Urban educators have. New Schools, Overcrowding Relief, and Achievement Gains in Los Angeles Strong Returns from a $19.5 Billion Investment

Transcription

1 AUGUST 2012 New Schools, Overcrowdig Relief, ad Achievemet Gais i Los Ageles Strog Returs from a $19.5 Billio Ivestmet William Welsh, Eri Coghla, Bruce Fuller, Luke Dauter Uiversity of Califoria, Berkeley Policy Brief 12-2 William Welsh, lead author, is a Ph.D. studet i the Departmet of Sociology, UC Berkeley. His work focuses o the social cosequeces of ew forms of schoolig. Eri Coghla is a Ph.D. studet i the Graduate School of Educatio, UC Berkeley. She studies issues of school choice ad decetralizig educatio reforms. Bruce Fuller, Professor of Educatio ad Public Policy at UC Berkeley, works o the effects of decetralizig public istitutios ad sociology of the family. Luke Dauter is a Ph.D. cadidate i the Departmet of Sociology, UC Berkeley. His work focuses o competitio ad istitutioal chage i public educatio. Urba educators have struggled with agig, overcrowded schools sice the late ieteeth cetury. I cotemporary Los Ageles, by the mid-1990s, the cout of packed campuses operatig year-roud o multiple shifts had reached uprecedeted umbers. Sice the 1960s the studet populatio of the Los Ageles Uified School District (LAUSD) had icreased by roughly 250,000 studets ad shifted dramatically from 85% white, mostly middle-class to four-fifths Asia, Black, ad Latio families, typically livig i low-icome eighborhoods (Kercher et al., 2008). L.A. had ot built a ew school sice the 1930s. So, by the 1990s almost 25,000 childre were bused out of highdesity areas to faraway schools with sufficiet space (Oakes, 2002). The Rodriguez coset decree brought by families livig i the most desely populated areas, alog with heated civic activism, pushed LAUSD to set goals for reducig overcrowdig. This pressig problem, which likely exacerbated achievemet gaps amog studets, caught the attetio of voters. The L.A. electorate evetually backed five local ad state ballot iitiatives, yieldig more tha $19 billio i fresh bod reveues, to fiace a immese facilities costructio program. This program ivolves over 130 ew facili- Executive Summary Aimig to relieve overcrowded schools operatig o multiple tracks, the Los Ageles Uified School District (LAUSD) has ivested more tha $19 billio to build 130 ew facilities over the past decade. District leaders asked researchers at Berkeley to estimate the achievemet effects of this massive iitiative beefits that may stem from eterig a bright, ew school or from exitig a agig, eve dilapidated facility burstig at the seams with studets. Previous reports from the project examied how ew schools ad alterative schools have affected rates of studet mobility ad teacher turover across LAUSD. 1 By trackig thousads of studets who moved from overcrowded to ew facilities over the period, we discovered robust achievemet gais but also ucovered questios related to future deliberatios. Key fidigs iclude: The steady opeig of ew schools dramatically relieved overcrowdig i elemetary ad secodary schools ad set i motio a complex migratio of Cotiued o page 2.

