A Pitch for School Geo-Demography Richard Lycan Population Research Center Portland State University, Oregon Association of Pacific Coast Geographers Olympia, Washington, October, 2012 Purpose of paper Provide a snapshot of the field of school demography as practiced by those of a geographical persuasion. Briefly outline the major approaches used in school enrollment forecasting. Show examples of how contract research in this area leads to interesting research issues. Describe the enterprise and how it can fit into a university setting. Who we are The Population Research Center (PRC) is a community service organization in the College of Urban and Public Affairs at Portland State University. PRC has some legislatively mandated functions, city and county estimates, and liaison with the U.S. Bureau of the Census. PRC has a staff of 6 full time professionals and support staff and employs graduate student assistants, mainly from Urban Studies and Planning and Geography. About half our income is from contract research. Most is for public agencies and non-profits. School demography is our largest source of contract income and employs about 1.5 FTE staff. Charles Rynerson does our school demography work. We have provided demographic services for school districts for nearly 40 years. For the last 20 years the school support has made use of GIS providing geographically detailed forecasts. The types of services to school districts The most commonly requested projects are enrollment forecasts for school districts, or districts and their attendance areas. Other services for school districts include: Attendance area boundary changes and boundary change scenarios School board member district reapportionment Estimates of student generation from new housing Field surveys to determine household composition The studies typically include a written report and a presentation to a school board committee and/or the general public. We have provided service to most of the mid to large school districts in Oregon and have a continuing agreement with a few. Where do school districts get demographic services? In house staff – Many large school districts have a staff, usually affiliated with facilities planning, that provide demographic services. The staff have skills in demography, GIS, planning, and database management. In a small district sometimes an assistant superintendent who deals with facilities does the forecasting. Turnkey Software – There are several firms that sell software tools to provide enrollment forecasts. Often they are an appendage to school bus routing software. Examples are Edulog in Montana and Davis Demographics in California. Private Consultants – There are several consulting firms that provide demographic services to school districts nationally. Examples are Gobalet and Lapkoff in California and McKibben Demographics in South Carolina, as well as some who work at a more local level. A&E firms sometimes provide forecasts as part of a capital plan. Universities – Several university population research centers provide this type of service. Examples include our center in Oregon and the Applied Demography Lab at the University of Wisconsin. The data environment Student record is central to building forecasts. We need historical data in order to calibrate enrollment forecasts. The student record data is carefully geo-coded to the students’ residential locations. Birth records also are geo-coded to mother’s residence and identify potential enrollees some five years later. Tax-lot data from the assessor’s office provides the information about the housing units and neighborhoods where students live. We link birth records and student records to their tax-lot. We make use of data from the decennial census and the American Community Survey. Our forecasts are informed by advice from various other organizations such as: Local planning departments The regional economists with the Oregon Employment Department Enrollment forecasts Long versus mid-range enrollment forecasts Long range forecasts - typically required for facilities planning and usually are for about ten years, but may extend as far as 15 or 20 years into the future. Mid range forecasts – typically required for budget planning, staffing, transportation planning, and use of existing facilities usually include extent five years into the future. Three types of forecasting models typically used The grade-progression model – driven by student enrollment trends for short term forecasts and for smaller geographies, such as elementary school attendance areas. The cohort cohort-component model – driven by births, deaths, and migration and used for larger geographies, such