Reaction-times / by Charles S. Myers.

  • Myers, Charles S. (Charles Samuel), 1873-1946.
Date:
[between 1902 and 1909?]
Catalogue details

Licence: In copyright

Credit: Reaction-times / by Charles S. Myers. Source: Wellcome Collection.

    10/24 (page 212)
    mean1. In the first place, its value is very considerably influenced by even a single aberrant reaction. Secondly, while the limiting value of a shortened reaction is zero, that of a prolonged reaction is infinity. Hence the fluctuations on one side of the average are liable to be very much greater than on its other side2. The median, on the other hand (that value which stands midway when the data are arranged in order of magnitude), is considerably less affected by these factors. Aberrant reactions affect its value only to the same extent as do normal reactions. Consequently the median, especially when derived from a fairly large series, may be held to indicate a truer mean than the average. Following the auditory and visual reaction-averages and medians, are two pairs of columns respectively marked m. v. and m. v. q. The columns of mean variations (m. v.) express the average deviation of the ten reaction-times from their average. The columns of mean variations of quartiles (m. v. q.) give the figures obtained by finding the median of each half of the series when arranged in order of magnitude of data, and by then taking half the sum of the difference of each of these two medians (quartiles) from the median of the entire series. The visual choice-reactions have been arranged somewhat differently. The columns headed median and of give respectively the median of, and the total number in the series. Few series were without one, two, or even more erroneous reactions in which the individual reacted with both hands or with the wrong hand. Their number will be discussed later. The column headed av. of 10 gives the average of those ten con¬ secutive right reactions in the series which agreed best with the median. As a rule, there was hardly any choice as to which group of ten should be selected. Where a choice was possible the result usually varied little, from whatever group the average was taken3. The columns headed v. c. contain a quantity which has so far received no name in psychological statistics. The variation-coefficient4, expresses the ratio between the mean 771. V. X 100 variation and the average. It is obtained by the formula v. c. — ——-. Its im¬ portance lies in the fact that the mean-variation depends not merely on the fluctuations of individual data from their average, but also on the actual magnitude of that average, varying directly with its value. Thus if a reagent, whose average reaction-time to an auditory stimulus is 120°, shew a mean variation of 10°, and if, reacting to a visual stimulus his average reaction-time be 180° and his mean variation be 15<r, he is reacting with an equal degree of constancy in each case, although the absolute values of the mean variation are not the same. The Arabic and Roman numerals in the columns headed series are easily explained by an example. For instance, 3 ii implies that the reactions in question were performed at the reagent’s third sitting, and that at that sitting they had been preceded by some other kind of reaction (viz., by 3 i). 1 I have used the word ‘ mean ’ throughout this section in the sense of the ideal average. 2 Cf. E. Krapelin, Ueber d. Beeiiiflussung einfacher psychischer Vorgdnge durch einige Arzneimittel, Jena, 1892, S. 23. 3 In two instances it was impossible to obtain more than nine consecutive correct reactions. The averages appear in Table XXXIV, but are preceded by the figure <9). 4 The term has been already employed by Professor Karl Pearson in a sense only slightly different from that in which it is used here.