In many research and product development contexts, there is an implicit assumption that data can reveal an objective truth. Dr. Karthikeyan’s talk about data triangulation challenges this assumption by offering a more nuanced perspective. Rather than searching for a single ground truth, we need to recognize that all measurements are partial, biased, and context dependent.

A central concept in this discussion is Umwelten, which describes how individuals perceive and experience the world through their own sensory and perceptual systems. Each person operates within a bounded perspective shaped by biology and experience. This means that no single user, sensor, or observer can fully capture reality. What we observe is always only one slice of a larger and more complex system. This leads naturally to the idea of data triangulation. Instead of relying on one source of data, triangulation combines multiple perspectives to reduce bias. The goal is not to find a perfectly exact answer, but to identify patterns that remain consistent across different kinds of evidence. The image of overlapping spaces is especially useful here. The region where multiple signals align is often the closest approximation we have to truth.

An important implication is that human behavior is inherently stochastic. It is not fully deterministic or perfectly predictable. Even under similar conditions, people may behave differently. This challenges the expectation that models or systems should always produce consistent outputs. A better goal is to design systems that respond robustly to variability rather than pretending variability does not exist.

Another key insight is the proxy problem. In human factors research, we rarely measure the phenomenon of interest directly. Instead, we rely on proxies such as physiological signals, behavioral performance, or subjective reports. Each proxy carries its own bias. Physiological data may indicate arousal, but not necessarily its cause. Subjective data may capture perception, but not actual behavior. Observational data may reveal patterns, but not the underlying intent behind them.

This is where triangulation becomes especially powerful. Objective data can provide direction by indicating what might be happening. Subjective data can challenge that direction by revealing discrepancies between perception and measurement. Observational data helps explain why those discrepancies exist. By integrating these sources, we can build a more complete understanding and generate more meaningful recommendations for product design.

The talk also emphasizes the importance of context and measurement uncertainty. Not every difference in data is meaningful. Measurement noise, variability, and standard deviation all shape whether an observed effect is large enough to matter. This requires careful interpretation rather than blind confidence in numerical outputs.

🧠Based on the discussions in class, I find it interesting that triangulation also applies to users themselves. When evaluating products, customers naturally perform their own form of triangulation. They combine personal experience, reviews, and observable performance to form judgments.  This perspective suggests that transparency is not just a design choice, but a way to support users’ decision-making process. By making system behavior and decision logic more visible, we can help users build confidence and trust.