as school districts, and longer term forecasts. The housing based model – driven by housing types and numbers and neighborhood turnover. We use combinations of the three types. Three types of enrollment forecasting models utilized component Grade progression Housing based Students per Home Cohort- GPR 2008/2007Chart 1 Enrollment TTSD Students per Single Home,06/05 Fall 2010 Births05/04 HS | MS | ES 01/00 02/01 03/02 Family 04/03 07/06 08/07 09/08 10/09 11/10 12/11 0.30 KG 1.10 1.00 1.01 0.89 1146|1144|1136 0.98 0.99 1.02 0.94 0.97 1.07 0.99 1.03 2000-2009 1146|1144|1137 1.11 1.02 Built 0.96 1.00 1.02 1.01 1.01 1.04 1.02 1.04 0.95 0.98 Population 01 0.25 1990-1999 1146|1144|1138 1.06 1.11 Built 0.94 1.57 1.00 1.00 1.30 1.00 1.33 0.63 0.89 0.75 1146|1144|1140 1.01 0.92 Built 0.93 1.04 00-04 before 1990 0.93 1.02 0.90 0.80 …. 0.92 1.01 0.87 0.97 1146|1145|1138 1.16 1.08 1.07 1.04 1.00 0.93 0.91 0.93 1.01 0.85 0.98 0.91 0.20 05-091.10 1.14 1.21 1.04 11 1.17 1.05 0.93 0.97 1146|1145|1139 1.12 1.14 1.09 1.10 1146|1145|4364 0.86 1.00 0.82 0.95 0.95 1.33 0.90 1.11 10-141.07 1.37 1.04 1.10 12 1.19 1.00 0.93 0.76 0.15 1301|1145|1138 1.75 1.35 0.69 1.21 0.80 0.90 1.08 0.58 1301|1145|1369 1.14 0.94 0.89 0.92 15-171.08 0.96 1.07 1.24 0.88 0.88 0.95 0.95 1301|1145|4364 1.02 1.10 1.18 1.15 1.11 1.04 0.98 0.98 1.07 1.00 0.98 1.00 0.10 ………. Migration 1301|1300|1135 1.04 1.12 0.98 1.00 0.96 0.94 0.96 1.00 0.94 0.91 1.08 0.94 1301|1300|1142 1.18 1.07 0.98 1.09 1.09 1.01 0.98 0.96 1.05 0.95 1.02 0.98 15-44 0.05 1301|1300|1143 1.02 0.96 1.01 0.91 1.01 1.01 1.02 1.00 0.93 1.04 0.92 1.02 1301|1300|1369 1.12 1.00 0.91 0.97 0.96 1.03 0.98 1.21 1.02 1.12 0.94 1.02 …….. 0.00 District 1.07 1.03 1.00 1.02 1.02 1.03 1.01 1.00 1.03 0.99 0.97 0.97 K-5 Combined models used for most of our forecasts 6-8 75-79 80-84 Grade Level 85+ Deaths 9-12 The Cohort Concept Many demographic models based on tracking a cohort over time. Lexis Diagram Can be viewed at the Cohort Over Time x+5 level of the individual Death of person x+4 x+3 Or for groups of common origin Age x+2 x+1 Age x x+5 Child born x+4 Child born x+3 Birth of person x+2 x+1 x t t+1 t+2 t+3 t+4 t+5 Time t t+1 t+2 Time t+3 t+4 t+5 The 1994 Cohort View Enrollment History by Grade and Year KG 01 02 03 04 05 06 07 08 09 10 11 12 Other 1990 980 1,220 1,183 1,171 1,154 1,074 1,151 1,159 1,120 1,098 1,029 993 875 30 1991 1,011 1,223 1,235 1,203 1,242 1,207 1,148 1,212 1,173 1,132 1,066 987 929 31 1992 1,041 1,230 1,224 1,293 1,243 1,287 1,307 1,210 1,259 1,232 1,155 1,021 949 18 1993 1,144 1,250 1,253 1,297 1,352 1,290 1,336 1,327 1,239 1,295 1,233 1,099 953 13 1994 1,175 1,304 1,318 1,304 1,383 1,406 1,420 1,436 1,358 1,303 1,283 1,111 1,028 8 1995 1,186 1,325 1,343 1,357 1,360 1,401 1,409 1,473 1,453 1,436 1,312 1,180 1,039 8 1996 1,291 1,342 1,385 1,386 1,391 1,396 1,474 1,466 1,533 1,533 1,428 1,214 1,149 19 1997 1,195 1,449 1,391 1,428 1,442 1,457 1,431 1,525 1,504 1,587 1,505 1,311 1,138 15 1998 1,248 1,359 1,527 1,462 1,496 1,490 1,502 1,471 1,550 1,601 1,598 1,419 1,201 29 1999 1,245 1,393 1,392 1,561 1,488 1,562 1,552 1,587 1,550 1,735 1,600 1,531 1,350 35 2000 1,296 1,359 1,442 1,433 1,656 1,552 1,614 1,588 1,612 1,632 1,709 1,502 1,435 37 2001 1,378 1,434 1,424 1,515 1,541 1,722 1,687 1,647 1,679 1,735 1,644 1,668 1,399 35 2002 1,381 1,523 1,478 1,492 1,536 1,600 1,779 1,707 1,687 1,834 1,708 1,613 1,517 57 2003 1,504 1,530 1,567 1,566 1,561 1,592 1,667 1,800 1,759 1,815 1,815 1,657 1,470 65 2004 1,516 1,596 1,619 1,620 1,652 1,651 1,676 1,730 1,852 1,897 1,828 1,782 1,621 65 1994 KG Cohort 2,000 Enrolled K-2 3,383 3,469 3,495 3,647 3,797 3,854 4,018 4,0351,8004,134 4,030 4,097 4,236 4,382 4,601 4,731 3-5 3,399 3,652 3,823 3,939 4,093 4,118 4,173 4,3271,6004,448 4,611 4,641 4,778 4,628 4,719 4,923 7-8 3,430 3,533 3,776 3,902 4,214 4,335 4,473 4,4601,4004,523 4,689 4,814 5,013 5,173 5,226 5,258 1,200 9-12 3,995 4,114 4,357 4,580 4,725 4,967 5,324 5,541 5,819 6,216 6,278 6,446 6,672 6,757 7,128 1,000 Other 30 31 18 13 8 8 19 15 800 29 35 37 35 57 39 65 Total 14,237 14,799 15,469 16,081 16,837 17,282 18,007 18,378 600 18,953 19,581 19,867 20,508 20,912 21,342 22,104 400 200 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 KG 01 02 03 04 05 06 07 08 09 10 Year, Grade Level Grade Progression ratio for Tigard School District Grade Progression Ratio for District Year2/Year1 01/00 02/01 03/02 04/03 05/04 06/05 07/06 08/07 09/08 10/09 11/10 12/11 2008/2007 1.07 1.03 1.00 1.02 1.02 1.03 1.01 1.00 1.03 0.99 0.97 0.97 2009/2008 1.05 1.01 1.00 0.96 1.00 0.98 0.99 1.01 1.01 1.00 0.96 0.98 2010/2009 1.07 1.00 1.01 1.02 1.04 1.01 1.01 0.99 1.04 1.01 1.02 1.04 1.08 1.06 1.04 1.02 