The Use of the Exploratory Sequential Approach in Mixed-Method Research: A Case of Contextual Top Leadership Interventions in Construction H&S

Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Associated Data

The data supporting the reported results can be received upon reasonable request, in accordance with the data policy of the University of Johannesburg and the prevailing legislation on data sharing.

Abstract

Quality and rigour remain central to the methodological process in research. The use of qualitative and quantitative methods in a single study was justified here against using a single method; the empirical output from the literature review should direct the current worldview and, subsequently, the methodologies applied in research. It is critical to gather contextual behavioural data from subject matter experts—this helps establish context and confirm the hypotheses arising from the literature, which leads to the refinement of the theory’s applicability for developing a conceptual model. This paper identified the top leaders in construction organisations as subject matter experts. Nine semi-structured interviews were conducted, representing the South African construction industry grading. The output of the refined hypothesis was followed by a survey that targeted n = 182 multi-level senior leaders to gather further perspectives and validate the conceptual model. The outcome resulting from the rigorous validation process adopted—the analysis process, which included Spearman rank correlation, ordinal logistic regression and multinomial generalised linear modelling—demonstrated that the lack of H&S commitment in top leadership persists, despite high awareness of the cruciality of H&S in their organisations. Contextual competence, exaggerated by the local setting, is one source of this deficiency. This paper provides guidelines for using the exploratory sequential approach in mixed-method research to effectively deal with contextual issues based on non-parametric modelling data in top leadership H&S interventions.

Keywords: exploratory sequential, mixed-method, qualitative data, quantitative data, reliability and validity, top safety leadership, safety leadership in construction, transformation of safety culture, safety culture in construction, top leadership

1. Introduction

Edmonds and Kennedy [1] defined the exploratory sequential technique as a progressive strategy that is used anytime that quantitative (QUAN) results are augmented by qualitative (QUAL) data. As a result, quantitative data analyses and explains the QUAL results in succession. The exploratory sequential technique is distinct from the explanatory sequential technique because it explores a concept before validating it, allowing for greater versatility in discovering novel ideas offered by the QUAL approach [2]. Numerous projects characterised by novel instrument creation choose this method as it enables the scholar to construct the instrument using QUAL information and afterwards verify it quantitatively [1,3]. As the sort of information generated by the first phase is uncertain—including whether it will emerge in a deterministic or non-parametric framework—and because the first phase is undertaken on a limited sample size, even though saturation would be achieved, the development of a new measurement instrument will be required. This is undertaken to handle the complexities of the resultant model characteristics because the contextual setting of top leadership is uncertain of the H&S culture’s consequences. Categorical data enables a greater level of precision and unambiguity [4]. Hence, it is advisable to perform validation or tests on the QUAN part of the model [3,5].

One of the distinct advantages of using an exploratory sequential approach is described by Heesen et al. [6] as a method that comparatively provides more robust validity. First, according to Flick, the interview-based QUAL methodological technique is suitable for resolving unresolved issues and developing and extending ideas based on such discoveries [7]. Interviews generate extensive data that allows subdomains of ideas to be studied. Furthermore, interviews are a direct data-collecting approach that is optimum for understanding issues’ complexity and depth. These collected ideas stemming from the rich data collected are used to reinforce the hypothesis [8]. When referring to the survey QUAN methodological approach, Bajpai [9] asserts that primary sources of data provide multitudes of benefits; it is noted that primary findings are frequently pertinent to the research objectives since they are collected on an individual basis. Applying both QUAL and QUAL approaches to single research works offers a more significant opportunity to establish more insight into the study subject, whilst a higher degree of validity and accuracy are achieved compared to applying a single approach [10,11].

This paper presents an interpretative, exploratory sequential methodology established on contextualism/a pragmatic worldview. Therefore, it is critical to establish the basis for this worldview as a start, to create a platform for the type of knowledge approach that this paper has adopted.

2. Theory

2.1. Establishing the Worldview

According to Crotty [12], a worldview or ontology is how the world is interpreted as existing. Research indicates the difficulty caused by the environmental context in the infrastructural development initiative in South Africa—particularly the necessary competency in upper-echelon leaders to lead a high-performance culture in organisational H&S [13]. In research associated with this, failure to select an appropriate tool in the beginning further increases methodological difficulties and causes severe confusion, leading to worthless study outputs [14,15]. Dumrak et al. [16] and Marle and Vidal [17] emphasise the complications brought by context by pointing to the extreme intricacies of major construction projects compared to smaller projects.

