Good System, Bad Scenario

Good System, Bad Scenario

Preliminary report says school system's projection methods sound but won't work for long.

The good news is the methodology used by Fairfax County Public Schools for enrollment projections is working as best it can. The bad news is it can't do any better and will, eventually, become unable to accurately project enrollment for the school system.

Jerome McKibben, senior demographer from McKibben Demographic Research, gave a preliminary presentation to the Fairfax County School Board on Monday, Sept. 11 stating that unless changes are made to the enrollment procedure, the unforeseen variables that caused the overcrowding at South County Secondary School will become more common.

"The methodology you use is good," McKibben told the board, along with Gary Chevalier, chair of the Office of Facilities Planning, and Assistant Superintendent of Facilities and Transportation Dean Tistadt. "The system you use is a variation of what is used by just about every school system in the country. It tends to be incredibly inaccurate, but it's the most cost effective."

McKibben said the problem with the procedure is that it limits the amount of variables from one community, or one school, to another, making it difficult for him to make general statements that would apply to every school.

"Each area has its own demographics, characteristics and population waves," McKibben said.

In addition, projections for elementary schools tend to run a little higher than actual enrollment, while the opposite is true for middle and high schools.

FROM ONE YEAR to another, any given school may see a projection error of 0 percent to 5 percent, which should be considered normal. As an example, McKibben pointed to Aldrin Elementary School in Reston which had a mean absolute percentage error of 4.85 percent.

While most schools fell into that range, the obvious differences were at South County, where the enrollment projections had error percentages of 9.09 percent at the middle school level and the high school saw a 14.38 percent error. In other words, the enrollment projections differed from the actual student population by just over 9 percent for middle school students and 14 percent at the high school level.

The situation at South County is different from other schools in the county because it only had one year of enrollment numbers to take into consideration. With other schools, McKibben was able to look at a five-year length of time to compute the error percentages. With more numbers to compare, the differences between enrollment projections and the actual population of a school tend to even out and look smaller.

Several factors contributed to the problem, McKibben said, including the number of students who originally went to Hayfield Secondary School but had gone to private school and returned when South County opened last September.

"There was a late burst of high income households with older kids which is a natural bubble of population that comes through the system," he said. This type of enrollment growth is to be expected when a new school, especially a middle or high school, opens, when between 85 and 95 percent of eligible children in the area will go to the new school.

"This is a classic example of a systemic error," McKibben said. "You need to have the data available to examine certain variables. In most cases, this system will work but it will not work in the future."

MCKIBBEN RECOMMENDED the projection model used by FCPS be expanded to take census data, existing home sales and new home construction into consideration when making long-term projections, as it will be better equipped to obtain a more accurate enrollment rate.

"You have to have a deft hand and know what you're doing or you can really mess things up badly," McKibben said.

With the reassurance that their numbers were as accurate as possible, the School Board asked McKibben where to go from here.

"As a board member, we have to make certain decisions and sometimes that means making long-term projections," said Board member Brad Center (Lee). "Is there anything we can do better in the out-years for better numbers? How can we identify how accurately we should take those numbers," he asked.

McKibben said it was easy to identify those areas, based on growth and census patterns and data and that his final report would include more detailed recommendations for changes to be made.

"All our enrollment numbers affect our buildings and boundary systems," said Board member Jane Strauss (Dranesville). "Thank you for presenting the census data ... each attendance area has different modeling bases that need different attention."

It also affects the Capital Improvement Plan, noted Board member Dan Storck (Mount Vernon). "I understand year-to-year changes, but why you're here is where we'll be in five to 10 years."

McKibben told board members that they're currently at the end of their enrollment peak, and within the next few years enrollment will most likely start to decline.

"It'll go back up again in 15 years or so, but you'll never quite get to this high again," McKibben said.

He recommended that the school system add three or four people to the planning staff to better track population trends and home sales to gauge the amount of school-age children.

"Honestly, I think you should redistrict the entire system," McKibben said.

Storck later said he was happy with McKibben's initial report and the identification of potential problems within the projection system. However, his biggest concern is taking McKibben's recommendations and making a real difference in the accuracy of long-term projections.

"Even if the (error) numbers settle down, they're still way too high," said Storck of the error percentages at South County.

Now facing what promises to be an emotional and hard-fought boundary study at the school, Storck said the School Board is hoping to be able to somehow use McKibben's suggestions to avoid overcrowding in the future.

"He clearly set the marker for where we want to go as a 21st century school system," Storck said of McKibben's work.

Both Storck and Olsezek agreed that McKibben's analysis was well worth the $49,500 the county paid.

"If we can avoid doing one boundary study because of his work, we've easily saved five times his fee," Storck said. "We could spend another million and be ahead of the game. This is an information economy, you pay a little for the information and get a huge return."