is the statement about the population that is being tested. As such, the outcome of a hypothesis test always leads to one of two decisions:

The null hypothesis is rejected.
The null hypothesis is not rejected.

A null hypothesis should only be rejected if it is highly unlikely to be true given the observed sample data.

An important caveat is that failing to reject the null hypothesis does not mean the null hypothesis is accepted to be true. It simply means that there is not enough evidence to disprove it.

Examples of null hypotheses are:

There is no treatment effect
There is no difference between groups
There is no relationship between variables

#\phantom{0}#

Alternative Hypothesis

The alternative hypothesis of a test, denoted #H_a#, predicts that there is an effect. That is, any observed pattern in the sample data is not the result of random variation, but rather a reflection of the underlying population's characteristics.

Generally speaking, the alternative hypothesis is the hypothesis you suspect to be true.

The alternative hypothesis cannot be tested directly. It is accepted if and only if the null hypothesis is rejected.

The null hypothesis and the alternative hypothesis are mutually exclusive, meaning there cannot be any overlap in the predictions they make.

Examples of alternative hypotheses are:

There is a treatment effect
There is a difference between groups
There is a relationship between variables

#\phantom{0}#

Resemblance to a Criminal Trial

Hypothesis testing bears a resemblance to a criminal trial. In court, the burden of proof is on the prosecutor, meaning that a suspect is assumed to be innocent until proven guilty.

Likewise, in hypothesis testing, the null hypothesis is assumed to be true unless very strong evidence is presented that suggests otherwise.