2 P O L I C Y B R I E F Executive Summary (Cot.) studets, both to ew facilities as well as to a growig umber of charter ad pilot schools. Sigificat achievemet gais are discerible for elemetary-school pupils who switched from a old facility to a ewly costructed facility. O average, these switchig pupils outpaced the average LAUSD studet by a gai equal to about 35 additioal days of istructio each year. Achievemet gais are most robust for elemetary studets who escaped severe overcrowdig by movig to a ew elemetary school. Relative to the rate of learig for the average LAUSD studet, this subset of studets ejoyed achievemet gais equivalet to about 65 days of additioal istructio per year. Studets migratig to certai ew elemetary schools experieced eve stroger gais. Across ew elemetary schools, we fid o relatioship betwee the per-pupil costructio costs directly tied to classrooms ad the magitude of achievemet ties, from early learig ceters to large, elegat high schools, as well as hudreds of reovatio projects (Fuller, Recios, & Scholl, 2008). With the exceptio of the federal iterstate highway system, LAUSD s costructio program is the largest public works gais. That is, pupils migratig to less-costly ew schools saw achievemet gais that were o differet, o average, from those movig to more expesive ew facilities. After a ew school opeed earby, studets who remaied i previously overcrowded elemetary schools experieced modest gais, compared with the average LAUSD studet. Although ew facilities featured slightly lower pupil-teacher ratios, higher shares of fully credetialed teachers, ad lower teacher turover, these features do ot explai the steeper achievemet growth of elemetary studets migratig to these ew facilities. Additioal research could ucover the deeper factors that explai the buoyat achievemets results. We could oly discer icosistet ad weaker achievemet gais for high school studets who moved from a overcrowded to a ew school facility. project ever udertake i the Uited States (Fuller et al., 2009). But does eterig a ew school, or movig studets out of severely overcrowded coditios, sigificatly boost studet learig? This is the bottom-lie questio asked by leaders of LAUSD s facilities iitiative ad the topic explored i this brief. Do Facilities Cotribute to School Quality i Potet Ways? Covicig evidece has show that higher-achievig studets atted higher quality facilities, while low-performig studets ofte gai access oly to agig, ofte overcrowded schools. However, this does ot ecessarily mea that shiy, iovative campuses directly raise test scores. Higherachievig studets ofte beefit from strog family support ad other aspects of school quality that cotribute to their achievemet growth. L.A. s massive experimet i movig thousads of studets from severely overcrowded to ew facilities offers a rare chace to look at the specific effects of this migratio, allowig us to move beyod correlatioal evidece ad make stroger causal ifereces. The estimated effects of migratig to a ew school also hold immediate implicatios, as LAUSD shapes ew ivestmets i school reovatio ad icreasigly shares ew facilities with alterative school maagers, icludig charter ad teacher-led pilot schools. Earlier correlatioal studies suggest that certai features of school desig such as clea air, good light, ad a comfortable ad safe learig eviromet result i stroger pupil egagemet ad achievemet (Scheider, 2002; Lemasters, 1997; Lackey, 1999; ad Cotto, 2001). Other elemets of school quality related to better 2 NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES

3 facilities, such as a eriched academic climate, or higher teacher morale ad lower staff turover, are also associated with better studet outcomes, stemmig from more robust workig coditios (Buckley et al., 2005; Ulie & Tschae-Mora, 2008). But to date little evidece exists to support the claim that the quality of school facilities directly iflueces educatioal outcomes. Earlier work suffers from small samples of schools, lack of cotrol for cofoudig factors, ad scarcity of logitudial data. More recet studies have a stroger empirical base, but results correlatig school facilities ad achievemet remai icoclusive (Bowers & Urick, 2011; Nielso & Zimmerma, 2011). This brief uses high-quality logitudial data recetly made available by LAUSD to derive more rigorous estimates of how ew facilities affect studet mobility ad achievemet. The study examies erollmet ad test scores for early 20,000 elemetary ad high school studets from , durig the first