HYPOTHESES

A hypothesis is a preliminary explanation that considers a collection of facts and is subject to further examination. In quantitative research, experiments are formulated to evaluate these hypotheses. We collect pertinent data and employ statistical methods to ascertain whether the hypothesis should be tentatively accepted or rejected. It is crucial to recognize that accepting a hypothesis is never absolute, as additional data may surface in the future, which could prompt its rejection (Sukamolson, 2007).

Experiments are conducted to test how introducing an intervention, also known as a variable, affects what happens. Hypothesis testing is used to test variable relationships. It is necessary to control all other factors to ensure that you are measuring the impact of the intervention you have introduced.

Experiments are used in explanatory research based on causal logic, which identifies causal relationships between variables. For instance, A causes B or A causes B under C circumstances. Certain necessary conditions must be present to support the presence of a causal relationship. The cause must precede the effect (temporal order), the reason must be related to the effect, and there must be no alternative explanation.

Explained in terms of variables (Leavy, 2022):

• The independent variable must precede the dependent variable, and a relationship must exist between the two.
• No extraneous variable can provide an alternative explanation for the dependent variable.
• Experimental groups receive the experimental intervention (the experimental stimulus), while control groups do not.
• In some cases, the control group may receive a placebo.
• All experiments have at least one experimental group, but not all experiments have control groups.
• Using control groups is necessary to accurately compare the results of the experimental group whose members received the intervention with those of a similar group whose members did not.
• Depending on the type of experiment, there may be one, two, or four groups in total.
• Some experiments involve pretests and/or post-tests in addition to the experimental intervention.
• A pretest determines a subject's baseline before introducing the experimental intervention.
• A post-test is given after the experimental intervention to assess the impact of the intervention.

When forming a hypothesis, it is essential to identify independent and dependent variables. The hypothesis should be a plausible statement of how the independent variable interacts with the dependent variable. Additionally, potential control variables must be identified.

The next step involves determining how to measure the independent, dependent, and control variables. During the operationalisation process, it is crucial to ensure high content validity between the numerical representation and the conceptual definition of any given concept.

Once the variables are defined and operationalised, the researcher must consider sampling. Which empirical referents will be used to test the hypothesis?

Stockermer (2019) point out that measurement and sampling are typically done simultaneously because the empirical referents that the researcher studies may affect the choice of operationalisation of an indicator over another.

After collecting the data, the researcher can conduct statistical tests to evaluate the research question and hypothesis. Ideally, the study's results will influence theory.

After constructing a set of hypotheses to test the initial theory, the researcher must also pinpoint other variables potentially impacting the phenomenon under investigation. These variables, such as socio-demographic, psychographic, and behavioural factors, should be controlled for in the study. With hypotheses and control variables in place, the researcher can then identify the best methods for measuring both the main variables of interest and the control variables before selecting an appropriate sample for the study.

The term "causation" refers to the idea that a change in one variable will result in another change. In this case, the definition of causation is expanded to include the idea that a precondition can influence a variable of interest. For example, one can imagine that a person's gender influences credit card usage. This means that gender could be seen as having a causal relationship with credit card usage, even though it is impossible to change a person's gender to observe whether credit card usage would change. The term "influence" is sometimes used instead of "cause" if it is more appropriate, but the logic of the analysis remains the same. If two variables are causally linked, then assuming they will be associated is reasonable. If an association provides evidence of causation, then the lack of association suggests that causation is not present. Therefore, an association between attitude and behaviour is evidence of a causal relationship: Attitude --> Behaviour (A. Aaker et al., 2013).

Researchers need to remember the vast array of causal relationships when attempting to establish causality in their studies. This requires implementing various methods and analysis techniques of varying complexity.

Both experimental and non-experimental quantitative studies can observe a comprehensive set of causal relationships. Such relationships may be direct, mediated, or mutual, with some being more complex than others. These complexities can range from simple linear regression to structural equation models (SEMs).

In Figure 7, nine types of causal relationships are depicted:

1. A direct linear causal relationship is one in which Y is a function only of A.

2. A mediated causal relationship is in which the influence of A on Y is mediated by B.

3. A direct causal relationship is one in which it is possible to estimate the total effect (direct and indirect) of A on Y.

4. Direct mutual linear causal relationship in which the influence of A on Y is reciprocal.

5. Indirect mutual linear causal relationship in which Y reciprocally influences the influence of A on Y (mediated by B).

6. Multi-mediated causal relationship (domino) in which A generates a sequential unfolding of effects over time on Y.

7. Moderate direct causal relationship in which the influence of A on Y is contingent on the conditions of C.

8. Causal relationship in which A (exogenous variable) initiates a complex structure (path) of influences (direct and mediated) on Y.

9. Apparent or spurious correlation refers to a statistical association between two variables that do not have a causal link. This type of correlation can arise due to mere chance or the influence of a third variable. It is important to be aware of spurious correlations to avoid incorrect conclusions or erroneous predictions based on statistical data.