II

THE COLLECTION OF DATA

Subjects. The data from which the following observations. were drawn were collected (January-May, 1924) by administering to approximately unselected family groups an hour-long team of tests selected from the more reliable "intelligence" and achievement tests. Factors making for possible departure from strict unselection were

(1) The fact that the investigation was necessarily conducted in territory close enough to the University to be accessible by automobile in an hour or so; this means that the level of intellectual ability was somewhat higher, in all likelihood, than the general average even of California.

(2) The method of selection of subjects. The precautions taken here proved unnecessary, but require citation: The geographical area from which subjects could be drawn was delimited by the physical necessity that the examiner should be able to travel conveniently by automobile from Palo Alto to the home in which a test was to be made, within a time which would still allow ample testing time before the bedtime of the younger children concerned. Within this area were situated, in Palo Alto, the Union High School, the Walter Hays School, the Channing Avenue School, the Lytton Avenue School, and the Stanford Campus School; in Mayfield the Grammar School; in Menlo Park the Grammar School, the Runnymede School, the Roman Catholic Convent School, and the Las Lomitas School. A few families living in Menlo Park but sending a child to the Sequoia Union High School in Redwood City were also tested. There are also in this area several private boarding schools; these were canvassed for possible day or local pupils, but none were found.

In each school a list was made of all pupils born between two dates a year apart; these dates were then adjusted so as

to make a child born at the middle of the said year exactly thirteen years old at the estimated middle of the testing period for that vicinity; that is, these subjects (designated hereinafter as "basics") were to have an age mean at thirteen years, no months, and a range of one year. Most of these refinements proved unnecessary; but it was planned at the time to use only the basic group in the computations, so that it was desirable to include all basics in the area;1 and to choose the basic age old enough to be as nearly comparable with their parents as possible, yet young enough so that all would be found still in school.

Requests were tendered personally to the parents of these children to cooperate with the investigation by devoting one evening to a test of the entire family group-every accessible person related to the child, and every person not related but willing to be tested. These requests were for the most part in the hands of an intelligent and tactful young woman field worker, and were made in a straightforward way as an invitation to assist in an enjoyable manner and without obligation in a strictly uncommercial research.

The actual administration of the tests calls for no extended comment. The experimenter ordinarily arrived at the home about 7:30 p. m., and proceeded immediately to the testing. The subjects were comfortably seated, provided with two of the experimenter's pencils and a lap-board, and informed in a few words (including usually the reading of the information printed on the first page) as to the purpose and mechanism of the test. Time was taken with a stop-watch. The auxiliary information was not asked for until the completion of the testing. An endeavor was made to use a little time between the individual tests to restore a sense of ease and the play spirit, unless some reason for the contrary existed. The testing was begun in January, 1924, and finished in the following May.

"Number of basics secured, 141; mothers, about 100; fathers, about 90; sibs, about 280.

(3) Refusals to cooperate. These ran in the neighborhood of 50%,2 and seemed to be scattered rather evenly over all levels both of mental ability and social competence. It is possible that refusals may have been motivated in part by a conscious or half-conscious recognition of marked discrepancy between parent and child in the abilities tested; such a supposition, however, seems to be crediting the subjects with a good deal more familiarity with and "faith" in the tests than they probably possessed. The tests were administered by the investigator and one assistant, under uniform conditions as to time (evening) but of course under a wide range of environments and special circumstances. The invalidating effect of this factor is a matter of conjecture. It may be pointed out that it is perhaps not so great as commonly supposed; the ideal test situation is perhaps one such as a private-school or small public school psychologist may occasionally be able to achieve -the children know him, like him, play with him, as part of their ordinary school life; they come occasionally to his office and play with the materials there, and visit with him; presently their tests have been completed and they remain quite unconscious of the occurrence of anything unusual-except that they may want to do it again soon. At the other extreme is the group-tester of a military turn of mind, who marches children into a strange place or one in which only state affairs occur, and conducts the test as though it were a battalion parade; in this case distraction in the sense of noise, activity not connected with the test, etc., is reduced to a minimumbut the element of possible psychological, i.e., inner, emotional, distraction, has been completely overlooked. The optimum would seem to be complete emotional freedom from strangeness. Somewhere between the best and worst is no doubt the usual test as given; the teacher gives a group test in the home. room-the environment is the same, the event somewhat exciting and disturbing; or the psychologist takes the child to his office-somewhat worse perhaps; the psychologist is apparently "Families approached, about 210; cooperating, about 105.