phase of the ew facilities costructio project. Takig the case of desely populated Local District 6 (LD6), we begi by detailig the extet of studet movemet from old, overcrowded buildigs to ew schools. Next we examie the achievemet effects for elemetary ad high school studets throughout LAUSD, focusig o the discrete beefits for studets who left severely overcrowded schools for ew facilities. We also ask whether studets remaiig behid i older schools beefit academically from erollmet relief. That is, gais may be experieced both by those who migrate to ew facilities ad by pupils who are ow served by less-crowded schools. How New Facilities Relieved Overcrowdig Studet movemet as ew schools ad charters ope Before estimatig achievemet effects from studet movemet, we cosider the extet of studet migratio sparked by the District s ambitious costructio program. To illustrate, we aalyze studet movemet i the LAUSD subdistrict that became ifamous for its high levels of overcrowdig: Local District 6, servig the Southeast Cities of Maywood, South Gate, Hutigto Park, Cudahy, Vero, ad Bell. We detail studet movemet from overcrowded high schools to two ew high schools that opeed i This case of studet migratio, icludig ito the risig cout of charter schools, focuses o high schools. For may years, the three large high schools i Local District 6 (LD6) were severely overcrowded. South Gate, Hutigto Park, ad Jorda each served early 5,000 studets, although they were desiged to serve far fewer; each operated multitrack caledars for over 20 years. I 2005, Maywood ad South East high schools opeed earby ad bega to alleviate overcrowdig. New facilities were sited ear overcrowded facilities so that their catchmet areas would iclude may of the studets previously assiged to the overcrowded facility. Whe the ew facilities opeed, catchmet areas were redraw such that some studets previously assiged to old facilities were assiged to ew facilities. Of course, some studets likely moved ito a ew facility s catchmet area, ad some parets likely foud ways for their studets to be assiged to a ew facility despite ot livig i the catchmet area. But for the purpose of our aalysis, o categorical differece exists betwee studets who could ad studets who could ot atted a ew facility. Eve small systematic differeces would ot affect the results of our achievemet aalysis, because our procedures cotrol for all uobserved characteristics of studets. Figure 1 shows erollmet chages i LD6 whe Maywood ad South East high schools opeed. The arrows show the umber of studets who trasferred from a old high school facility to a ew school. (Movemet from middle schools ito ew high schools is ot show.) Withi the first year of opeig, South East high school erolled over 2,000 studets, drawig a majority of its erollmet from severely overcrowded South Gate High School, eighborig David Starr Jorda, ad Hutigto Park. Maywood erolled fewer studets i its first year of operatio, but drew studets from earby Bell ad Hutigto high schools. Thus, although Figure 1 oly shows movemet of studets to Maywood ad South East, may of the schools i LD6 saw erollmet drops, due to the variety of ew schools opeig. NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES 3

4 Compto Ave Sata Fe Ave Pacific Blvd Soto St P O L I C Y B R I E F FIGURE 1. Local District 6: Studets Move from Overcrowded to New High Schools, 2005 Cetral Ave Machester Ave Cetury Blvd Los Ageles Uified School District Local District 6: Iitial Shift of Studets from Overcrowded to New High Schools, st St Vero Ave LOCAL DISTRICT 5 Florece Ave LOCAL DISTRICT 7 July 22, 2012 HUNTINGTON PARK SENIOR HIGH SCHOOL Erollmet: 4,598 SOUTH EAST HIGH SCHOOL Erollmet: 2, > Chage i studet mobility over time Wilmigto Ave 38th St By 2008, two ew charters, three ew mixed schools, ad two ew middle schools opeed i LD6. By this time, Maywood ad South East cotiued to pick up erollmet from the overcrowded schools as well. South Gate had set over 1,000 studets to South East, David Starr Jorda had set early 200, ad Hutigto Park had set over 160 studets. Maywood received over 160 studets from Bell ad 30 from Hutigto Park. By 2008, the majority of middle ad seior high schools i LD6 had experieced erollmet drops or stability from the previous year, as relief played out. Alameda St 103rd St 37th St <-- 86 DAVID STARR JORDAN SENIOR HIGH SCHOOL Erollmet: 2,478 LOCAL DISTRICT 6 < Log Beach Blvd Califoria Ave Gage Ave SOUTH GATE SENIOR HIGH SCHOOL Erollmet: 3,726 Abbott Rd Firestoe Blvd Tweedy Blvd