1.00 0.98 0.96 2008/2007 0.94 2009/2008 0.92 2010/2009 0.90 01/00 02/01 03/02 04/03 05/04 06/05 07/06 08/07 09/08 10/09 11/10 12/11 The grade progression ratio all can be computed spatially The dots show the location of students enrolled in Tigard Schools in Fall 2010. The color shades show the ratio of the Fall 2010 enrollment to the same cohort one year earlier. Green shaded areas show that the cohort declined Blue shaded areas show growth in the cohort. The cohort component model Enrollment is determined as a share of the forecast school age population. The school age population is determined by earlier births and net migration of families with school age children. The model is recursive, the forecast for each time period building on that 5 or ten years previous. Provides age detail, but best used for larger geographies. Births Population 00-04 05-09 Enrollment 10-14 KG 15-17 01 ………. …. 15-44 11 …….. 12 75-79 80-84 85+ Deaths Migration Data that inform the cohort-component model Fertility rates – by age Net migration rates by age and sex, perhaps the most difficult to estimate Public school capture rates – the share of the school age population enrolled. 1.0 100 90 0.8 80 In-migration of young households 70 0.6 Percent race and Hispanic status. Shift from births from younger to older mothers. Net Migration Rate 1990-2000 2000Portland Capture Rates - ACS & Student Records Public Schools Net Migrants 0.4 as share of population 0.2 60 50 2000 - Student Records 2000 - NCES/Census ` 40 30 20 0.0 10 00- 05- 10- 15- 20- 250 04 09 05 14 0619 07 24 29 08 30- 35- 40- 45- 50- 5534 59 09 3910 44 1149 1254 13 -0.2 6064 14 65- 70- 75- 80- 85+ 69 15 7416 79 1784 18 19 83.8 86.0 84.7 82.6 80.5 69.6 38.3 2000 - Student Records 35.9 80.2 83.7 83.2 82.7 87.3 84.8 86.2 Out-migration of older households Out-migration of households with 81.9 80.5 82.7 83.2 84.0 85.5 85.8 85.2 84.9 85.1 84.5 82.6 78.3 64.5 46.1 2000 - NCES/Census and empty nesters school aged children -0.4 Age 5 Year Age Group Capture rate for Portland Public Schools Computed from grid mapping of student record and census block data. In 2000. Lower values tend to be in high income areas where more students are enrolled in private schools. In 2010. High values tend to be in low income and minority areas where most students are enrolled in the public school system. The change from 2000 to 2010. The housing based model We look first at birth mothers who’s children will appear in school in about 5 years. Chart 1 TTSD Students per Single Family Home, Fall 2010 0.30 From work for Portland Public Schools we show what type of housing is most common for highly educated age 30+ mothers. OMA is a gentrifying neighborhood and Remainder is a lower income area with large Hispanic population. And for less educated mothers under age 30. From work for Tigard School district we show the loading of students into single family housing by age of the structure. Note how the loading of KG-5 students declines as the housing ages. Built 2000-2009 Built 1990-1999 0.25 Built before 1990 Students per Home 0.20 0.15 0.10 0.05 0.00 K-5 6-8 Grade Level 9-12 Enrollment forecasting leads raises interesting issues worth of further exploration Example One – The Hillsboro School District in suburban Portland has a large and increasing Hispanic population as well as the families of employees of Intel. It has experienced rapid population, and school enrollment, growth but recently the enrollment growth peaked. Example Two – In some areas of older but substantial housing in Portland enrollment has been declining and schools have been closed. However in 2006 some of the schools in these neighborhoods began to show enrollment growth in kindergarten and the lower grades. A gentrification process was at work involving older, highly educated, and affluent mothers. Example One: An enrollment forecast for Hillsboro School District focused on student generation rates Hillsboro SD has a large Hispanic population many of whom reside in multifamily housing. Student generation rates are relatively high in Hispanic neighborhoods but also in new suburban areas. Most areas of newly built housing show high student generation ratios. Areas where seniors outmigrate result in turnover and in-migration of younger households. As new housing ages the students advance in age and eventually leave home. Example Two: Draws from a forecast for Portland Public Schools School enrollment for the Portland District reflects the economic and housing epochs of the past 150 years, recently declining after the echo of the post WWII baby boom passed through. However, in just since about 2005 the District’s enrollment has shown a