WORDSUM Vocabulary Test

The ten-word vocabulary test that the General Social Survey has given to representative samples of American adults since 1974 (in the GSS it is the variable WORDSUM). Each item shows a word in capitals and five candidates; pick the one that comes closest in meaning. The items were drawn by Thorndike from the Gallup-Thorndike vocabulary test of the 1940s. Short as it is, the score is a fair proxy for general intelligence: its g-loading is about 0.65 as observed and about 0.8 once corrected for the unreliability of a ten-item test (Crémieux, How g-loaded is the Wordsum?, using the same items in the NLSY). About two minutes.

The score is the number correct out of ten. Your percentile is against non-Hispanic White respondents in the 2018 GSS (N = 978, survey-weighted), the last round in which everyone was asked all ten words in a face-to-face interview; from 2022 the GSS moved online and dropped three words for half the sample, which makes the later rounds hard to compare. The GSS offers a “don’t know” option, scored as wrong; this version leaves it out and has you guess, which is worth about a third of a word on average.

The IQ band converts the percentile to an IQ scale within that reference group. Ten items cannot pin a score down finely, which is why it is a range.