Badii Blvd Slauso Ave 14 --> Imperial Hwy > Atlatic Ave MAYWOOD HIGH SCHOOL Erollmet: 554 BELL SENIOR HIGH SCHOOL Erollmet: 4,737 Atlatic Blvd I summary, we see that both ew school costructio ad the opeig of charter schools spurred may families to switch schools. At the same time, erollmets shrak i the previously overcrowded schools. So, the school eviromets chaged for studets who migrated ad for those who remaied behid i ow-smaller, older schools. Did Studets Ejoy Achievemet Beefits Whe Movig to New Schools? Iterstate 710 The logitudial studet data system, built by LAUSD, allows us to estimate achievemet beefits for studets who switched from a previously overcrowded school to a ew facility. This iitial aalysis focuses solely o the over 20,000 studets who made this switch durig the period. We report results for elemetary ad Garfield Ave Easter Ave Telegraph Rd New School Major Seder Washigto Blvd Mixed School Middle School High School Charter School Maget Program Local District Miles high school studets separately, usig what are called withi-idividual fixed-effects models. This statistical procedure gauges chage i test scores for idividual studets as they move from a overcrowded to a ew facility. The methodological appedix at the ed of this brief provides more details. After cotrollig for studets previous test scores, the age of the school, whether it was a studet s first year of school, the umber of istructioal days i the school caledar, ad all uobserved characteristics of each studet (such as parets press for achievemet, studet motivatio, ad peer effects), we estimated average year-to-year growth i studet scores o the Califoria Stadards Tests (CSTs) as a result of movig to a ew facility. Table 1 gives summary statistics for our models. Good ews for elemetary school studets Figure 2 summarizes our results. We foud that ew elemetary school facilities, after their iitial two years, provided a average boost to achievemet of about 0.18 of a stadard deviatio (SD) i math ad 0.20 SD i laguage arts for each year that the studet was i the ew facility. Box 1 explais why we stadardized the magitude of achievemet effects i this way, equatig the gai to additioal istructioal days. Studets switchig ito ew high school facilities was associated with a statistically sigificat average gai i 4 NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES

5 TABLE 1. Elemetary ad Secodary Studets Switchig to New Schools (excludig middle schoolers) laguage arts of about 0.13 SD uits. I mathematics, studets migratig to ew high schools performed at lower levels, although this differece is ot statistically sigificat. These results did ot differ by ethicity, reducedprice meal eligibility, special educatio or Eglish proficiecy level status. Pupils switchig to a ew elemetary school experieced their smallest gais i the first year after opeig. I subsequet years, achievemet gais icreased, which we attribute to educators hittig their stride i the ew facilities. However, high schools did ot show comparable year-by-year Elemetary School High School Math Laguage Arts Math Laguage Arts Number of studet 6,105 6,113 6,554 7,148 switchers Average umber of observatios (years per studet) FIGURE 2. Achievemet Chage i Studet s Secod Year or Greater Elemetary High School Math Laguage Arts Math Laguage Arts Average stadard-deviatio chage i achievemet attributable to beig i ew facility Whe cotrols for teacher educatio ad experiece are cosidered icreases i the gais of studets who switched schools. Difficult to explai why ew facilities lifted achievemet The positive effect of switchig to a ew elemetary school may have bee due to ew schools attractig more effective teachers tha old schools. To test this hypothesis, we reestimated our models, this time cotrollig i each year for teachers average level of educatio ad years of teachig experiece at each school. Addig these cotrols reduced the ew school effect across the board (see Figure 2). I particular, the beefits experieced by studets HOW MUCH IS A STANDARD DEVIATION GAIN? Because the tests that studets take i differet grades do ot measure precisely the same abilities, our aalysis depeds o stadardizig studets scores. That is, we scale their raw test scores so that the average score i each grade is zero. The, the uit we use to measure scores chages from poits to stadard deviatios. Istead of sayig that Studet A scored te poits better o a test tha Studet B, we say that Studet A scored, for example, 0.25 stadard deviatios better o the test. Although