modest uptick. The setting for the Portland School District The Portland District is largely built out and surrounded by growing suburban areas. There has been some infill and replacement single family housing in the District, but most of the construction has been multifamily and does not generate much student enrollment. Only about 50 students come from the 20,000 units in the trendy Pearl District (A) but high value new housing (such as at B) has contributed some growth. The recent up-tick in enrollment is due more to turn-over than to new construction. B A An important factor in the turnaround has been the larger share of births to older mothers In 2001 the number of births in the District to older (age 30+) mothers equaled those to younger mothers (under age 30) Total births stayed about the same Proportion of Births to Older Mothers There was growth in the number of births to older mothers across the District, but that on the east side was mainly after 2000 Of particular interest was the “East Older Mothers area (East OMA). East OMA 1990 East OMA 1994 East OMA 2002 East OMA 1998 East OMA East OMA 2006 2008 The four combinations of mother’s age and education Here is a map showing the births per acre for older moms (age 30+) who have a baccalaureate degree of higher (42% of the District’s births from 20072009). The counterpoint map shows mothers under age 30 with less than baccalaureate level of education. This area include the areas with large Hispanic populations (31%). Mothers under age 30 with baccalaureate level of education. This may be a younger group of gentrifying households, but perhaps many of them are childless (10%). The group of older mothers age 30 plus with less than a baccalaureate education (16%). And finally, a combined map of older more educated and younger less educated mothers (combined 73% of births). Gentrification based on Percent College Educated and Employed in MTP Occupations – following Ley, 1996 From 1990 to 2000 the average change in proportion with college degree and employed in MTP occupations rose around the edge of the East OMA. From 2000 to 2008 the measure of gentrification spread out beyond the East OMA. Ley depicts gentrifying areas as being in the top 20% of tracts with respect to change in the composite education and MTP employment index. East OMA Conclusions Yes, the shift in births from younger to older mothers and gentrification of some of Portland’s 100 year old housing did help turn around the decline in the district, and likely saved some schools from closing. Our school demography work has been successful with many districts returning for new or updated forecasts. PRC has a reputation for unbiased and well thought out forecasts, although we have not always been right. We support about 1.5 researchers doing enrollment forecasts. The individual projects bring in from $3,000 for a simple update to $60,000 for a metropolitan district. A large part of the cost of a school forecast is in the database development. This provides employment for students with GIS skills. Repeat business with districts lowers the cost for this part of the work. Should you take on this type of work? Good – Learn about the social geography of communities, provide employment for students, satisfaction with providing a valuable service, discover puzzles to be solved. No so good – Need long term commitment, patience in working with school administrators and committees, may be hard to manage time commitments if teaching. Sources of information The website for the Population Research Center. Examples of forecasts: http://www.pdx.edu/prc/school-enrollment Papers and presentations on school demography: http://www.pdx.edu/prc/publications Books Plane, David and Peter A. Rogerson. The Geographical Analysis of Population with Applications to Planning and Business, John Wiley and Sons, 1993. Siegal, Jacob S. and David A. Swanson. The Methods and Materials of Demography, Second Edition. Elsevier Academic Press, 2004. Conferences to learn, present AAG, Population Specialty Group Population Association of American, Applied Demography Specialty Group Applied Demography biannual conference ESRI Education Conference Richard Lycan – lycand@pdx.edu Southern Demographic Association Charles Rynerson – rynerson@pdx.edu Fee Schedule .016d(02) - Population Research Center – Fee Schedule District-wide forecast by grade level $6,000 & up Update district-wide school enrollment forecast by grade level. $3,000 & up District-wide forecast and individual schools or attendance areas forecast $8,000 & up Update district-wide school enrollment forecast and individual schools or attendance area forecast. $4,000 & up