5.  The selection and reliability of these sources of data

By employing data secured only from official school records and in the manner stated, this study has been limited to those schools that provide the cumulative pupil records, with continuity and completeness, for a sufficient period of years.  Some schools had to be eliminated from consideration for our purposes because the cumulative records covered too brief a period of years.  In other schools administrative changes had broken the continuity of the records, making them difficult to interpret or undependable for this study.  The shortage of clerical help was the reason given in one school for completing only the records of the graduates.  In addition to the requirements pertaining to records, only publicly administered and co-educational schools have been included among those whose records are used.  It was also considered important to have schools representing the large as well as the small city on the list of those studied.  Since many schools do not possess these important records, or do not recognize their value, it is quite probable that the conditions prescribed here tended to a selection of schools superior in reference to systematic procedure, definite standards, and stable organization, as compared to those in general which lack adequate records.

The reliability and correctness of these records for the schools named are vouched for and verbally certified by the principals as the most dependable and in large part the only information of its kind in the possession of the schools.  In each of these schools the principals have capable assistants who are charged with the keeping of the records, although they are aided at times by teachers or pupils who work under direction.  In three of the larger schools a special secretary has full charge of the records, and is even expected to make suggestions for revisions and improvements of the forms and methods.  In view of such facts it seems doubtful that one could anywhere find more dependable school records of this sort.  It was true of one of the schools that the records previous to 1909 proved to be unreliable.  There is no inclination here to deny the existence of defects and limitations to these records, but the intimate acquaintance resulting from close inquiry, involving nearly every factor which the records contain, is convincing that for these schools at least the records are highly dependable.

However, there is some tendency for even the best school records to understate the full situation regarding failure, while there is no corresponding tendency to overstate or to record failures not made.  Not infrequently the pupils who drop out after previously failing may receive no mark or an incomplete one for the last semester in school.  Although a portion or all of such work may obviously merit failure, yet it is not usually so recorded.  In a similar manner