Accordingly, in a literature review, this study has applied an approach promoting contextualisation, according to Pepper [18]. This theoretical paradigm is therefore applied systematically throughout the whole study. Perception may be classified into four theoretical aspects: formism, mechanism, organicism, and contextualism, according to the orientation to cognition by Pepper [18]. Contextualism is a theoretical paradigm that presupposes a definitive understanding of a phenomenon categorisation occurs once it is placed in its main context [18]. As a result, it is logical for this paper to be aligned to a worldview that demonstrates the effects of the national and economic sectoral environment on the capacity of top leadership to change H&S culture and influence H&S results.

Zikmund [19] views contextualism as pragmatism, asserting that pragmatism as a philosophy is based on behaviour, circumstances, and outcomes rather than past conditions. It is supported by a paradigm focused on what constitutes logic and how to resolve issues.

2.2. Epistemology

True perception is seldom universal but rather illative, interpretative, and speculative. The criterion by which existence is measured is mostly pragmatic [3,14,20]. Crotty [12] describes epistemology as a conceptual viewpoint followed by a logical position that informs methodology and thus brings purpose to a technique that specifies the study’s logic and variables selection. The respondent (knower) and the individual cognitive bias (the known) in the H&S leadership commitment, in the context defined by the worldview, are the criteria in this case. As Morris [21] suggests, this “known” knowledge acts as a precursor to the efficacy of the interpretative paradigm, which is based on the notion that all knowledge is contextual. The interpretative method is concerned with perceiving nature via one’s subjective impressions. These rely on a mutual engagement between the researcher and the issues and use perception procedures (rather than QUAN) such as interviews. This approach backs up the notion that substance is formed via first-hand opinion; it believes that forecasts are difficult to come by. Per this concept, individuals have free will, aspirations, emotions, and thinking [11,22]. By conducting interviews with subject matter experts—the top leaders—and by also conducting surveys with the top management team (TMT), this study fits well within the interpretative scientific logic concerning the nature of knowledge—hence the adoption of the exploratory sequential technique for gathering and analysing the required data.

2.3. Research Design

Research design is a thorough description of the steps that must be followed during the data gathering and analysis to produce a satisfactory answer to research questions [5,23]. Additionally, research design may be defined as the overarching principle that the study will adhere to for the many components of the study to be applied logically and succinctly, assisting the scholar in reaching an ideal outcome [24]. Figure 1 shows the research design for this paper.

Research design outline (Adapted from Zikmund et al. [19]).

To accomplish a best-fit conceptual model, which is one of the outputs of this paper, the procedure for adopting the conceptual model’s hypotheses, description, and assessment requirements follow Elangovan and Rajendran’s [25] seven-step rigorous conceptual modelling framework and integrate an acclimation of Zikmund et al.’s [19] scientific approach. The background knowledge was derived from the literature review’s output and synthesised into eight hypotheses, forming a typology. Gogo and Musonda [26] state that this typology forms the basis for the background input for the methodology section described in this paper. The literature demonstrated that leadership is implicitly and explicitly related to H&S outcomes. The eight hypotheses developed from the literature review showed that leadership positions are challenging and require greater comprehension to ensure a decrease in the rates of injury to workers. These hypotheses are all anchored on the actions emanating from top leadership commitment in the South African construction industry.

2.4. Conceptual Typology

The functional content measure for the typology in Table 1 is the synthesis of the project’s eight propositions and how these contribute to one another and the study’s four core principles, namely: top leadership participation, regional cultural background, H&S culture and H&S performance. The H&S competency creation and audit framework contribute to top leadership engagement in the typology. In contrast, external influences such as regional culture and the business field are used as the backdrop for top leadership engagement. Table 1 shows the interactions between the model variables.

Table 1

Functional content measures for the typology.

Source: Gogo and Musonda [26].

The contextualism approach to philosophy is reinforced by the national culture and construction industry for this typology. The description of conceptual restrictions by Babbie [27], which range between Meta (South Africa), Macro (Construction sector), Meso (Top organisational leadership) and Micro (Leadership commitment), is applied consistently in the typology [26].

