STUDI PERENCANAAN PENDIDIKAN

9 forecasting
Civilized men have long been occupied with the problems of projection. P.E. Rosove has identified the following twenty-two for forecasting:
1. Brainstorming (the authors)
2. Delphi technique
3. Expert opinion (the many contributors to this book)
4. Literary faction
5. Scenarios (chapter 1)
6. Historical analogy (chapter 2)
7. Historical sequences (chapter 2)
8. Content analysis (the whole book)
9. Social accounting (chapter 9)
10. Primary determinant
11. Time series (chapter 9)
12. Extrapolation (chapter9)
13. Contextual mapping
14. Morphological analysis (the comprehensive educational planning model of this book)
15. Relevance trees
16. Decision matrices (chapter 9)
17. Deterministic models
18. Probabilistic models
19. Gaming
20. Operational simulation
21. Cos-benefit analysis
22. Input/output tables (chapter 9)
At once we see the problem of forecasting as being intimately bound up with scientific methods and theory building. Many methods and theories are available but answers to cartain basic questions are neccssary to a forecasting technique.
How far should the forecast be carried? Are different periods recommended for different activities, e.g., projecting population, capital improvements, budgeting, and so on? How are the steps for wich the forecasts are made determined? Will the way the plan will be implemented affect the forecast?
To answerthese question, the following sections include the various considerations required for forecasting, some methods for forecasting for various aspects (people, plans movent, economics, and activities) of educational systems, and a projection of the educational system as a whole.
CONSIDWRATIONS REQUIRED FOR FORECASTING
Forecasting involves assumptions about likely condition. To determine the assumpation, two categories basic and special assumption can be established. Basic assumptions are about such factors as the number of births to occur, deaths, rate of population migration, the form of government, and political, economic, and social organizations.
Special assumptions center on essentially local conditions. Will there be a major recession or not? Will full employment be a reality? Will business continue in an up-down cycle? Will public policy permint exploitation of the local environment? What changes planned for will become a reality? Based on these variables and others, the problem then becomes one of selecting those assumption to be used for educational planning purposes.
Other considerations are related to observed data and their cause-and-effect relationships. There levels exist: first, a deterministic causality in which if one event occurs another occurs also; second in the probabilistic causality in which if an event occurs, a probability (p) exists another event will also happen; third, there is acirrelation in which an event occurs in association with another event but no observable cause-and-effect relationship exists between them.
Consideration must also be given to how far forecasts should be carred. Sometimes a twenty-five-to thirty-year plan (long-range planning, is demended; or a ten-to fifteen-year plan (middle-range planning) is required; or a more detailed plan (short-range planning) is needed. These three types of plans depend on wwhat is needed. In the past, short-range planning has been the dominant plan method with almost complete exclusion of the dominant plan method with almost complete exclusion of the other two types. Therefore no paths were available for the educational system to follow. As a result, educational programming was nearsighted in providing for the immediate needs of individuals. Comprehensive educational planning involves a selected path to be followed. The path will be precise for the early stages and increasingly fuzzy as the time period extends into the future.
The way the plan is implemented constitutes another consideration. As mentioned above, a plan is a charted path or a trajectory of changes that requires occur, thy must be noted in time for corrective actions to be applied. Also, if there are unforeseen conditions detrimental to the individual’s environment, controls should be such that remedial action must be measurable at suitable time intervals. The best milestones occur in census-taking years but conditions in the future may force shorter counting intervals.
Projection of varios aspect (people, employment needs, educational needs and so on) must be made simultaneously. The time intervals of these projections must be the same to allow comparisons in the actual state of affairs. A common practice is to set n,n+5, n 10,n+15. n +20, and so on, to fit the decennial count.
SOME FORECASTING METHODS
The most common forecast in use by educational planners is the population projection. Its use indicates the overriding priority already established for determining future educational resources. It forms a framework for a great deal of subsequent work in devising, testing evaluating, and implementing an educational plan. Most of the important decisions about educational services derived from population forecasts
One method for forecasting populations used by many educational planning agencies is the cohort-survival method. (see figure 9.1)
Age groups are extracted from the latest census. The first step requires the educational planner to take the appropriate age group and take from it the number of deaths appropriate for the group available from an abridged-life table. The next step estimates the number of net in-and-out migrations by age groups for the same interval of time ; the results are entered as the second adjustment. The sum of survivors and the in-and-out migrants from the original age group furnishes the estimate for the next higher age group at the beginning of the next five-year period.
