Ability tilt refers to a within-individual difference between two abilities (X-Y), e.g. differences between tech and verbal or verbal and math abilities. Studies have found associations between ability tilts and their constituent abilities (X or Y). Here we show that such associations may be spurious due to the non-independence of the two measures. Using data from the 1997 National Longitudinal Survey of Youth (NLSY97), we find that associations between ability and ability tilt may simply be due to more positive associations between two measures of the same or similar abilities compared to two measures of different or dissimilar abilities. This finding calls into question theoretical interpretations that have proposed that ability tilt correlations are due to differential investment of time and effort in one ability at the expense of the other ability.