Part 3: TYPES OF QUANTITATIVE RESEARCH DESIGNS

Research design is divided into quantitative, qualitative, and mixed methods. The researcher must decide on the most appropriate design for the type of research work (Fig. 10). The field of social sciences and education research employs a variety of quantitative research designs and methods, including experimental, correlational, survey, and quasi-experimental designs.

Quantitative research encompasses various experimental designs, including true experiments, quasi-experiments, applied behavioural analysis, and single-subject experiments. On the other hand, non-experimental quantitative research involves causal-comparative research, where the investigator compares multiple groups based on an independent variable that has already occurred, and correlational design, where investigators use correlational statistics to measure the association between variables or sets of scores.

Furthermore, these designs have evolved into more intricate relationships among variables, such as structural equation modelling, hierarchical linear modelling, and logistic regression techniques. Quantitative strategies have become more complex in recent years, with experiments featuring numerous variables and treatments, such as factorial designs and repeated measure designs. Elaborate structural equation models have also been developed to identify the collective strength of multiple variables and causal paths.

Each design has distinct characteristics and goals. In this section, we will explore these designs in-depth, providing a thorough understanding of their suitability for various research inquiries and highlighting their distinctive approaches.

Figure 10. Summarized form of Research Design Types (Asenahabi, 2019).

The methodology employed by a researcher can significantly affect the outcome and accuracy of their study. The research design encompasses the framework, structure, and strategy researchers use to address a research question. By carefully considering these elements, researchers can establish their hypothesis, conduct their study, and interpret their data (Leavy, 2022). In quantitative research, it is pivotal to maintain control and mitigate any factors that could sway or skew the findings. Reviewing the key characteristics and assumptions that underpin quantitative research is vital to appreciate how control plays a crucial role in quantitative research design (Asenahabi, 2019; Bloomfield & Fisher, 2019).

Quantitative research is a formal and systematic process for describing variables, testing their relationships, and examining cause-and-effect associations between variables. Quantitative research generates numerical data, predominantly informed by positivist or post-positivist paradigms and underpinned by various assumptions, such as the belief in a single truth or reality, objectivity, and deduction (Hair et al., 2021; Sukamolson, 2007).

Quantitative research is a scientific method that employs objective and impartial techniques to test hypotheses and find the correct answer. The research process involves drawing a representative sample of participants from a known population and measuring the variables of interest. Typically, researchers begin by testing the null hypothesis, which posits no connection between independent and dependent variables. The hypothesis is scrutinised, and the findings are assessed through statistical analysis. Ultimately, the null hypothesis is deemed acceptable or unacceptable based on the results of the statistical evaluations. Once the null hypothesis is accepted or rejected, inferences or generalisations can be applied to the population of interest. The research design must be reliable and have internal and external validity, collectively known as rigour, to allow confident generalisations about a population (Watson, 2015).

Rigour in quantitative research can be described as the amount of control the researcher exerts to prevent the effects of extraneous or confounding variables on the dependent (test or outcome) variable.

To assess the effect of a single factor on a particular outcome, the researcher must account for any other variables or external factors that could potentially affect the outcome. This can help isolate the specific impact of the independent variable being studied.

For example, to create a fall risk profile, a researcher needs to compare the characteristics of a sample of patients who have experienced a fall with those of a sample of patients who have not.

In this situation, if there is an error in the sample selection of the non-falling patient group and this group accidentally consists of a higher mean for age, then the difference – or not – between the two groups may be due to age due to the sampling error.

Quantitative research uses a range of different study designs. These may vary in how they are categorised according to four significant types of quantitative research: descriptive, correlational, quasi-experimental, and experimental (Tab. 6).

Table 6. Type of Quantitative Research Design

Source: Adapted from Bloomfield and Fisher (2019).

A descriptive quantitative study aims to examine variables in a single sample and systematically measure, describe, and interpret them. It is commonly used to gather data about a specific phenomenon or attribute of interest within a known sample or population in its natural environment without controlling or manipulating variables.

Research studies that describe and examine variables in two or more groups are comparative descriptive designs. The variable/s of interest are measured and defined in both groups and are then compared. For example, researchers might conduct a comparative descriptive study to describe the differences in educational qualifications between male and female nurses employed at a local hospital.

