Formulating Impactful Research in Information Systems

This article provides a comprehensive overview of developing research topics, particularly within the field of Information Systems, and then delves into a detailed plan for studying the efficiency impact of warehouse automation using quantitative methods, drawing directly from the provided sources and our conversation history.

The sources outline two primary pathways to formulating a research topic: the Method-Follows-Question approach and the Question-Follows-Method approach.

• Method-Follows-Question (Problem-Driven): This is described as the more common and often more impactful approach. Here, a researcher begins with a research problem or question that they aim to address. Once the question is clearly defined, the researcher then selects the method that is most suitable for answering it. An example provided illustrates this: if the research question is “How do warehouse workers adapt to automation?”, an ethnographic study would be a fitting method due to its capacity to provide rich, contextual insights into the lived experiences and adaptations of the workers. This path is driven by a real-world problem that the researcher cares about, making the research inherently relevant.
  
• Question-Follows-Method (Method-Driven): In this approach, a researcher’s training or interest in a particular method takes precedence. The researcher then seeks out a research question that can be effectively answered using that chosen method. For instance, someone interested in applying sequence analysis might look for settings where sequences of actions or events are critical, such as warehouse picking routines or chatbot interactions. This approach is common in areas characterized by methodological innovation, such as text mining, social network analysis, or design science. While driven by methodological enthusiasm, it is still crucial to choose a topic where the application of the method truly matters and can yield valuable insights.
  

In Information Systems, a winning research topic typically possesses three key characteristics:

• It solves a relevant problem.
• It uses the right method, not merely a preferred one.
• It makes a theoretical contribution.

To further illustrate these concepts let’s talk about some of the research method I plan to adopt in my PhD journey 

Ethnography: This method is useful for understanding culture, meaning, work practices, or sensemaking in context. Examples of research questions include:

• How do warehouse workers make sense of their roles in automated environments?
• What forms of resistance or adaptation emerge during robotic transitions?
• How is power restructured in warehouses undergoing digital transformation? These types of questions could be explored in settings like Walmart fulfillment centers or pilot sites testing robotics.

• Sequence Analysis: This method is employed when the research interest lies in patterns over time, such as actions, behaviors, or routines. Relevant research questions include:
  
  • What recurring patterns of interaction exist between humans and robots in warehouse operations?
  • How do order-picking routines evolve before and after automation implementation?
  • What sequences lead to successful or failed fulfillment cycles? Conducting this type of research necessitates event logs, sensor data, or digital traces.
• Design Science: This approach is used when the goal is to build and evaluate an artifact, such as a decision-support tool, interface, or robot interaction prototype. Examples of research questions are:
  
  • How can we design an interface to improve human-robot coordination in warehouses?
  • What features enhance trust in AI-driven warehouse management systems?
  • How can warehouse performance dashboards be improved to support real-time decision-making? Research using design science involves designing a solution, evaluating its effectiveness, and connecting it to relevant theories such as the Technology Acceptance Model (TAM) or affordance theory.

Alternatively, research can begin with a specific problem area. For instance, if the core problem identified is: “Warehouse automation is disrupting how humans and machines work together,” various methods can be applied to investigate different facets of this problem:

• To understand the culture and sensemaking aspects, the question could be: How do workers interpret and respond to robotic automation? The most suitable method for this would be Ethnography.
• To analyze behavioral routines, the question might be: What actions lead to bottlenecks in robotic picking processes? Sequence Analysis would be the appropriate method here.
• To address interface design, a relevant question is: How can we improve task allocation between humans and robots? Design Science would be the chosen methodology.
• To assess the efficiency impact, the question is: Does automation improve fulfillment speed and accuracy? This question calls for Quantitative methods such as regression or experiments.

Following this framework, we then focused on how to study the efficiency impact question quantitatively. A detailed research plan was developed, encompassing the following key elements:

• Research Focus: The main question is clearly stated as: Does warehouse automation improve fulfillment speed and accuracy?
  
• Conceptualize Your Variables:
  
  • Independent Variable (IV): Automation Level, which can be measured as a binary variable (manual vs. automated warehouses) or a continuous variable (e.g., percentage of processes automated, number of robots deployed).
  • Dependent Variables (DV): Fulfillment Speed (e.g., average time per order, time to dispatch) and Order Accuracy (e.g., percentage of orders picked correctly, customer complaints/returns due to errors).
  • Control Variables: To isolate the effect of automation, it’s crucial to control for factors such as warehouse size, order volume per day, staff-to-order ratio, and seasonality.
• Formulate Hypotheses: Two testable hypotheses are proposed:
  
  • H1: Higher levels of warehouse automation are associated with faster order fulfillment.
  • H2: Higher levels of warehouse automation are associated with increased order accuracy.
• Research Design Options: Two primary quantitative research designs are considered:
  
  • Option A: Quasi-Experimental Design: This involves comparing warehouses before and after automation implementation (within-subjects) or comparing automated versus non-automated warehouses (between-subjects). This design is suitable when random assignment is not feasible. Techniques like difference-in-differences (DID) or panel data regression can be used if time-series data is available.
  • Option B: Cross-Sectional Regression: This involves collecting data from multiple warehouses at a single point in time. Regression models can then be used to analyze the relationship between automation level and the dependent variables while controlling for other factors.
• Data Sources: Relevant data can be collected from:
  
  • Internal warehouse data (if partnerships with companies like Walmart or Symbotic are possible).
  • Industry datasets (if available).
  • Time logs, inventory logs, and delivery logs.
  • Performance KPIs from before and after automation rollout.
• Tools & Analysis: Suitable tools and techniques for analysis include:
  
  • Statistical software such as Stata, R, SPSS, or Python for regression modeling.
  • Tools like Excel or Power BI for data visualization.
  • Propensity Score Matching to mitigate selection bias when comparing warehouses with and without automation.
• Theoretical Lens (Optional but Great): Anchoring the study in relevant theories can provide a deeper understanding of the findings:
  
  • Sociotechnical systems theory: to understand the co-evolution of technology and human systems.
  • Resource-Based View: to frame automation as a strategic capability.
  • Process Theory: to analyze how structural changes impact workflows.
• Outcome: Such a study has the potential to provide evidence-based insights into the real operational gains of warehouse automation and is highly publishable in relevant academic journals within the business and Information Systems communities.
  

Finally, the information provided in the sources regarding the “Two Paths to a Research Topic” and the detailed plan for a quantitative study on the efficiency impact of warehouse automation has been confirmed as fundamentally correct and aligned with standard research methodology practices. The explanation of the two paths to research topic formulation accurately reflects common approaches. The examples provided for brainstorming research questions based on method or problem are relevant and well-structured. The detailed plan for the quantitative study correctly outlines key steps such as variable conceptualization, hypothesis formulation, research design options, data sources, analysis tools, and the importance of a theoretical lens.