For a certain common type of experimental situation, the better the subject's performance — i.e., the more skilled he is, or the easier his task (perhaps due to good human-engineering design) — the more markedly skewed his distribution of scores tends to be and the more nonrobust parametric tests tend to be. So when performance conditions are optimal, test conditions tend to be pessimal and vice versa.
Subjects were given one, two, or three cues with which to make an either-or decision. Certain hypotheses were formulated to describe the subject's thought process in utilizing the multiple cues, and mathematical models were constructed to simulate them. The models were then used on data for the single-cue case to predict performance in the multiple-cue case. Predicted performance "data" were then compared with actually observed data for the same multiple-cue case, thus testing the predictive validity of the mathematical model and the tenability of the corresponding hypothesis.
When frictional resistance is heavy, the optimum knob diameter is about 2 in. The curve relating turning time to knob diameter is roughly U-shaped, and diameters outside the range of 1 3/4 to 2 1/2 in. are significantly inferior to the 2-in.-diameter knob. When frictional resistance is reduced to a moderate level, the curve becomes essentially flat in the region from 3 1/4 in. (the largest diameter tested) to 1 in., but rises steeply and significantly with decreasing diameter at diameters smaller than 1 in. (i.e., 3/4 in. and 1/2 in). There is an indication that if friction were reduced to a very low level, the curve would flatten out still further to include in its flat region all practical diameters smaller than 1 in. Evidence is cited in support of this indication. Evidence is also presented to the effect that the above findings are valid over a wide range of rotary inertias. Reach time was independent of frictional resistance, but increased with decreasing diameter at diameters smaller than 1 1/2 in.
Reach time, turning time, and inadvertent touching of adjacent controls (i.e., errors) were measured while a standard setting was made with one of several closely spaced knobs. The variables manipulated were spacing between knobs, knob diameter, and knob configuration. Performance improved rapidly with increasing distance between knob edges up to an interperipheral distance of 1 in., after which performance continued to improve but at a much slower rate. For equal amounts of panel space consumed by several closely crowded knobs, ¼-in. diameter knobs were more nearly error free than were the larger diameter knobs tested. For equal distances between knob edges, however, performance improved with increasing knob diameter. These results apply only to knobs capable of being operated by moderate torque. It was found that the frequency with which a crowding knob is inadvertently touched is strongly affected by the angular position which it occupies with respect to the operated knob but is practically independent of the presence of other crowding knobs at the same distance from the operated knob.
A series of experiments was performed to determine the minimum allowable dimensions of circular, non-detent knobs, mounted on concentric shafts, when frequent inadvertent operation of adjacent coaxial knobs cannot be tolerated. A standard setting was used, and measures were taken of reach time, turning time, and inadvertent touching of adjacent coaxial knobs. Manipulated variables were thickness, diameter, and difference in diameter between the operated knob and the adjacent knobs. It was concluded that if three knobs are to be concentrically ganged and if the middle knob is about 2 in. in diameter, (a) the diameter of the front knob should be at least 1 in. smaller and that of the back knob 1¼ in. greater than that of the middle knob, and (b) the front and middle knobs should each be ¾ in. thick, whereas the back knob may be as thin as ¼ in.
Both objective measurements and subjective ratings were made of the degree to which each of 18 widely differing gloves possessed each of the following characteristics: tenacity (i.e., resistance to sliding over a grasped surface), snugness of fit, suppleness, and protection against injury to the enclosed hand. Twenty-two subjects performed each of five different control operations while wearing each of the 18 gloves and while barehanded. For both objective and subjective measurement of the characteristics, it was found that degree of tenacity is correlated with speed of gloved operation of on-off controls, that amount of suppleness is correlated with rapidity of gloved operation of adjustable controls, and that increasing snugness of fit improves operation time for both types of control. The difference between operation times, gloved and barehanded, was found to depend strongly on the type of control operation required.
Five types of control (push buttons, toggle switches, knobs, horizontally operable levers, and vertically operable levers) were operated at room temperature with the hand clothed as follows: no glove, wool glove, double glove, i.e., leather glove over wool glove. Operation time was measured. It was concluded that the effect of gloves on control operation time depends on the type of glove worn, the physical characteristics of the control, and the type of control operation required.
Tactual coding of knobs by use of bizarre shapes is frequently achieved at the expense of manipulability and setting precision, which appear, in many cases, to be optimal when knobs are cyclindrical. In order to be able to maximize both discriminability and manipulability, certain parameters of cylindrical knobs were investigated as bases for tactual coding. Rim surface, diameter and thickness were all found to be useful for this purpose. When feeling one of two knobs whose pictures were before them, subjects rarely (less than 1% of the time) identified the wrong picture as the felt knob in any of the following situations: diameters differ by 1/2 inch or more, thicknesses differ by 3/8 inch or more, rim surfaces belong to different ones of the three families: smooth, fluted, knurled.
Seventy-five male college students and twenty-five human engineering psychologists were given a questionnaire presenting diagrams consisting of three concentrically ganged knobs and three dials which they were told the knobs operate. They were asked which dial they thought should be operated by each of the three knobs. Knob-dial associations were obtained with dials in horizontal and vertical arrays above, below, to the left of, and to the right of the knobs, and with dials differing in size, shape and distance from the knob axis. Knob-dial associations were found to be influenced by all of these factors except dial shape. Associations which were both strong and relatively unrivaled were found for dial position in a horizontal array (except when the array is to the left of the knobs), and for dial size. Subjects associated the spatial knob progression, front knob to back knob with the spatial dial progression, left dial to right dial and with the dial size progression, smallest dial to largest dial. Strong, but strongly rivaled, associations were found for dial position in a vertical array and for dial distance from the knob axis.