THE EFFECT OF ANSWER SHEET FORMAT ON TEST PERFORMANCE BY CULTURALLY DISADVANTAGED FOURTH GRADE ELEMENTARY SCHOOL PUPILS

not reported in the retrievable record; the paper is a two-page research note (J Educ Meas 8(4), 289-290)

US fourth-grade pupils described as culturally disadvantaged

A two-page note comparing answer-sheet formats. Sample size, randomisation and power are not recoverable from the abstract, and a null from an underpowered two-page note is close to uninformative about the size of any real effect. Recorded because it is the most direct empirical test of the mechanism this topic invokes, and because recording a null that cuts against the topic's own hypothesis is the point of the disposition system.

Answer-sheet format had NO effect on the mean test performance of fourth-grade pupils. This is a boundary marker on the mis-bubbling explanation for an anomalous score, and it should be read narrowly: it is a null on the AVERAGE effect of a format change across a group, which is a different quantity from the probability that one particular child mis-tracked one particular answer sheet. A mean null is compatible with a small number of individually catastrophic administrations, and the archive should not read it as evidence that clerical failure does not happen.

Not applicable - a null on a test-administration format manipulation.

No identified replication in over fifty years. The wider answer-sheet-format literature is small, old, and reports both nulls and small effects.