12. SPSS Classroom - How to Write Research Hypotheses?

A research hypothesis is an “educated guess” expressed as a logically conjectured statement that links variables (or constructs) in a way that can be tested. It matters because it turns a broad review of prior literature into a clear, focused claim about what will be examined. After reading the literature, researchers still need to state what they want to test—what relationship they expect, what difference they anticipate, or what mechanism they believe will operate—otherwise reviewers often push back with questions about the study’s purpose and proposed connections.

Hypotheses come in two core forms: the null hypothesis (H0) and the alternative hypothesis (H1 or HA). The null hypothesis predicts no relationship or no difference between variables, while the alternative hypothesis predicts a relationship or a difference. In business and social research, work typically reports the alternative hypothesis, treating it as the substantive expectation. The two hypotheses are opposites: if one is rejected, the other is supported. The transcript also stresses that a hypothesis must be both testable and predictable—if it can’t be tested or can’t be meaningfully anticipated, it fails as a proper research claim.

To make the distinction concrete, the transcript uses a gender comparison example. If the study examines self-esteem across male and female respondents, the null hypothesis states there is no significant difference in self-esteem between the groups (self-esteem is equal). The alternative hypothesis flips that claim: there is a significant difference in self-esteem between male and female respondents.

The same logic applies to relationship hypotheses. For instance, if the study looks at information quality and perceived usefulness, the null hypothesis says information quality has no significant impact on perceived usefulness, while the alternative hypothesis says it does have a significant impact. The wording can vary—“impact” and “influence” are interchangeable in meaning—so long as the direction of the claim matches the hypothesis type.

Beyond direct relationships, the transcript explains how to write hypotheses when models include moderators and mediators. A moderator changes the strength or direction of an existing relationship. For example, information quality can moderate the relationship between perceived usefulness and benefits by strengthening the positive link when information quality is improved. Conversely, a moderator like knowledge hiding can weaken a positive relationship at high levels of knowledge hiding.

Mediation describes a different mechanism: an effect that travels through an intermediate variable. In a chain where information quality affects perceived satisfaction, and perceived satisfaction affects benefits, the mediation hypothesis states that perceived satisfaction mediates the relationship between information quality and benefits—meaning information quality has an indirect effect on benefits through perceived satisfaction. In all cases where moderation or mediation is claimed, the alternative hypothesis is the relevant one because it asserts that relationships and effects exist, not that they are absent.