The emptied slot of the previous age group for the same year is filled also from the previous age group of the previous year. The original slot, now empty, is filled by the application of the approximate fertility ratio to the number of famales of child-bearing age (usually fifteen to forty-four) at the beginning of the last period. Computer programs are available for quick processing.
The migration natural-increase method is similar in many respects to the one just described. Two components rater than three, as in the cohort-survival method, are used. As a result only total populations are capable of being forecast. Current estimates of population are taken and adjusted migration and natural increases on ayear-by-year until the forecast date is reached. Past net-migration rates are determined by past relationships between the direction and volume of net migration for both slow and rapid economic expansions. These figures are than projected to fit the economic outlook. Other approaches for determining net migration rates are used, such as employment forecasts or educational forecasts. Other resulting ratios based on total population are then extended by mathematical or graphical extrapolation. Natural increases are generally based on subjective extension of past minimum and maximum rates. These may be computed from recorded birth and death statistics. Caution should be used in projecting these figures, as past records are based on depression and wartime fluctuation in births. Table 9.1 illustrates the forecast procedure.
The mathematical and graphic method referred to are trend extrapolations by the least-squares method and by logistic curves. The assumption is made that some forces that operated in the past will also operate in the future. The arithmetic approach results in a straight-line projection and assumes the same numerical change for similar periods of time. The geometric projection is also a straight line but on semilogarthmic graph paper, signifiying similar rates of change over similar periods of time. The last-squares method for fitting a straight line is commonly used for representing total population data derived from points representing births, deaths, and net migration over time. This approach is most useful for short-term forecasting. (see figure 9.2) the ratio and apportionment methods are applicable to successively wider geographical ares, c.g., from county to subregion to state to regional to national levels. The requirements include a time series of population for the county level and a forecast for the national level. In ratio method using county and national levels the population data for the national level are noted. Next plotted are the population data over the same time period for the county level compared with the national level and the ratio derived are squares. A curve is fitted to the data and by least squares or graphic methods extrapolated to intersect the projected value for the national level at the given forecast date. The apportionment method is broadly similar. Instead of one level of difference, a series of ratios representing cach level are developed and plotted, also in a time series. Curves are fitted and these are then projected to the forecast date.
The matrix method, one of the latest and potentiality the most fruitful method, uses matrix algebra. The initial age and sex distribution is represented as a column vector. The differences between births and deaths are handled by a “survivorship matrix” which operates on the original vector to project the population through successive time periods.
Table 9.2 (matrix method) illustrates the point ;in this table b2, b3,b4,b5…bn are the age specific birth rates and s1, s2, s3, s4…..sn are probabilities of surviving from n to n+5, if a five-year interval is being used. Using a set of matrices makes possible a variety of forecasting procedures.
Economic forecasting methods fall into two groups the analitycal and the short-cut methods. One analytical method is input/output analysis. After the use of estimates for the various economic goods and services listed in the known input/output tabulation, and adjustment for dollar value and anticipated changes in the economy, the refined coefficients are than applied industry by industry to obtion the output for the forecast date. These data give total economic outputs. Estimates of forecast data on labor productivity are obtained by a projection of the dollar output per worker for all listed industries in the initial year of the study. The actual employment estimates are obtained by the diving of values for estimated future output by the appropriate values for output per worker. Obviously, this method depends on the availability of data and on technical know-how, but, if these problems are solved, educational planners can readily plan educational programs that are more appropriate for increasing efficiency in the economic development of education.
Other economic forecast tools are methods that use income statistics. The basic requirements of these methods are trend analysis of past production and past worker outputs and suitable national forecasts of both production and output per warkers. These data are untized by the ratio and apportionment and matrix methods but with minor modifications to fit the problem.
