Analyzing Supply Chain Uncertainty to Deliver Sustainable Operational Performance

Read this study, which surveyed supply chain managers to understand how they address supply chain uncertainty. Section 5 identifies the study's solutions to overcoming uncertainty.

Research Methodology

Data Collection Process

Survey data were collected from firms based in Thailand. The respondents were typically the supply chain decision makers of their firms. Phillips suggests that high-ranking informants tend to be more reliable sources of information than low-ranking ones. Respondents were randomly selected from the list of registered companies in the Thailand Business Directory published by Teleinfo Media Public Co. Ltd., Bangkok, Thailand in 2015–2016. The sample firms included businesses of various sizes e.g., small, medium, and large. A combination of mail, e-mail, and telephone survey was used to collect data. In this study, respondents were requested to evaluate the extent, on a 5-point Likert scale, with which their firms practice the various aspects of measures. Although the use of single informants may result in method variance, as well as informant bias, the logistics or supply chain manager is most likely to be the most knowledgeable informant on the issue. Consequently, of the 307 surveys sent out, 155 were returned. There were 146 usable responses included in the subsequent analysis. The overall response rate was 47.56%. The response rate is reasonably acceptable when compared to that of recent studies in operations management. The survey was conducted over a period of two-and-a-half-month. Two techniques were used to improve the response rate, which was: (1) following up with reminder phone calls, and (2) promising to mail a final summary of the study's results to responding firms for their reference.

Measurement Scales

The measurement of the constructs is based on existing validated scales. All the constructs in this study are operationalized in a reflective-reflective type 1 model based on theoretical considerations. The scale of SCU draws on eight-item scales from Chen and Paulraj, which many prior studies have also used. Similarly, based on prior research, the scale of SCS draws on fourteen-item scales. Finally, based on earlier studies, the scale of OP draws on eight-item scales from Bayraktar et al.

Symmetrical Modeling Approach

The research model (Figure 1) is analysed by employing a partial least square–structural equation modelling (PLS-SEM), as implemented in SmartPLS 3. SmartPLS 3 assesses the psychometric properties of the measurement model and estimates the parameters of the structural model simultaneously. PLS-SEM is a widely accepted variance-based, descriptive, and prediction oriented technique to SEM. Using PLS-SEM is particularly more suitable when the research objective focuses on prediction and explaining the variance of key target constructs by different explanatory constructs; the sample size is relatively small, and/or the available data is non-normal; and, if covariance-based SEM provides no, or at best questionable results.

Asymmetrical Modeling Approach

The hypothesized paths suggest a causal relation leading from SCU, through SCS, to OP (Figure 1). However, this relation may be more complex. That complexity is empirically scrutinized by a conventional correlational method (i.e., PLS-SEM) and an innovative configurational method based on set-theoretical approaches (i.e., fsQCA). Indeed, configurations play a crucial role in management research. FsQCA is often applied in management research in conjunction with conventional statistical methods (regression for instance). However, it differs from them in the way that it prompts the researcher to go beyond mono-causality rationale and brings out the multiplicity of causal paths underlying management phenomena by using fsQCA software. FsQCA approach uses Boolean algebra to generate a combination of causal conditions leading to an outcome. Central to fsQCA approach is the calibration procedure and the truth table analysis. The calibration is a transformation process consisting in converting conventional measures into fuzzy sets.