Enrollment Forecast Methodology

 

            All forecasts have several things in common; all are an extrapolation of the past, all involve some level of judgment, and all forecasts are wrong.  This is why judgment is so important when creating and evaluating the forecast.  The real goal in the forecasting process is to reduce the range of error and monitor the process over time so it can continually be improved.

 

            The forecasting methodology used to predict the number of students who will be enrolling in Prince William County Schools for the next 6 years is a combination of methods and analyses.  These methods include a birth to kindergarten analysis, the cohort survival method for grades 1 through 12 for one year ahead, the generation factor analysis for all projections more than one year ahead, and judgmental adjustment to fine tune the forecasts.  This combination of methods was chosen because they provide for very accurate forecasts and at the same time are relatively inexpensive to produce.  This forecasting process is a two step process using the cohort method for the first year out and the generation Factor method for years 2 through 6.

 

            The accuracy of this forecasting method has been very good in the past, with an average error of less than .4% countywide.  Success at the school level has not been as great but the forecasts have been well within the acceptable range.  Countywide forecasts have been more successful than the school level forecasts because the population being forecasted is much larger. This phenomenon is common to all forecasting methodologies and is therefore expected.  One of the drawbacks of the methodology is its lack of explanatory information.  Unlike an explanatory or causal method, the cohort method shows no correlation to other variables and therefore reveals no possible reasons for the shifts in student population growth.

 

 

Phase I – First-Year Out Forecasts

 

            The first-year out forecasts are created using a birth to kindergarten analysis, and the cohort method, on a school by grade basis, for 1st through 12th grades.  These forecasts are then used as a baseline for calculating the forecasts for the out years. 

 

            Birth To Kindergarten  The method used to forecast incoming kindergartners is a modified cohort procedure.  Instead of using the grade to grade progression used in the standard cohort progression method, births countywide are used to forecast kindergarten enrollment 5 years later.  This procedure projects the total number of kindergartners who will attend Prince William County Schools in the first year forecasted.  These students are then allocated to the elementary schools on a weighted average basis.

 

            Cohort Progression  The cohort method is, in basic terms, the application of an average growth rate over time to the current year’s enrollment.  This growth rate can be taken from the previous year, or from an average of several previous years.  The resulting ratio or growth rate is then applied to the current year's enrollment to project the next year’s enrollment. 

 

            In Prince William County, a cohort forecast is done for the whole county on a grade by grade level.  These forecasts have been historically very accurate.  Additionally, a cohort forecast is done at the school level on a grade by grade basis.  These forecasts have been less accurate but are still within the acceptable range.

 

            After each of these forecasts is calculated there is a reconciliation of the two using the judgment of the forecasters.  The result is a forecast that is used as a baseline for the out-year forecasts.

 

            The cohort ratio (generation factor) calculation in the following example describes the progression of sixth graders to seventh grade on average over 3 years.  The same could be done over 2, 5, or any desired number of years.  A greater number of years in the cohort calculation lessens the effects of any given year on the forecast, effectively “smoothing out” the historical data.  Keeping this in mind, the forecaster can adjust the number of years used to calculate the cohort ratio to adjust for current and expected economic conditions and produce better forecast results.

 

 

 

Grade

 

1995

 

1996

 

1997

 

1998

 

Cohort Ratio

 

Forecasted

1999

 

6th

 

3,782

 

3,790

 

3,892

 

3,888

 

 

 

 

 

 

7th

 

N/A

 

3,857

 

3,830

 

3,957

 

1.0139

 

3,949

 

              

 

1.      Cohort Ratio = (7th 1996 + 7th 1997 + 7th 1998) / (6th 1995 + 6th 1996 +6th 1997)

               7th 1998 indicates the 7th grade cohort of 1998.

2.      Cohort Ratio = (3,857+3,830+3,957) / (3,782+3,790+3,892) = 1.0157

3.      Forecasted 1999 7th Grade Cohort = 3,888 x 1.0157 = 3,949

 

            The number of years used to calculate the cohort ratio has in fact changed several times over the period starting with 1980 in Prince William County.  These changes were made to help adjust for the increasing growth rates in the late 80's.  A three-year cohort was used to calculate the forecast for 1999-2000 because it is believed that the growth pattern over those years is a good representation of the future.  In the future the use of mixed cohorts may also be employed to adjust for differing growth rates within the county.

 

 

 

Phase II – Out-Year Forecasts  (See Attached Flow Chart)

 

 

 

 

 

Single-Family

 

Townhouse

 

Multi-Family

 

Elementary

 

.365

 

.331

 

.161

 

 

Middle

 

.230

 

.132

 

.049

 

 

High

 

.289

 

.140

 

.044

 

 

 

            Generation Factor  The out-year forecasts are created using the generation factor method combined with the cohort progression method. Together, these two means produce additional students using the first projected year (from Phase I) as a baseline.  The process is detailed in the attached flow chart.  This is an effective method that can, with accurate data, provide highly accurate forecasts of future enrollments. 

 

            Generation factors are created by calculating the number of students by housing unit type to the number of housing unit types countywide.  This ratio is then applied to the number of predicted available units by year to produce the out-year projections of new students in the Prince William County Schools.

 

            The table above shows the student generation factors used by the Prince William County Public Schools.  These ratios are countywide averages, calculated by school level and housing unit type.

 

            The projections presented in this document were created using a combination of each of the listed methods. Birth to Kindergarten analysis was used to forecast Kindergarten enrollments. The Cohort progression method was used to project the 1999-2000 enrollment, with the out-years being projected using a combination of the cohort progression and generation factor methods. Where appropriate, judgmental adjustments were made.