3. Methods

3.1. Data Collection Approaches

The initial phase of the data collection and analysis was the interview stage. It was characterised by non-probability purposive sampling to establish a theory based on the conceptual model and hypothesis. This was the QUAL phase, where the nine interviews were conducted. The second phase, QUAN, comprised the pilot stage of the survey study, where non-probability convenience sampling was applied to 10% of the target sample as a pilot study to establish the tool’s validity. The third stage, characterised by random probability sampling, used the developed survey questionnaire to gather perspectives on top leadership commitment to H&S by applying QUAN.

3.2. Population and Sampling of the QUAL Study

Patten and Newhart [28] describe a population in research as the people, things, wildlife and vegetation involved in the research. Typically, a sample is drawn to represent the population [29]. This paper considered the target population as the top leaders of the construction organisations in South Africa at all nine levels of the CIDB.

Representation, subject-matter expertise and thematic saturation are critical for determining sample sizes [30]. For qualitative studies, Guest et al. [31] describe the adequacy of a sample size as reaching saturation at 6 to 12—where 30 has been defined as the upper limit. To reach QUAL saturation in this paper, a sample of n = 9 was aimed for. Additionally, in the QUAL study, subject-matter expertise was ensured by targeting top organisational leaders for their primary prowess in the upper-echelon leadership of contractors and, in particular, their H&S business liability, as per Galvin [32]. Although the target sample for reaching QUAL saturation is small, it still has to fulfil the representation criteria to ensure that the population is well represented [29,32]. In the case of the QUAL study, this representation is achieved by first ensuring that all nine levels of the CIDB are included, and then secondly, by applying a non-probability, purposive sampling method. Non-probability purposive sampling creates a direct method for targeting a subject-matter expert based on defined criteria (CIDB grade, position in company, legal appointment in company and more) [31].

3.3. Population and Sampling of the QUAN Study

Consistency is key to quantitative studies [33]. Accordingly, comprehensive insight into a specific phenomenon is validated by many respondents, demonstrating consistency in supporting a defined proposition. Cooper and Schindler [2] posit that of the many variables that define a sample size, the size of the population, uncertainty, variance and confidence interval are among the most influential.

In the QUAN study of this paper, a pilot survey that targeted n = 18 respondents was achieved by non-probability convenience sampling. The main survey applied random probability sampling to n = 180 respondents. Non-probability convenience sampling was selected for its versatility and to limit the selection of multiple members in the same group—thus ensuring full representation in a smaller sample size. On the other hand, random probability sampling was selected because the likelihood of each member being selected is known—thus ensuring greater participation in the larger sample size [22,33,34]. This is particularly useful when targeting multi-level respondents, as was the case in this paper, to ensure that perspectives in top leadership commitment to H&S are gathered from all levels of the upper-echelon construction organisations’ leadership.

3.4. Interview Data Collection Procedure

To ensure clarity in the keynote and the spheres of the enquiry and assessment, Arksey and Knight [35] support the idea of two interviewers; however, this view is refuted by Whiting [36], who posits that the use of a single interview conductor is sufficient. Whiting [36] substantiates this assertion by further providing robust interview guidelines. This paper uses a single interviewer following Whiting [36].

For this study, the design of the semi-structured interview questionnaire followed the protocols and guidelines by Scheele and Groeben [37], Graneheim and Lundman [38] and Whiting [36], which emphasise that questions should be based on the reviewed literature. Accordingly, all interview questions are based on a conceptual typology proposed by Gogo and Musonda [26], which depicts how each leadership aspect relates to each respondent’s H&S aptitude for each contextual determinant. This allows the researcher to examine how the model depicts leadership commitment related to H&S in its full context by accurately representing the respondent [39].

Furthermore, these guidelines included the setting, which in the case of this paper was the respondent’s office—or online in case of constraints for a physical meeting. The respondents were also provided with a short description of the research, and the purpose of the interview was explained clearly. The divisions of the interview questionnaire were explained, and the entire meeting session was kept aligned with the ethical boundaries set beforehand. The interviews were verbal while incorporating the probing techniques shown in Table 2 to reach a clear response. The respondent’s answers were recorded verbatim on both tape and interview answer sheets by the interviewer. Recording the interview answers verbatim offers a robust method for data collection [3].

Table 2

Techniques for probing which can be used during interviews.