<p>Classical statistical methods are based on precise data and well-defined distributional assumptions, but many real-world datasets involve inherent imprecision or vagueness, which are better represented through fuzzy sets. This paper addresses the challenge of performing statistical inference when data are fuzzy. The primary objective is to develop a robust approach for fuzzy hypothesis testing and the construction of fuzzy confidence intervals. We propose a bootstrap-based methodology for these tasks, defining fuzzy hypotheses, computing fuzzy p-values, and introducing decision rules tailored to the fuzzy framework. To illustrate the practical application of this method, we conduct an empirical analysis of depression indices in elderly populations, using data from the easySHARE dataset. By comparing the Swiss population with neighboring countries, we demonstrate the practical utility of fuzzy inference methods. Our findings reveal that fuzzy confidence intervals and p-values may lead to differing conclusions compared to traditional crisp methods, highlighting the importance of selecting an appropriate inference technique that aligns with the underlying uncertainty and decision-making context. This work advances classical statistical inference by extending it to handle fuzzy data in a flexible, principled, and computationally feasible manner.</p>