SPECIAL
ARTICLE
AMERICAN
BOARD
OF
PEDIATRICS
The written examination of January 11, 1952 was taken by 462 candidates, a larger
number of candidates than on any previous occasion. A sample of the total number was
selected for statistical analysis, the sample being made up of the grades earned by 284
candidates who by July 1, 1952 had stood the oral examination and could therefore be classified as having passed or failed the over-all examination. Two additional candidates
who appeared for oral examination were given a ‘‘condition’ ‘ in ‘‘Growth and Develop-ment’ ‘; the grades earned by these candidates are not included in the analysis. Of the total of 286 candidates the final results show : passed, 238 ; failed, 46 ; conditioned, 2.
The examination was divided into five sections, designated as A, B, C, D and E.
Sec-tion A was composed of 100 false and true statements. All of these questions had been
asked in previous Board examinations and all had been answered correctly by from two-thirds to three-fourths of the candidates. They were so selected to conform with the opinion of the Committee that the purpose of its examination is to discover the poorly qualified candidates and not to assign rank-order preference to the best qualified
candi-dates.
Section B was designed to test the candidate’s knowledge of infant feeding. Twenty
questions, answerable by ‘‘yes” or “no,” were presented in connection with three feeding
formulas to be assessed by the candidate. The questions of both sections, A and B,
offered the candidate only two choices (false or true ; yes or no) ; a correct answer to any question was therefore assigned one-half of a grading point and a perfect score on
Sections A and B would have been 60.
Section C consisted of 40 incomplete statements followed by five suggested comple-tions. In common parlance such items are spoken of as ‘‘multiple-choice questions.’ ‘ For
a candidate completely lacking knowledge the chances are that one out of each five
ques-tions could be answered correctly by guessing. In grading the examination this chance
was ignored and each correct answer was assigned 1 grading point. Sections A, B and C
therefore furnished 100 grading points.
Section D was made up of 20 sentences each consisting of two parts : a statement and a reason for the statement. The candidate was presented with a choice of five different responses to each question. He was requested to enter the letter ‘‘A’‘ if both statement
and reason were true and related as cause and effect ; to choose the letter ‘‘B’‘ if both statement and reason were true but not related as cause and effect ; to write the letter
“C” if the statement was true but the reason false ; to select the letter “D” if the
state-ment was false but the reason an accepted fact or principle; to inscribe the letter “E” if both statement and reason were false. Since an answer to each question demanded two correct responses from each candidate, a value of 2 grading points was assigned to each
correct answer. The Committee believe that questions of this type demand more of the
candidate than simple possession of factual knowledge, that in addition correct
re-sponses require clear thinking regarding cause and effect relationships.
Section E was made up of 10 questions, each consisting of five numbered words or
st. QUARTILE 47.4
MEDIAN 55.2
3rd. QUARTILE 62.9
MEANIRD. 47.51
MEAN 55.08
MEAN I PD. 62.66
3O
I’
MID-POINT OF EACH GROUPING OF GRADES
722 AMERICAN BOARD OF PEDIATRICS
each correct response was assigned 2 grading points. Four of the items in the numbered
left hand column suggested a feature common to one of the three lettered items in the
right hand column. One of the items in the left hand column did not belong in the same category. The candidates answer embraced a number and a letter: the number identified the item which did not belong and the letter the item characterizing the other four num-bered items. The Committee believe that correct responses to this type of question also
demand clear thinking.
The total grading points from the 190 questions of all types were 160. A candidate’s
U) Lii 4 Lii I.-4 z U) Lii z > Lii I C.) 4 U) Lii I-4 0 0 z 4 C-) Li. 0 Lii z
AMERICAN
BOARD
OF
PEDIATRICS
WRITTEN EXAMINATlON JANUARY 1952
grade was obtained by dividing his score in grading points by 1.6.
The distribution of the grades earned by 284 candidates is presented in the diagram.
The lowest grade was 23.1 and the highest was 77.5. Mean and median grades were
both approximately 55. First inspection of the distribution columns of the diagram
AMERICAN BOARD OF PEDIATRICS 723
grades. A chi-square test of the significance of the deviations in the frequency of grades
between those observed and those calculated on the assumption of a normal distribution
does not reveal a statistically significant difference. The odds are 3 in 20 that a deviation as great or greater could have resulted from chance alone. Stated the other way the odds are 17 in 20 that the examination is beginning to realize its objective in range sensitivity.
As in previous years the reliability of the examination has been assessed by comparing
the grade made by each candidate on his odd numbered questions with that recorded from his even numbered questions. The comparison reveals a ‘‘probable error of estimate”
(P.E.m) of 2.78. The figure must be compared with that of 3.33 for the 1951
examina-tion and of 3.39 for the 1950 examination. The increase in reliability is ascribed largely
to the greater number of grading points on the 1952 examination. The examinations of
1950 and 1951 each furnished 100 grading points; this examination of 1952 yielded 160 grading points ; the forthcoming examination of 1953 will have 250 grading points.
The reliability of the examination in predicting whether or not a candidate would
pass the deciding oral examination is presented in table 1.
TABLE 1
Grade on Written Examination Decision after Oral Examination
Passed Failed Passed
(no.) (no.) (%)
Lowest quartile 36 31 54
Second quartile 61 13 82
Third quartile 60 4 94
Highest quartile 78 1 99
Totals 235 49 83
The Examinations Committee believe that the foregoing statistics indicate progress in
the production of an improved written examination. Efforts to continue the progress are
constantly being made. Currently, the analytic study of an examination includes the
com-putation of an “index” of the actual value of each question asked. The index consists
of the percentage of candidates passing the oral examination who knew the correct answer minus the percentage of candidates failing the oral examination who also knew the correct answer. It is already apparent that certain questions of all grades of difficulty are not effective in distinguishing between candidates who pass the oral examination and
those who do not. Such questions will be eliminated in the future.
The American Board of Pediatrics as a whole has now decided that a candidate must
earn a passing grade on its written examination before being permitted to appear for oral examination.