Investigators' Instructions and Experimenters' Misrecording of Questionnaire Data

As defined by Silver (1968), an experimenter is one who administers the experimental treatments, records data, and may do the statistical calculations. The investigator is the person who initiates the study, supervises the experimenters, and interprets the data. The experimenter misrecording effect refers to the failure of the experimenter accurately to record the data. Such effects have been identified in a number of studies (see Barber & Silver, 1968a, 1968b, for a review). The present focus was whether such effects can be induced by the investigator's instructions. The 12 Ss were college students enrolled in introductory psychology courses. The experimental task was to score questionnaire data which had been provided by other subjects. All 800 items on the questionnaires were independently scored and verified by the authors (one Ph.D. and 2 psychology majors, one a B.A.). Identical protocols were given to four Ss (experimenters; undergraduates in introductory psychology) in each of three groups. Ss in Group 1 were asked simply to re-check the scores of previous scorers. Those in Group 2 were told by the main investigator and first author that the scores were too high as originally scored and not in accord with his expectations, and that they must have been mis-scored. Ss in Group 3 were asked to confirm the data and were told that their work would be subject to an accuracy check. Each S received 20 protocols with 40 items on each. The number of errors (disregarding direction or magnimde) was tabulated independently and verified by the second and third authors. l'he total number of errors (out of 800 possible) was for Group 1 21 errors, for Group 2 23 errors, Group 3 22 errors. These results suggest that the experimenter misrecording effect is minimal regardless of the investigator's instructions. In the case of borderline statistical significance these errors could be large enough when the data were treated by analysis of variance, for example, to alter the conclusions. Although there was little evidence for misrecording, the present data were easily scorable. A second limitation is that the instructions and investigator-experimenter interactions may have been too brief to induce such effects

Library science · Mathematics education · Psychological research · Subject (documents · Task (project management · Academic and Historical Perspectives in Psychology · Applied Psychology · Computer Science · Psychology · Social Psychology · Statistics Education and Methodologies

• How often are our numbers wrong

  Robert Rosenthal•American Psychologist•1978

  Add to list

• Fact, fiction, and the experimenter bias effect

  Theodore Xenophon Barber, Maurice J Silver•Psychological Bulletin•1968

  Add to list

• Pitfalls in data analysis and interpretation

  Theodore Xenophon Barber, Maurice J Silver•Psychological Bulletin•1968

  Add to list