Heycke and Gawronski (2020) developed a multinomial processing tree (MPT) model to separate the respective contributions of stimulus relations and of stimulus co-occurrences to relational evaluative conditioning (EC) effects. In a series of experiments, Heycke and Gawronski (2020) found the C parameter (the MPT model’s estimator of co-occurrence effects) to be unaffected by manipulations of encoding time, pairing frequency and the delay between learning and test phases, but (paradoxically) increased by a longer response window in the test phase (speeded choice task). This pattern of results is incompatible with current theorizing on the processes underlying stimulus co-occurrence effects (association formation, failures in memory retrieval) and thus requires further research on the sources and properties of the C parameter. In the present research, we tested whether a significant moderation of the C parameter by encoding time can be obtained when the test phase features a longer response window (allowing for overall larger C parameters). Across three experiments, we obtained robust evidence for a moderation of the C parameter by encoding time (with larger C parameters when participants were given more encoding time during the learning phase). The theoretical and methodological implications of these findings are discussed.