a friend, but the environment is new and strange, and the friend obviously wants something. It is not, probably, sufficient to lump all distractions, inner and outer, together and say that such evidence as we have indicates that its effect on score is negligible; on the contrary, we seem to know very little about it, and common sense would indicate that these inner distractions may be of considerable importance. Just where, on such a scale, to place the present procedure is not easy to discover; but much might be said for placing it well up, on the basis that the subject is on his own ground, that the surroundings are those of even minimum distraction for him, and that these considerations plus the endeavor of the tester to put him at ease and to bring him to a perfect understanding of what is required of him tend to produce more nearly uniform conditions of performance than had they been standardized from the standpoint of the tester alone. In the more conspicuous cases, such as marked language handicap, doubtful performances were disregarded in the evaluation.

The Testing Instrument. The abilities tested were standard ones, comprising

(a) Antonym-synonym discrimination and discrimination of identical and non-identical digit and name pairs (Tests 1, 11).

(b) Completion of symbol and number series (Tests 4, 2). (c) Science-nature and history-literature information (Tests 9, 10).

(d) Arithmetic reasoning (Test 3).

(e) Vocabulary recognition (Test 5).

(f) Recognition of the relationships of simple geometrical forms (Test 6).

(g) Perception of analogies presented verbally (Test 7). (h) Digit-symbol substitution (Test 8).

Tests 5, 9, and 10 (Vocabulary, Information) are from the Stanford Achievement Test, as they stand.3 Tests 2, 6,

"The times in all cases were so adjusted that the brightest university students could barely finish; this resulted in a general shortening from the times under which the original tests are ordinarily administered.

and 8 (number series completion, form combination, digit-symbol substitution) are from Army Alpha and Beta, as they stand. Test 3 (arithmetic reasoning) is from the National Intelligence Tests, as it stands. Tests 1, 4, 7 and 11 (opposites, series completion, analogies, comparison) are from the National Intelligence Tests, adapted in 1, 4 and 7 by the addition of harder elements (from Terman Group Test and Army Beta) and in 11 by the omission of pictorial elements. The directions were somewhat varied to suit the immediate conditions, but the aim in all cases was to see that every subject (so far as possible) understood what was required of him, but that nothing was actually done for him.

The reliabilities of these tests, taken individually, are, with minor exceptions, between 8 and 9 for groups ranging in size from about 40 (sister-sister, tests such as history-literature information) to about 110 (sons of mothers, tests such as form-combination); that of the entire battery is, accordingly, in the neighborhood of .95. Although such a refinement later proved to be unnecessary, separate reliabilities were computed (by correlation of converted-score halves-see infra, page 258 and "Brown's formula" for each group involved; e.g., for fathers of boys and sons of fathers in the fatherson correlation, for younger sisters and older sisters in the sister-sister correlation, etc., and for each test separately. The detailed reliability coefficients resulting are cited on page 260.

Heredity Techniques. It was originally hoped to derive from the data some clear evidence bearing on mental inheritance; there is, however, no certain or simple technique for excluding, or estimating, the influence of environment. The most obvious method-that of comparing the coefficients for groups alleging a longer and a shorter time spent with their children -is unsatisfactory because (1) clinically the data seem unreliable; the parents evidently are only guessing at the truth because there is no way of arriving at a genuine approximationthey do not know how much time they spend with their chil*Kelley [13], p. 206, formula 158.