Hypotheses

Research Hypothesis

A RESEARCH HYPOTHESIS is a clear, specific, testable statement or prediction about the relationship between two or more variables (i.e., independent variable X explains variation in dependent variable Y).  The research hypothesis guides the direction of the study and outlines what the researcher expects to find.  A good hypothesis is:

• based on existing knowledge (i.e., theory drives your predictions)
• FALSIFIABLE (i.e., can be proven false if it is incorrect)

Some hypotheses are directional, positing that as values of one variable increase or decrease, values of the other variable increase or decrease:

• There is a POSITIVE RELATIONSHIP between population density (independent variable) and crime rates (dependent variable) — as population density increases, crime rates increase
• There is a NEGATIVE RELATIONSHIP between education level (independent variable) and teen pregnancy (dependent variable) — as education increases, teen pregnancy decreases

These types of hypothesis are appropriate when working with variables that have direction (i.e., ordinal-, interval- or ratio-level variables).

Some hypotheses merely posit that there is a relationship between two variables:

• Women are more likely to vote than men (respondent sex is the independent variable) to vote (dependent variable) — there is a relationship between sex and voting

This type of hypothesis is common when working with nominal variables.  Because nominal variables have no direction, the relationship between a nominal variable and another variable cannot be stated in directional terms. Instead, the hypothesis should specify the type of relationship between the variables in terms of how differences in the dependent variable are linked with differences in the independent variable.

If, based on theory and existing knowledge, there are control variables that further explain variation in the dependent variable, they should be included the research hypothesis: 

• if all variables involved have direction (i.e., ordinal-, interval-, or ratio-level variables), you would simply add the phrase “while controlling for” and then specify the control variables
• if the control variable is nominal, you should specify the expected relationship

The research hypothesis is presented as H₁.  If you have more than one research hypothesis, they would be presented as H₁, H₂, H₃, etc.

Null Hypothesis

Null means having no value.  By extension, the NULL HYPOTHESIS is a statement that there is no relationship between the variables being studied. While the research hypothesis serves as a foundation for conducting empirical research by guiding the direction of the study and outlining what the researcher or administrator expects to find, the null hypothesis forms the basis of hypothesis testing.  The null hypothesis is presented as H₀.

Alternative Hypothesis

An ALTERNATIVE HYPOTHESIS is proposed as an alternative to the null hypothesis; it indicates that there is a relationship between the variables being studied. Alternative hypotheses can be:

• directional (one-tailed), specifying a direction of the effect or difference
• non-directional (two-tailed), merely stating that there is a difference, without specifying the direction

Research hypotheses and alternative hypotheses are conceptually similar but serve different functions: a research hypothesis’s broader context is more aligned with scientific inquiry and theory testing, whereas an alternative hypothesis is specifically formulated for statistical testing against the null hypothesis. The alternative hypothesis is presented as H_a.