Theoretical Framework

Fault diagnosis represents a central area of professional competence in the occupation of electronics technicians for automation technology (EA) at both national and international levels (Nickolaus & Geißel, 2009; Zinke, Schenk & Kröll, 2014). Curricular frameworks emphasize that trainees must be able to analyze complex relationships between symptoms and underlying causes and apply systematic strategies of troubleshooting and fault elimination (KMK, 2003). These requirements reflect the increasing complexity of industrial automation systems, in which diagnostic failures can lead to substantial downtime, safety risks, and economic losses. Despite the curricular emphasis on diagnostic competence, empirical studies indicate that many trainees experience difficulties in accurately identifying fault causes (Walker et al., 2015). This discrepancy suggests that the curricular goals regarding diagnostic proficiency are not fully met in practice.

A key gap in the current state of research concerns the lack of detailed analyses explaining why trainees struggle with fault diagnosis. Existing findings describe performance outcomes but do not sufficiently explore the underlying cognitive processes, diagnostic strategies, or information-processing patterns that shape successful or unsuccessful diagnoses. This lack of process-oriented understanding is problematic from a subject-didactic perspective, as meaningful instructional interventions require insights into the mechanisms of diagnostic thinking, not merely its outcomes.

To address this, the present study introduces a theoretical and methodological approach for analyzing diagnostic processes. Drawing on occupational and cognitive psychology (Schaper & Sonntag, 1997a, 1997b) as well as subject-specific didactics (Benda, 2008), fault diagnosis is conceptualized as action within search spaces (Klahr & Dunbar, 1988). According to this framework, diagnostic behavior unfolds through the navigation of different problem spaces, guided by specific cognitive strategies.

The literature suggests two central diagnostic strategies:

1. Symptomatic Strategy
   This strategy aligns with pattern-recognition theories and involves the activation of pre-compiled knowledge structures, such as familiar diagnostic scripts or previously encountered fault patterns. While efficient for routine or typical faults, it is limited when applied to unfamiliar or structurally complex problems.

2. Topographic Strategy
   This strategy emphasizes the construction of a mental representation of the system’s functional architecture. It relies on systematic information gathering, hypothesis testing along the system’s topology, and the integration of diverse data sources. Research suggests that this strategy is particularly effective in complex or non-routine diagnostic situations (Konradt, 1995; Hoc, 2000).

Whereas initial causal analyses have been conducted in other vocational domains—such as business education (Rausch et al., 2017) or automotive mechatronics (Abele, 2018)—no comparable research exists for electronics technicians for automation technology. Consequently, little is known about whether EA trainees rely predominantly on symptomatic, topographic, or hybrid strategies, and how these strategies relate to diagnostic success.

Research Questions
Addressing this deficit, the study systematically investigates:

a) To what extent can symptomatic, topographic, and hybrid diagnostic strategies be empirically identified in the diagnostic processes of EA trainees?
b) How are these process characteristics related to diagnostic outcomes, particularly the accuracy and quality of the final diagnosis?

*Methods* To examine the research questions, the study applies a multi-method design combining written protocols and logfile data. This triangulation allows for a comprehensive analysis of both self-reported diagnostic reasoning and objectively recorded interaction behavior. The first data source consists of written diagnostic protocols collected from n = 318 trainees in their third year of vocational training. Participants worked with an ecologically valid computer simulation that reproduces industrial automation processes and typical faults. During troubleshooting, trainees were instructed to verbally document their reasoning, including generated hypotheses, information sources accessed, intermediate conclusions, and final diagnostic decisions. Written protocols capture consciously accessible reasoning steps and are widely used to study problem-solving processes in vocational education. The second data source consists of logfile data from an additional sample of n = 89 trainees, who completed the same three fault scenarios within the simulation. The simulation software automatically recorded all user actions—such as opening schematics, performing measurements, selecting components, or navigating subsystems—together with precise timestamps. Logfiles provide high-resolution process data that reflect actual diagnostic behavior and allow for detailed analyses of information-processing patterns. Data analysis followed a cluster-analytic approach using the Cascade Simple k-means algorithm (Calinski & Harabasz, 1974) implemented in WEKA 3.7, supplemented by Classification via Clustering (Hall et al., 2009). Variables included action frequencies, temporal sequences, navigation paths, and transitions between search spaces. This approach allows identifying latent strategy patterns and relating them to the theoretically derived model of symptomatic vs. topographic strategies. To examine the relationship between strategy type and diagnostic success, cluster membership was compared with accuracy measures derived from the final diagnosis entered by the trainees. This enabled an integrated interpretation of diagnostic process and diagnostic performance.

The analysis of the written protocols resulted in a stable three-cluster solution across all fault scenarios. In addition to purely symptomatic and topographic diagnostic strategies, a hybrid pattern emerged, characterized by an initial reliance on symptom-based heuristics followed by a systematic topographic approach. This hybrid strategy appears particularly relevant when trainees initially apply known scripts but then transition to deeper structural reasoning when these scripts do not yield a solution. Regarding diagnostic success, the findings align with theoretical expectations: trainees applying a topographic strategy, or those who executed a strategy shift from symptomatic to topographic reasoning, achieved high success rates. In contrast, trainees relying exclusively on symptomatic strategies demonstrated significantly lower success probabilities, particularly when dealing with complex or unfamiliar faults. These results highlight the central role of mental model construction and systematic information processing for effective fault diagnosis. Preliminary logfile analyses, based on one fault scenario, reproduce the same cluster structure identified in the protocol data. This convergence suggests that both methodological approaches capture comparable underlying cognitive processes. The final study will present complete logfile analyses, offering additional insights into micro-level diagnostic behavior, such as navigation patterns, measurement strategies, and temporal dynamics. The study’s findings provide a theoretical and empirical foundation for enhancing diagnostic competence training in vocational education. They highlight the need to support strategy flexibility and to foster topographic reasoning through targeted instructional designs.

Abele, S. (2018). Diagnostic Problem-Solving Process in Professional Contexts. Vocations and Learning, 11(1), 133-159. Benda, D. (2008). Das große Handbuch Fehlersuche in elektronischen Schaltungen. Franzis. Calinski, T., & Harabasz, J. (1974). A dendrite method for cluster analysis. Communications in Statistics, 3(1), 1–27. Hall, M., et al. (2009). The WEKA Data Mining Software: An Update. SIGKDD Explorations, 11(1). Hoc, J.-M. (2000). A Method to Describe Human Diagnostic Strategies. International Journal of Cognitive Ergonomics, 4(4), 297-309. Klahr, D., & Dunbar, K. (1988). Dual Space Search During Scientific Reasoning. Cognitive Science, 12(1), 1-48. KMK (2003). Rahmenlehrplan für den Ausbildungsberuf Elektroniker/Elektronikerin. Konradt, U. (1995). Strategies of failure diagnosis in computer-controlled manufacturing systems. International Journal of Human-Computer Studies, 43(4), 503-521. Nickolaus, R., & Geißel, B. (2009). Electricians. In Feasibility Study VET-LSA (pp. 48–70). BMBF. Rausch, A., et al. (2017). Problemlöseprozesse sichtbar machen: Analyse von Logdaten. ZBW. Schaper, N., & Sonntag, K. (1997a/b). Modelle diagnostischen Handelns. In Störungsmanagement und Diagnosekompetenz (pp. 39–76 / 155–172). Vdf. Walker, F., et al. (2015). Berufsfachliche Kompetenzstrukturen bei Elektronikern. ZBW, 111(2). Zinke, G., et al. (2014). Ergebnisse einer Online Befragung zur Berufsfeldanalyse. BIBB.