Citation

Graham JL. J. Exp. Psychol. 1937; 20(6): 597-608.

Copyright

(Copyright © 1937, American Psychological Association)

DOI

10.1037/h0060831

PMID

unavailable

Abstract

In this bar graph analysis, six variables were analyzed: horizontal or vertical mode of presentation, coarseness or fineness of the scale units, spacing distance between the bars, width of bars, bar length, and nearness or remoteness from the scale. The factors compared were equated either in identical comparisons or in groups where the factors and degrees were equated (but not in identical combinations of degrees). The probability of an illusory trend holding consistently was expressed in terms of its ratio to the sigma of the difference. Results reveal that the factors which should give the maximum amount of overestimation in combination are short bars, presented with wide spacing distances, coarse scale units, and narrow bars viewed vertically if the most important comparison is a long bar, and horizontally if it is short. The bar positions remote from the scale, also, give greater overestimation. (PsycINFO Database Record (c) 2012 APA, all rights reserved)