This study used population-representative longitudinal data from the 2005-2006 and 2010-2011 waves of the National Social Life, Health and Aging Project-a probability sample of US adults aged 57-85 at baseline (N = 650 women and 620 men)-to examine the causal direction in linkages of endogenous testosterone (T) with sexual activity and relationship quality. For both genders, our autoregressive effects indicated a large amount of temporal stability, not just in individual-level attributes (T, masturbation) but also dyadic ones (partnered sex, relationship quality)-indicating that a need for more nuanced theories of relational processes. Cross-lagged results suggested gender-specific effects-generally more consistent with sexual or relational modulation of T than with hormonal causation. Specifically, men's findings indicated their T might be elevated by their sexual (masturbatory) activity but not vice versa, although androgen levels did lower men's subsequent relationship quality. Women's T, in contrast, was negatively influenced not just by their higher relationship quality but also by their more frequent partnered sex-perhaps reflecting a changing function of sexual activity in late life.
Keywords: Challenge hypothesis; Older adults; Relationship quality; Sexual activity; Testosterone.