this metric caot precisely covert score icreases to additioal days i school, prior research gives us a rough idea. Usig Hill et al. s (2008) results from atioal logitudial data, the elemetary ew-school beefit correspods to about 35 istructioal days i math ad 45 days i laguage arts. Usig Rathbu ad West s (2004) results revises the estimate dowward betwee 15 ad 25 istructioal days i both math ad laguage arts. i ew high schools ca be almost etirely explaied by the educatio ad experiece levels of their teachers. I elemetary school, a additioal statistically sigificat effect of ew facilities remais eve whe teacherquality cotrols are icluded. This result idicates that ew facilities boosted elemetary studets achievemet growth above ad beyod what would be predicted by simply attractig more qualified teachers from elsewhere i the district. NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES 5

6 P O L I C Y B R I E F TABLE 2. Measures of Resource Edowmet i Old ad New Facilities What could accout for this additioal effect? Oe hypothesis is that ew schools ejoy stroger resources tha old schools, such as better classroom eviromets ad istructioal equipmet. While our data do ot cotai highly precise measures of resource edowmet by school, we were able to costruct four measures that capture differet elemets: the mea ratio of studets to teachers i each school, the mea rate of teacher turover, the mea percet of teachers with full credetials, ad (for ew facilities) a average cost per studet-year (total costructio cost adjusted for iflatio ad divided by umber of seats). Table 2 shows that ew ad old facilities did ot differ appreciably o ay of the first three measures, which are weighted by umber of studets i each school. Although the average effect of switchig to a ew elemetary school was positive, differet schools showed variatio aroud this mea (agai, cotrollig for school age, studet etrace status, prior scores ad uobserved Elemetary School High School Old New Old New Average Percetage of Teachers with Full Credetials 97% 98% 86% 81% Average Ratio of Total Studets to Total Teachers Average Percetage of Teachers Leavig School per Year 12% 8% 25% 19% Teachers Average Years of Experiece Teachers Average Level of Educatio (Stadardized).06 SD +.06 SD.33 SD +.32 SD pupil characteristics). We attempted to assess whether such variatio was associated with the above-metioed school-level factors. Were the more effective ew facilities those that were more expesive, or that had better studet-teacher ratios, teacher turover rates, or teacher credetials? We foud that oe of these factors had a statistically sigificat associatio with achievemet gais. I particular, although iflatio-adjusted costructio cost per seat raged from $12,000 to $22,000, studets i higher-cost facilities did ot show higher growth tha peers i lower-cost facilities. Figure 3 displays the achievemet gais associated with each idividual ew elemetary school (the same model reported i Figure 2, except ru for each ew facility separately), plotted agaist iflatio-adjusted costructio cost per seat. Similarly, at the high school level, resource edowmets were ot sigificatly associated with high-beefit schools. Bigger gais whe movig from a severely overcrowded school The oe factor that sigificatly predicted stroger achievemet gais was the degree of overcrowdig i the school from which studets switched. We quatified overcrowdig by measurig the percet erollmet declie i each old-facility school from 2002 to This procedure is justified by the fact that few ew schools opeed before 2002, ad was a period of extesive facility costructio. Thus, the schools that lost the greatest percetage of erollmet over this time were the most overcrowded oes. We the divided our sample of studets ito two groups: those who switched from severely overcrowded schools, which lost betwee 37% ad 61% of erollmet, ad those who switched from moderately or ot overcrowded schools, which lost less tha 37% of erollmet (or icreased their erollmet over time). We ra the same model o each of the two groups of studets separately. 2 Those from severely overcrowded schools saw the largest math gais from eterig a ew school, while the average gai for studets from less-overcrowded schools was smaller ad statistically isigificat (Figure 4). Eve greater differeces were observed for laguage arts, ad the effect became more dramatic whe schools were broke dow ito fier gradatios of overcrowdig (ot show). However, disaggregatig high school studets i this way revealed o sigificat effects o achievemet. 6 NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES

7 FIGURE 3. Elemetary School Achievemet Gais Not Liked to Direct Classsroom Costructio Costs (per studet) $10,000 $12,000 $14,000 $16,000 $18,000 $20,000 $22,000 FIGURE 4. Studets Switchig from the Most Severely Overcrowded Elemetary Schools Saw Greater Beefits Math Did Relief from Overcrowdig Also Yield Achievemet Beefits? To further ivestigate the hypothesis that overcrowdig relief was a additioal factor i raisig studet achievemet, we looked at studets who remaied behid i the old facilities Elemetary Laguage Arts Achievemet gai for studets from schools with below-media overcrowdig Achievemet gai for studets from schools with above-media overcrowdig that set the greatest umber of studets to ew facilities. If overcrowdig relief was a major factor boostig the test scores of studets who switched schools, we wated to kow if the studets who stayed behid also experieced a boost. For this aalysis, we isolated the old facilities which set at least 1 percet of their studets to a ew facility i the years observed. We aalyzed the test score growth of studets who stayed i these major sedig schools. 3 We compared test score growth i the years before the earest ew facility opeed agaist subsequet years. We foud that stay-behid elemetary studets also ejoyed achievemet gais after ew facilities opeed that were statistically sigificat i the case of laguage arts. These gais were greater tha the gais that all other LAUSD studets experieced i the same time period cotrollig for all studet-level characteristics. See Figure 5 for the compariso, otig the modified vertical scale. These results suggest that the studets who stayed i the major sedig schools also felt the beefits of relief from overcrowdig. However, the scores of high school studets who stayed at these schools did ot differ from average LAUSD studet scores at the high school level. The fact that studets i major sedig schools saw above-average beefits from havig a ew school ope earby suggests that overcrowdig relief cotributed to the achievemet gais that switchers experieced. However, the fact that switchers saw much greater beefits tha the studets that stayed behid i sedig schools shows that the ewess of the school also cotributed. All-Star Elemetary Schools We idetified a small umber of ew elemetary facilities whose studets cosistetly showed achievemet NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES 7

8 P O L I C Y B R I E F growth at rates far above the mea LAUSD pupil. These results, compared with the mea ew-school studet achievemet gai i three time periods, appear i Table 3. At the high school level, Academic Leadership Commuity ad Academic Performace Excellece Academy stood out as showig positive gais over ad above studets previous trajectories i both math ad laguage arts over all three years. Summary ad Implicatios for Facilities Policy Our fidigs reveal strog achievemet beefits from L.A. s $19.5 billio ivestmet i ew schools at least for studets migratig to ew elemetary schools. These youg studets who switched from overcrowded to ew schools durig the years displayed achievemet gais equal to about 35 days of additioal istructio. The magitude of this gai was eve greater for elemetary studets who exited the most severely overcrowded schools. We did ot fid similar gais for high school studets who moved from overcrowded to ew schools, a empirical mystery that requires additioal research. As erollmet relief was accomplished by opeig ew schools earby, studets who remaied behid i less overcrowded schools also experieced sigificat gais i achievemet. This suggests desigers of future efforts for example, whe reovatig old schools should aticipate the tadem beefits expected i both the ew ad old facilities. FIGURE 5. Studets who Stayed i Major Elemetary Sedig Schools Experieced Modest Gais from Havig a New Facility Ope Nearby TABLE We could ot pipoit the determiig factors that explai the positive effects experieced by studets. The collateral improvemet i teacher qualificatios displayed by ew schools appears to have played a role, especially i attractig youger teachers with masters-level traiig. However, more research is required to uderstad the igrediets of quality or social relatios that mark ew or less crowded facilities that i tur pay off i higher achievemet. Math Elemetary Achievemet gai after 2005 for all other elemetary LAUSD studets Achievemet gai after a earby ew facility opes for studets i