Résumé : Gamma-ray bursts (GRBs) are traditionally classified into long (lGRBs) and short (sGRBs) durations based on their $T_{90}$, with lGRBs widely used as tracers of the cosmic star formation rate (SFR) due to their observed association with core-collapse supernovae. However, recent detections of kilonovae accompanying some lGRBs challenge this assumption, suggesting potential contamination from compact binary mergers. Here, we move beyond the conventional $T_{90}$-based classification and focus exclusively on GRBs directly associated with supernovae - the most direct signatures of massive stellar collapse - to reassess their connection to the SFR. Using a sample of SN/GRBs, we construct the luminosity - redshift ($L$-$z$) plane and uncover a significant correlation between these variables. To account for observational biases, we apply the $\tau$ statistic and Lynden-Bell's $C^{-}$ method to derive the intrinsic luminosity function and formation rate. Our analysis reveals that even among this well-defined subsample, the SN/GRB formation rate still exceeds the SFR at low redshifts ($z < 1$). These findings suggest that GRBs at low redshift may not serve as reliable tracers of the SFR, and that larger samples are required to further investigate this discrepancy.