Variations in methods of calculating reliability of duration data

W. Joseph Wyatt, Edward J Callahan, Joel Michael

Research output: Contribution to journalArticlepeer-review


The traditional percentage and correlational methods of estimating reliability of duration recording were compared to the reliability obtained with an event‐by‐event examination of observers' records in which the actual percentage of time that the observers were in agreement was calculated. While observing standardized videotapes, eight volunteer undergraduate students recorded “studying” using a specific definition and “not eating” using a nonspecific definition. The order of persentation and level of definition specificity were reversed for eight additional volunteer observers. Recordings were done on an event recorder. It was found that the traditional percentage reliability scores were significantly higher than actual (event‐by‐event) observer agreement for both behaviors, at both levels of definition specificity. Correlational reliability coefficients were significantly greater than event‐by‐event scores for specific definitions only. Implications of the findings include the possibility that duration recording, which does not involve a permanent product (e.g., when using a stopwatch), may be periodically checked for reliability with an event recorder, or that interval recording may be used.

Original languageEnglish (US)
Pages (from-to)473-479
Number of pages7
JournalPsychology in the Schools
Issue number4
StatePublished - 1985
Externally publishedYes

ASJC Scopus subject areas

  • Education
  • Developmental and Educational Psychology


Dive into the research topics of 'Variations in methods of calculating reliability of duration data'. Together they form a unique fingerprint.

Cite this