Part 2: COMPONENTS OF QUANTITATIVE RESEARCH DESIGNS

Comprehending the role and significance of research designs is crucial for effective research. The design encompasses the entire research process, from framing the question to analysing and reporting data.

Two fundamental research questions exist: descriptive research, which explores what is happening, and explanatory research, which focuses on why things are happening.

Descriptive research can be advantageous, especially when exploring new areas, as it can provoke "why" questions for explanatory research. Explanatory research involves developing causal explanations that argue that a specific factor affects a particular phenomenon. For instance, gender may affect income levels. However, the complexity of causal explanations may vary, and hidden or unmeasured variables may be at play.

It is important to note that people often mistake correlation for causation. When two events are linked, it does not necessarily imply that one causes the other. The link between them may be coincidental rather than causal. Therefore, it is crucial to understand the distinction between correlation and causation to conduct effective research.

Aaker et al. (2013) organise the process/design of a study as shown in Fig. 6. It all starts with specifying the Research Question, that is, the problem that the project will try to solve and the knowledge to which it will contribute or initiate.

Directly resulting from the literature review, it must immediately be "translated" into research questions, that is, hypotheses that will determine what will be measured, from what sources of information and with what methodologies. Research hypotheses are systems of variables whose sets, although only partially exhaustive, cover the main dimensions of the phenomenon under analysis. They also clarify the relationships proposed between such variables that need testing. With this, the following very relevant and demanding task is operationalising (making measurable) the variables whose relationships will be tested (scales).

Once the research question has been specified, the concepts (variables), latent or directly observable, have been defined and whose relationships will be tested, and the measures that capture them have been specified, it is essential to determine which units of information will contain the required information (secondary or primary).

Quantitative studies (experimental/non-experimental) must also define the sampling method (random/non-random) that will be applied to this "population" and the size and characteristics of the groups (non-experimental; experimental; control) that will be heard.

With this knowledge, the researcher must decide which concrete information collection plan to adopt: correlational/survey (cross-sectional; longitudinal) or experimental.

Collecting information (questionnaire) is complex, susceptible to adding "errors," and dependent on the researcher's experience. For all these reasons, it is advisable to use scales that have already been validated in previous studies whenever possible, reinforcing their reliability and validity.

Once organised information is available, the data will be subjected to adjusted and planned analyses to test the research hypotheses (descriptive, univariate, multivariate, inferential). The results obtained must then be described and interpreted to, in conclusion, be "converted" into an answer(s) to the initial Research Question that triggered the entire process.

Understanding whether a research question is descriptive or explanatory is crucial as it significantly affects the research design and the information collected. Researchers must develop causal explanations when answering the 'why' questions. Causal explanations attempt to prove that a particular factor X, such as gender, affects a phenomenon Y, such as income level. While some causal explanations can be simple, others can be more complex.

When making predictions, researchers must distinguish between correlation and causation. It is a common mistake to assume that two events are causally related just because they occur together or one follows the other. The correlation is likely coincidental and does not indicate a causal relationship.

Distinguishing between causation and correlation is essential to accurately understanding prediction, causation, and explanation. It is important to note that an accurate prediction does not always require a causal relationship, and the ability to make a prediction does not necessarily prove a causal relationship. Confusing these concepts can lead to a lack of understanding and incorrect conclusions.

Recognizing the difference between correlation and causation is essential because we can observe correlation but directly observe causation. Therefore, we must infer the cause, making avoiding invalid inferences a primary goal of explanatory research design.

There are two approaches to causation: deterministic and probabilistic. In deterministic causation, variable X causes Y without exception if it reliably produces Y. This approach aims to establish causal laws, such as the rule that water boils at 100ºC.

However, most causal thinking in social sciences is probabilistic rather than deterministic. We can enhance probabilistic explanations by specifying the conditions under which one factor is more or less likely to affect another. However, we will never achieve complete or deterministic explanations. Two events are causally related because they occur together or follow the other. The correlation is likely coincidental and does not indicate a causal relationship.

Mistaking causation for correlation can lead to understanding prediction, causation, and explanation. Accurate prediction does not necessarily require a causal relationship, and the ability to predict does not prove causality.

Research objectives and questions can be single or multiple and can be covered synchronously  one at a time, or diachronically from the 1st to the 4th (Fig. 4).

Quantitative research focuses on measurement and assumes that the subject of inquiry can be quantified. Its main aim is to provide comprehensive data through measurement, analyse the data for patterns and connections, and verify its accuracy. The scope of quantitative research ranges from easily quantifiable attributes such as height and weight to more intangible elements such as human emotions and thoughts.

The quantitative research approach is highly precise and logical, utilising statistical analysis to its fullest extent. Its ability to test theories through hypothesis formulation and formal statistical analysis sets it apart as a methodology. It is especially useful in measuring variables such as height, weight, attitude, and well-being, differentiating between independent and dependent classifications and capturing the influence of the former on the latter. Multiple hierarchical measurement theories are also employed to acquire diverse measurement types (Tab. 4).

At its most basic level, nominal classification categorises data without quantitative analysis. As we move towards ordinal measurement, we introduce a hierarchical structure to the data, although this method may require more precision. We rely on interval and ratio-level measurements for increased accuracy, although generating a ratio can be challenging when studying social phenomena. Ordinal and interval measurements are the most commonly used techniques in quantitative research.

Regardless of the method of measurement, errors are bound to occur. These errors can stem from various sources, including instrument, human, and random errors.

Although it is possible to reduce instrument and human errors, it is impossible to control random errors. Therefore, it is essential to consider random errors when designing and using any instrument. Instrumental and human errors can manifest in two ways: within the instrument (or within the human operator), which means that the same instrument may produce varying results in different settings, or inter-instrument (or human-to-human), which means that two seemingly identical instruments may yield different results.

Similarly, human errors imply that individuals using the same instrument may obtain divergent results with different advantages. On the other hand, instrument errors imply that two people using the same instrument may obtain different depths simultaneously. While errors cannot be eliminated, they can be minimised.

Effective instruments must be designed to minimise instrument errors. In social research, this means ensuring that observational questionnaires and checklists are easily comprehended and that questions are answered precisely.

When designing instruments, it is crucial to balance "authenticity" and "directivity." An authentic instrument measures as much as possible about a phenomenon but risks becoming indirect. In contrast, a direct instrument focuses only on items directly related to the phenomenon, potentially losing some authenticity (Watson, 2015).