Findings from descriptive research studies are most valuable in determining the frequency to which something exists. They are also helpful in describing a new or little-known phenomenon. However, while findings cannot be used to establish cause and effect, they may help develop hypotheses that can be tested in future studies.

Researchers must use methods to ensure the data collected is reliable and valid. This includes using a probability sampling technique to select a sample of adequate size and accurately represent the target population. 

The instruments and methods used to collect data in descriptive studies include surveys, checklists, observations, interviews, and equipment to measure physiological variables, such as weight scales and thermometers. To ensure internal validity, these must also be calibrated, standardised, and piloted before use.

The primary objective of correlational research is to establish a relationship's existence, strength, and direction between two or more variables. That is the degree to which changes in one variable correspond to changes in another. By analysing the correlation coefficient, researchers can determine the nature and direction of the relationship between the variables, which is crucial in making informed decisions based on the data collected.

Like descriptive research, correlational studies do not manipulate the variables investigated and do not seek to determine cause or effect. Instead, they can describe or predict relationships or test theoretical models of relationships. Causal inferences regarding the relationships between independent and dependent variables are not drawn without random selection or manipulation of the independent variable. There is no attempt to make such inferences without following the abovementioned procedure (Rumrill, 2004). The findings from correlational studies can be statistically explained in three ways: positive, negative, and no correlation.

In statistics, a positive correlation refers to a connection or association between two variables. When one variable increases, the other variable also increases, or when one variable decreases, the other variable also decreases (Bloomfield & Fisher, 2019). This means that the two variables move in the same direction. For example, the amount of food a person consumes might correlate positively with weight.

A negative correlation between variables occurs when an increase in one variable results in a decrease in another and vice versa. For example, the more food a person consumes, the lower their hunger levels will be. Two variables are said to be uncorrelated when an alteration in one does not lead to an alternation in the other and vice versa.

Researchers typically rely on a statistical measure called the correlation coefficient when reporting the results of correlational studies. This value ranges from +1 to -1, with a figure near +1 indicating a robust positive correlation and a value close to -1 signifying a significant negative correlation. A value close to zero denotes that the variables are not correlated.

The third category of quantitative research design is quasi-experimental studies. These are similar to experimental studies in that they aim to test the effectiveness of interventions and, therefore, involve manipulating an independent variable.

However, unlike a proper experimental study (e.g., a randomised controlled trial), it lacks the random allocation of participants to certain conditions, such as an intervention/experimental or control group. This may have considerable implications because factors other than those being investigated may affect the findings. These are known as confounding or extraneous variables.

When conducting a randomised controlled trial is not feasible or ethical, quasi-experiments are usually performed. Different types of quasi-experimental studies include, among others (Tab. 7.):

• non-equivalent control pre-test post-test design
• non-equivalent control post-test-only design
• one group pre-test–post-test design and
• time-interrupted series.

Table 7. Features of Quasi-Experimental Research Design

Source: Bloomfield and Fisher (2019).

The experimental study design has the most significant level of control. It has frequently been identified as the gold standard of quantitative research due to its ability to determine a cause-and-effect relationship between an intervention (the cause) and the study outcome (the effect) (Rogers & Révész, 2020).

Scientific research widely recognises experimental designs as the gold standard. This method, known as true experimentation, establishes a cause-and-effect relationship between variables within a study. Despite common misconceptions, true experimentation is not exclusive to laboratory settings.

Experimental research provides a structured approach to establishing causal relationships between variables. Using this approach, the researcher is actively involved in deducing and testing hypotheses. The researcher manipulates an independent variable (cause) and observes its effect on a dependent variable while attempting to control for extraneous variables. This is achieved by administering the treatment to one group while withholding it from another and then analysing the resulting scores of both groups.

In the realm of research, an experiment involves selecting participants randomly and exposing them to different levels of one or more variables, known as independent variables. The researcher then observes the impact of this exposure on one or more outcome variables called dependent variables. The aim of conducting an experiment is to establish a correlation between the independent and dependent variables and deduce conclusions regarding the effectiveness of the intervention and its causal connection. A key aspect of this process is controlling for extraneous variables. This control is essential as it ensures that any observed effects are solely due to manipulating the independent variables, enhancing the study's validity. Experiments are a potent tool for investigating cause-and-effect relationships in diverse fields, including psychology, medicine, physics, and engineering (Mizik & Hanssens, 2018).

