Per cent solved is a variable, not a constant unit of difficulty, and is not comparable across groups¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 1919 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? Per cents state order of difficulty exactly but cannot establish exact amounts of difficulty, because increments of difficulty are unequal across the percent range and per cents lack a common unit comparable between groups. → Herring 1919

Evidence¶

Herring 1919¶

Herring, J. P. (1919). Derivation of a scale to measure abilities in scientific thinking. The Journal of Educational Psychology. https://psycnet.apa.org/journal/edu

q2 · i? · design · r2

Theoretical argument in the article's methodological section: per cents indicate the order of difficulty but are a variable unit, so they are safe to use only once their variations are determined. No effect size is printed.

"The percent has often been used to measure mental abilities just as if it were a constant unit. It is a variable, and as such, it can be used safely only when the amounts and effects of its variations have been determined."

Discussion¶

Related Claims¶

• Problems whose per cents solved do not increase monotonically across grades are eliminated as not measuring grade-to-grade progress — related
• In the Yerkes-Bridges alteration of the Binet Scale, equal point increments are assigned to digit-span steps whose growth intervals, as placed in the Binet Scale, are approximately 1, 3, 3, and 4 years. — related
• Yerkes-Bridges coefficients of intelligence of 1.20 or more are attained by very different percentages of children at successive ages (34.5% at ages 4-5 down to 0% at ages 16-17 and 18-on), so they do not represent the same degree of brightness at all ages. — a narrower finding that bears on this claim
• Arbitrarily assigned point values in the Yerkes-Bridges scale yield implausible growth-rate results, such as intellectual ability increasing more than fifty per cent during the fourth year and only approximately fifteen per cent during the seventh. — related
• A problem's diagnostic value as a measure of general ability can be indexed by the slope of its pass-rate ogive at the 50% point — related
• Scaling is a sine qua non for experimental use of tests, and unscaled tests interpreted as scaled do harm — a narrower finding that bears on this claim