Structural equation models of memory performance across noise conditions and age groups

Competing models of declarative memory were tested with structural equation models to analyze whether a second-order latent variable structure for episodic and semantic memory was invariant across age groups and across noise exposure conditions. Data were taken from three previous experimental noise studies that were performed with the same design, procedure, and dependent measures, and with participants from four age groups (13-14, 18-20, 35-45, and 55-65 years). Two noise conditions, road traffic noise and meaningful irrelevant speech, were compared to a quiet control group. The structural models put to the test were taken from Nyberg et al. (2003), which employed several memory tests that were the same as ours and studied age-groups that partly overlapped with our groups. In addition we also varied noise exposure conditions. Our analyses replicated and supported the second-order semantic-episodic memory models in Nyberg et al. (2003). The latent variable structures were invariant across age groups, with the exception of our youngest group, which by itself showed a less clear latent structure. The obtained structures were also invariant across noise exposure conditions. We also noted that our text memory items, which did not have a counterpart in the study by Nyberg et al. (2003), tend to form a separate latent variable loading on episodic memory.