Sutarlinah Soekadji
(Submitted 25 August 2015)
(Published )


The answers of those questions are found through computerized guessing simulations, using TPA items answered by 1395 examinees in response to instruction that discouraged guessing. The results show that guessing shortened the score effectivity range of the subtests, and increasing average subtest scores, which means increasing the average index of difficulty levels (decreasing item difficulties). Subtest scores penalized for guessing tend to be justly corrected, and subtest scale scores based on norms for guessing option B only, or C only, with or without penalty yield high correlations ( .92 to .99) with the original scale scores (without guessing nor penalty). Using random option guessing slightly lowers the correlations (.88 to .97). Guessing penalized or unpenalized yield the same slightly lower alpha reliability (around 0, 03). Thus guessing should be encouraged, especiallyfor the benefit of those who are too anxious to guess, or unwilling to take the risk. Even in the process of scoring, guessing could be provided for blank responses.

Keywords: Guessing, penalized, unpenalized.

Full Text: PDF

DOI: 10.22146/jpsi.6993


  • There are currently no refbacks.

Copyright (c)