Wordsum is weakly g‑loaded

Using item‑level tests from the National Longitudinal Survey of Youth and demographic matching to the General Social Survey, a reconstructed 'quasi‑Wordsum' suggests the 10‑item Wordsum is short, range‑restricted, and has lower loading on the general intelligence factor (g) than longer batteries like the AFQT. Much of its poorer predictive power for political and social outcomes arises from test length/reliability and limited g‑coverage rather than absence of any cognitive effect. — If the Wordsum underestimates g, many published associations between 'intelligence' (as measured in the GSS) and social outcomes may be attenuated or mischaracterized, affecting policy and cultural narratives that rely on those findings.