Open Access

JSON Export

{"created":"2022-01-31T14:17:51.950928+00:00","id":"lit28772","links":{},"metadata":{"alternative":"Studies from the Yale Psychological Laboratory","contributors":[{"name":"Scripture, Edward W.","role":"author"}],"detailsRefDisplay":"Studies from the Yale Psychological Laboratory 2: 101-104","fulltext":[{"file":"p0101.txt","language":"en","ocr_en":"REMARKS ON DR. GILBERT\u2019S ARTICLE,\nBT\nE. W. Scripture.\nIn preparing for publication Dr. Gilbert\u2019s Researches on the mental and physical development of school-children, several matters seemed to me to require a further statement.\nSuggestion-test. The large and the small standard are of the same weight, namely, ic=55e. The diameter of the small standard was d= 2.2om, and of the large standard J)= 8.2m. The weights of the 14 blocks were successively\ni?,=15S; ^=20\u00ab; .\t.\t.;i>14=80*,\nthe difference between any two successive blocks being pi\u2014pj\u20145B. The diameter was 8=3.5s for all. The result of the child\u2019s two judgments was that w=pk for the small standard and uo\u2014pi for the large standard. The amounts of difference vk-=pk\u2014w and v,=.p,\u2014w can be taken as measures of the effect of the different sizes of the standard and the blocks of the 14-series. As all blocks were cylinders of the same length, we can, if we neglect effects of contrast of length to diameter, express the difference in size by the areas of the ends. Thus the visual suggestions can be indicated by J(ir8J\u2014ird*) \u2014Uk and J(7r2>a\u2014!r82)=M; respectively. As the differences vk and v, disappear when the blocks are lifted without being seen, we can put\nvk\u2014fk(uk) and \u00ab,=/,(\u00ab,).\nFull expressions for fk and f for a given individual would give the law of suggestion for the given case. The determination of this law was not the object of the investigation, in which uk and U, were taken constant. For the sake of brevity the difference between the large block and the small block was taken as the amount of suggestion ; thus the constant suggestion was taken as\nS=Uk + W;=i(7rZ?a \u2014 7rcf).\nThe result of this suggestion is\nH=Vk + V,..\nIf by A we indicate the age then","page":101},{"file":"p0102.txt","language":"en","ocr_en":"102\nJS. W. Scripture,\n&=f(S, A), and by taking S \u2014 constant, we obtain\nH=f{A)\nor the force of suggestion as a dependent on age.\nMethod of computation. The\tmethod of computing the results\ncan be thus indicated. Let the results be\t\n\u2022\t. \u2022,\nK K \u2022\t\u2022 -, K\n\u2022\t\u2022 -, L\nwhere the letters a, b, .\t. I refer to different children and the\nindices 1, 2, .\t. n refer to experiments on the same child.\nIf we express by R,z=f(x) the fact that R is determined from the i powers of x, then for the median\twe have the general expression\n=/\u201e(*\u201e x\u201e\t.\t.\txn)\nand for the arithmetic mean\nA*=fi(xi> xr \u25a0\t\u25a0\t\u2022> *\u00bb)\u2022\nFor the reasons indicated on p. 23 the mean variations are deter-. mined as f.\nWe have thus for the individual children the medians\nc.=/o(\u00ab,\u00bb\t\u2022\t\u2022\t\u2022\u00bb\t\u00ab\u00bb)\ncb=fn{b\u201e bt,\t.\t.\tbn)\n\u00b0i\u2014h\u2019 \u2022 \u2022 \u2022) y\nThe mean variations for the individual children will be\nK-O, \u2022 \u2022 \u25a0. [\u00ab\u00bb-<])\nIA-Cd; \u2022 \u2022 \u2022> [\u00e0n\u2014Ct])\nDi\u2014fAUi ~ci\\> P\u00bb - cJ> \u2022 \u2022 \u2022>\n[L -c,])","page":102},{"file":"p0103.txt","language":"en","ocr_en":"Remarks on Dr. Gilbert's article.\t103\nThese mean variations can be regarded as psychological quantities expressing the accuracy of each child\u2019s judgment. The median accuracy of judgment for the particular age r will be\nDr=f0(Da, Db, . . ., D,).\nThis can be called the mean personal insecurity.\nThese individual quantities are quite different from the statistical mean variations, which are taken in order to show the homogeneity of the children for any given age. The median for a given age r is found by\nC,=/\u00ab(c<o\tc;).\nThe mean variation of the children from the general mean will be\n-Dr=/x([tf\u00ab\u2014CJ> [Cb~C'\\ \u2022\t\u2022\t'\u00bb\nLimited series. In tests (1), (2) and (3) it was discovered too late that the series of blocks and colors did not extend far enough to include extreme cases. In each table there is a column with the percentage of cases where the least perceptible difference or the force of suggestion went beyond the limits of the apparatus, and in the charts dotted curves are given for these cases. Gilbert states that the column of mean values does not give the quite correct result unless it be considered in relation to the column of percentages just mentioned. This statement would be true if the mean value used had been the arithmetic mean but is not true for Gilbert\u2019s results as the mean value used was the median. As explained on p. 32 the separate values influence the median only as being above or below it. The median child remains just the same whether an extreme child exceeds the recording power of the apparatus or not. The columns of means and their curves are thus completely correct in themselves. The coltimns of percentages of no-discrimination give really another representative value of no particular use in itself but quite important when compared with the medians as indicating the form of the frequency-curve and the even course of the curve of results according to age. In this respect it is as important as the mean variation.\nThe column of these percentages, although having no influence on the median, does have an influence on the mean variation. The mean variations are all too small by a quantity \u00a3 following the law t,\u2014f (P). If the positive and negative mean variations had been calculated for the (100\u2014P)^ of the cases separately, a deduction on the assumption of (11), p. 12, would have rendered it possible to","page":103},{"file":"p0104.txt","language":"en","ocr_en":"104\nJE. W. Scripture.\ncalculate the actual mean variation for the whole number independent of the limitations of the apparatus. The gain, however, would have been incommensurate with the labor; the mean-variation-curves would resemble those actually given and would simply be steeper at the left.\nThe columns headed PB and PG give us the means of answering the question as to when the difference between the boys and girls recorded for any given age is to be considered as a true difference between the sexes or as merely the result of the finite number of cases considered.\nThe method is as follows. Let the percentage of the total n boys within the limits of the test be jo=(100\u2014PB)<fo and of those beyond be q = PB<fo ; likewise for n girls let the percentages be p'= (100\u2014PG)</c, q'=.PG<fo. According to a well-known theorem1 we can assert with a probability of 4>(y) that the two classes are different, provided\nwhere 4>(y) is the function expressed on p. 20.\nI have tested some specimen cases in this way and find that for these three tests it cannot be asserted in many cases with a practical certainty of $(y)=0.999978 that there is a real difference between the two sexes.\nAnother test is furnished by Bayes\u2019s theorem used on p. 38. This theorem can be applied to all the tables.\nIn a like manner the differences between successive ages can be tested.\n1 Illustrated In Lexis, Einleitung in die Theorie der Bev\u00f6lkerungstatistik, 103, Strassburg 1875.","page":104}],"identifier":"lit28772","issued":"1894","language":"en","pages":"101-104","startpages":"101","title":"Remarks on Dr. Gilbert\u00b4s article","type":"Journal Article","volume":"2"},"revision":0,"updated":"2022-01-31T14:17:51.950933+00:00"}