True experiments randomly allocate subjects to treatment conditions, while quasi-experiments utilise nonrandomised assignments. To ensure the similarity of subjects, cases are matched on various characteristics and randomly assigned to control and experimental groups. Only observable facts are considered, and inferential statistics produce precise numerical results. Different experimental designs have been developed, from simple before-after to complex multivariate factorial designs, including:

• Parallel design - In a parallel design, participants are randomly assigned to either the intervention or control group.
• Crossover design - In a crossover design, participants are initially assigned to either the intervention or control group and then switch over to the other group after a certain period of time. This design helps negate bias from individual differences since each participant becomes their own control.
• Cluster design - In many research contexts, it is not always possible to randomise individuals to receive different interventions. To overcome this, groups or clusters of individuals (for example, wards, units or hospitals) can be randomly assigned to either the control or intervention, and all cluster members will receive the allocation.

The randomised controlled trial (RCT) is a highly valued research approach that embodies these virtues (Styles & Torgerson, 2018) (Fig. 11).

Figure 11. RTC-Randomised Controlled Trial (Crano et al., 2014).

In this type of experiment, all variables are identified and controlled except for one. The independent variable is manipulated to observe its effects on dependent variables. Additionally, participants are randomly assigned to experimental treatments instead of being selected from naturally occurring groups, which ensures the validity of the research.

The fundamental principles of experimental designs include random assignment, variable manipulation, and control groups. While experimental designs effectively establish causal relationships, they have limitations, such as ethical considerations and practical constraints.

The fundamental framework of a quantitative design is rooted in the scientific method, utilising deductive reasoning. This involves the researcher developing a hypothesis, investigating to gather data on the problem, and then analysing and sharing the conclusions to demonstrate that the hypotheses are not untrue.

To follow this procedure, one should:

• Observe an unknown, unexplained, or new phenomenon and research the current theories relating to the issue.
• Create a hypothesis to explain the observations made.
• Predict outcomes based on these hypotheses and create a plan to test the prediction.
• Collect and process data. If the prediction is accurate, proceed to the next step. If not, create a new hypothesis based on the available knowledge.
• Verify the findings, draw your conclusions, and present the results in a suitable format.

Crano et al. (2014) outline the steps of a classic true experimental research design, which involve gathering a group of participants, conducting a pre-test on the dependent variable, randomly assigning participants to either the experimental or control group, closely controlling the application of the experimental treatment between the two groups, and measuring both groups again on the dependent variable after the experimental manipulation (Fig. 12). Variations exist, such as removing pre-testing, including multiple experimental treatments, or using the same participants across all experimental conditions.

Figure 12. Pre-test - Post-test Control Group Design (Crano et al., 2014).

Following crucial steps is essential to designing and conducting experiments effectively. These steps include the following (Kuçuksayraç, 2007):

• Sampling participants for the study.
• Randomly assigning participants to groups.
• Randomly assigning groups to experimental or control conditions.
• Defining the independent variable, which refers to the aspect of the environment being studied that varies between the groups.
• Defining the dependent variable, which measures any resulting behavioural changes.
• Controlling all other variables that may affect the dependent variable while keeping the independent variable consistent.
• Conducting statistical tests to confirm or refute the hypothesis to determine whether there are any differences between the two groups regarding the dependent variable measurements.
• If the hypothesis is confirmed, explain and generalise the findings.
• Finally, predict how the findings may apply to other situations, potentially through replication.

In conclusion, quantitative research is an essential method for measuring variables and evaluating the effectiveness of interventions. Unlike qualitative research, quantitative research is conducted objectively, emphasising reducing bias. Researchers who seek to adopt evidence-based practices must have a strong understanding of quantitative research design. This knowledge allows them to comprehend and evaluate research literature better and potentially integrate study outcomes and recommendations into their work.

Table 8. summarises the alternatives for implementing quantitative and experimental research projects adjusted to different research conditions.