The associative deficit hypothesis posits that older adults perform worse on long-term memory tasks than younger adults because they have difficulty creating and retrieving associations between different units of information (Naveh-Benjamin, 2000). We aimed to replicate this effect and to determine if it is driven by a difficulty in storing (i.e., creating) and/or retrieving the associations. Previous studies on the locus of this deficit have yielded mixed results. Thus, we used two paradigms with different materials, allowing us to test the robustness of our results. Specifically, in the first paradigm, we presented clusterable words from a category (e.g., cat & dog) and singleton words without a clusterable associate, and after a brief distractor phase, participants completed a free recall task (pair-clustering paradigm; Batchelder & Riefer, 1980). The second paradigm used cue-target learning of weakly associated pairs (e.g., flame - heat) and tested memory with free and cued recall (free-then-cued-recall paradigm; Rouder & Batchelder, 1998). The behavioral data from our younger (18-35 years) and older adults (at least 60 years) replicate the associative deficit. However, our modeling results suggest different processes to drive this deficit. This may indicate that the storage and retrieval estimates from the two multinomial models do not represent the same processes. Or maybe clusterable category words are easier to remember than associations between previously unrelated words, or maybe older adults engage more in the pair-clustering paradigm because they are actively searching for the related pair word.