Some short-cut forecasting methods for economis are the apportionment technique and direct and indirect ratio procedures. The apportionment technique is superior the ratio procedures as it accounts for employment trends in related areas that share in overall trends. Another method for economic forecasting is the social accounts methods. Even though its development is in the initial stages, its potential value to the educational planers understanding of the economic environment warrants mention. The method is similar to the input/output analysis but it uses monay rather than production and employment as the measure of the interrelations within the economic system. it includes capital formatioan, ivestment, trade, and production. The matrix for income and expenditure relationships among all sectors is used to obtion projections of income each sector. Estimates of productivity by projected income per worker at the levels of employment are forecast.
Population and economi projrction have a wide range of application. The number of potentially new or rehabilited sepaces that will be required is indicated. This, in turn, influences activities, such as education. This process can be refined, for istance, one can compare the demand for social activities with the supply of students derived from population projections. The degree of sophistication for projecting place and activity needs will be controlled by the from of the population forecast. If a simple total population is used then a crude activity rate will result; if males and females are projected then separate male and female activity rates woll result ; if cohort-survival is used then activities can be projected in terms of age and sex groupings. If one knows the projected activities, the projected population, and the projected economic environment, than the number of projected places or spaces can be determined by a technique such as the matrix-method.
Projecting spaces must be done so that the activity and the space are related to the same areal unit. Second, the time intervals between the steps in projection should be the same. Third, it is the place or cost values, adaptations, physical appearances, sensory qualities, and so on. The object of projecting is to have the best possible data throughout the planning stages. This need is poiganantly demonstrated by the approximately 50 per cent of the schools now in operation that are incapable of housing new educational programs.
In summary, all of the relevant aspects of a place require projecting and at some trial interval such as a five-year interval. The objective is to show potential places that are capable of accommodating educational activities at future times. A combanition of various factors-people, economic, cultural, and others are involved.
Last, the projection of movement is primarily concerned with the patterns and volumes of future interaction. Patterns of movement are generally assoctied with a particular activity. This being so, the projrcted activities described by their characteristics-for example, number of students at a certain school, their parents’ per capita income, car ownership, and so on-make projecting patterns of movement possible. In this case the number of projected movements becomes the dependent variable X, and the characteristics by activity of the independent variables Y1, Y2, Y3 and so on. Using multiple regression analysis with the following from
X= A+B1Y1+B2Y2+B3Y3………Bm Ym
With A’s and B’s coefficesients derived in the cours of statistical analysis, the number of movements of each kind of projected activity by frequency by pupose, and by mode of movement is possible. Clearly movement forcasting is not a simple matter. Careful analysis, therefore, requires that projrctions of people, places, economics, and activities must be done in order to derive relationships among them
PROJECTING THE EDUCATIONAL SYSTEM AS A WHOLE
Projecting certaing parts of an educational system without considering the other parts impinging on the whole educational system is unrealistic. Our concern in comprehensive educational planning involves giding an controlling the educational pattering of people, places, movements, economic, and so on. It follows than that in projecting educational systems as a whole one must include all the parts. Given such- and such an educational policy with such-and-such assumptions about people’s behavior (a,b,c and so on) what changes in places (p,g,r and so on) and movements (x,y,z and so on) can be expected at time t2 t3 and t4? A model, as lowry wrote “…..consist of names, variables embedded in mathematical formulae…., numerical constants…. And a computational method….the pattern generated is typically a set of values of interest to the planner or decision- maker, each value tagged by geographical location and/or calendar date of occurance
Projecting a whole system must include an understanding of educational patterns and changes in the past assumption is that future relationships and coefficients can be derived. In our case the system as a whole is best modeled in a way that simulates changes in the five central featurs (people, place, movement, economics and activity) as they occur through time and that simulates the reaction of each to changes in the other.
A recursive model is the most promising for handling these conditions. It simulates the evaluation of the educational system in a series of steps by taking the output of one time-stage as the input of the next. This then becomes a trajectory showing the way the way system is expected to evolve through time.
Data describing the educational system are in many cases capable of being reduced to matrix and map forms. The matrices show the number of people in various places, or activities in various places, or a cost consist of a set of matrices and maps (people, places, movements, costs, activities) for each of the years 1972, 1977, 1987, and so on.