Error correction, Scoring, Test reliability, Testing problems
The following comments are the result of preliminary analysis of some errors made by a random sample of teachers in marking the test papers of pupils who, in 1946, took part in the testing programme in connection with the Curriculum Survey being conducted by the A.C.E.R.
All schools where children did the tests were asked to select a certain number of papers at random from each grade and forward them to the A.C.E.R. The numbers sent in from each school were roughly proportional to the size of the school. Thus large schools sent in more papers per grade than the smaller ones. [p.1, ed]
The tests in which the error was greatest were Reading for Meaning, Sentence Structure, Word Usage, Arithmetical Processes and Spelling. [p.3, ed]
Owing to the limitation of the re-marking to the papers from one Grade, there were insufficient papers marked in each test in each State to justify presenting detailed interstate comparisons test by test. For each state however, an examination was made of the net effect of marking errors on each test. In those cases where this appeared to give an average error of 1 mark or more to the papers marked, a further check was made. In all but one of these, it was found that the large error was due to misinterpretation of instructions for scoring in one or two schools.[p.5, ed]
Australian Council for Educational Research. (1947). Errors in the marking of printed tests (ACER Information Bulletin No. 9). https://research.acer.edu.au/ib/9/
Copyright Australian Council for Educational Research 1947
Place of Publication