J—s

In the first case (i) the use of a figure close to the average percentage or average ratio appeared justified. In the second case (ii) it was considered advisable to follow the trend observed in the base years, and for the enrolment percentage or the attendance ratio a figure was used which resembled that for 1949 rather than that for the average of the five base years. Finally, the percentages and ratios were multiplied by each other for each age and each type of school, to yield one factor which as a percentage might be applied at once to the appropriate birth number. The factors are tabulated in columns (7) to (10) of Table D. Examplels-16-year-olds at public post-primary schools in 1957. (a) Enrolment percentages of birth numbers — 1949 .. .. 56-26 1948 .. .. 54-12 1947 .. .. 53-47 1946 .. .. 54-59 1945 .. ..51-57 Average percentage 54-0 ; Percentage adopted = 56-25. (b) Attendance ratios for public post-primary schools —- 1949 .. .. 74-5 1948 .. .. 74-2 1947 .. .. 74-2 1946 •• .. 74-2 1945 .. .. 74-3 Ratio adopted = 74-5 (c) Product of (a) 56-25 and (6) 74-5 over 100 = 41-91. (d) The corresponding birth number for 15-16-year-olds in 1957 is that of 1941-42 = 39,628. By applying the factor (c) = 41-91 per cent, to this number the estimate of 16,605 is obtained. In this way the estimates for the years 1950-1960 were computed for each age and for public primary (Table CI), private primary (Table C 2), public post-primary (Table C 3), and private secondary schools (Table C 4). The summary estimates of the school population in all public and private primary and post-primary schools are given in Table C 5. SPECIAL POINTS TO BE NOTED The method described above, resting as it does on the statistical observation of enrolment percentages and attendance ratios in five base years, and on the combination of these percentages and ratios in one factor, has the advantage that it permits a simple and direct forecast of future enrolments by reference to the births in the corresponding years. It has the disadvantage that, among the variables of infant and child mortality, net gains (or losses) in external migration, ratios of attendance at different types of school, and proportions of school attendance below and above the compulsory school ages, it does not take into account any change that is not indicated in the base years. To this extent the estimates may require adjustments in the light of the behaviour of those variables in later years. Since the estimates are made for individual years of age it remains possible that changes in one age-group may be cancelled out by changes with the opposite effect in another age-group. For example, an influx into primary schools of five-year-olds that increases beyond 87 per cent, of the number of births five years earlier might be offset partly or wholly if less thirteen and fourteen-year-olds attend the primary schools, and more attend the post-primary schools. The following five points should be noted in particular:— (a) To carry the primary-school population estimates beyond 1955—that is, the latest year of known births (1949-50) —it was necessary to assume the numbers of births for the years 1950 to 1955. These figures were marked by an asterisk (thus " * 46,000 "), and it has been assumed that the decline of births after the peak in 1946-47 will continue at a slightly higher rate. The anticipated decline appears justified by the fact that the number of women in child-bearing ages is at present declining. The assumed numbers of births affect about one-tenth of the estimated 1956 primary school enrolment, and their influence increases each year until 1960, when they affect approximately 60 per cent, of the estimated total. Even then, however, the margin of error will not be great, whether it leads to an under-estimate or an over-estimate. (b) Since the Maori school population represents approximately one-tenth of the total school population the reduction in Maori infant mortality (114-92 in 1939, as against 76-67 in 1948 per thousand live births) will probably lead to an improvement in the enrolment taken as a percentage of birth numbers. It is therefore probable that enrolment percentages used for forecasting will gradually become too low by several points, and will result in a slight under-estimation.

21