major sedig schools All-Star New Elemetary Facilities: Average Chage i Studet Achievemet Growth (SDs) Math Laguage Arts Laguage Arts Year 1 Year 2 Year 3+ Year 1 Year 2 Year 3+ Charles White Elemetary Maywood Elemetary Madiso Elemetary Mea for All New Elemetaries We foud o relatioship betwee the direct cost of costructio related to classrooms ad istructioal space ad achievemet gais amog ew schools. More work is required here as well. But this ull fidig does suggest that LAUSD ad other urba districts might systematically experimet with less costly facilities, perhaps maaged by charter or pilot school leaders, to arrive at more cost-effective buildigs. Such careful experimetatio could lead to more cost-effective ivestmet of reovatio dollars as well. 8 NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES

9 The fact that some ew elemetary schools achieved eve stroger achievemet effects for studets ivites more fie-graied research o how facilities quality iteracts with teacher quality ad school commuity to yield remarkable results. LAUSD faces several policy decisios with implicatios for its wide array of more tha 850 campuses ad allied facilities. The District has haded off ew schools to charter ad pilotschool maagers. These alterative forms of schoolig already experimet with storefrot ad other ovel facilities. Billios of additioal dollars will become available, as the bod market ad tax rates permit, to reovate old, ofte dilapidated schools. As District erollmet cotiues to slip, some facilities are ow uderutilized, a dramatic turaroud from the severe overcrowdig that the massive buildig program has all but alleviated. Policy Implicatios As these issues are take up ad policy optios weighed, these fidigs offer importat evidece ad prompt future lies of iquiry: Higher quality facilities offer ecessary but isufficiet coditios for raisig achievemet. Teacher quality ad relief from overcrowdig play sigificat roles as well. The distict role of facilities i cocert with other teacher ad istructioal resources should be cosidered, especially whe the District cosiders hadig schools off to alterative providers. The fact that costructio costs per pupil are urelated to the magitude of achievemet gais for elemetary studets suggests that margial returs to more expesive facilities may be low. Both charter ad pilot school leaders are experimetig with lower cost facilities. Studyig the discrete achievemet patters associated with such iovative facilities would be iformative As LAUSD moves from costructig ew schools to reovatig old facilities, scope ad cost optios should be weighed carefully. Maiteace remais a competig eed across the District as well. Uderutilized facilities represet a emergig issue, while how iovative desigs are bleded with ivetive teachig methods especially i high schools offers a ew frotier, icludig small learig commuities ad the Liked Learig iitiative. The positive effects for elemetary studets whose schools experieced relief from overcrowdig suggests that takig further steps to reduce erollmet i still desely packed schools could result i additioal gais. The lack of robust achievemet beefits for studets who moved to a ew high school facility is cause for cocer. Other studet outcomes might be studied, ad data should be updated by LAUSD to check for effects as the fial third of ew facilities have come o lie sice Still, somethig is missig beyod fresh facilities as the district attempts to lift achievemet iside high schools. As other urba districts attempt to remedy overcrowdig or to reovate old facilities, the evetual effects o achievemet should be carefully studied ad ot take for grated. NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES 9

10 P O L I C Y B R I E F Edotes Warm thaks go to Gle Daley, Steve Eglish, Mary Filardo, Neil Gamble, Kathy Hayes, Cythia Lim, Guy Mehula, Paul Miller, Babatude Oguwole, Rea Perez, Amada Rios, Jeff Vicet, ad Jeff White for their uwaverig support ad steady advice. Russ Rumberger, members of the LAUSD Research Committee, ad two aoymous reviewers offered helpful commets o earlier drafts. This work was supported by the Ford ad Specer foudatios. The Hewlett Foudatio supported Mr. Dauter s time via the research etwork, Policy Aalysis for Califoria Educatio. 1 All research reports from the Los Ageles School Ifrastructure Project appear at: 2 We tested this operatioalizatio for robustess by re-ruig the aalysis usig the district s schoollevel classificatio, a biary variable idetifyig 80 schools as previously overcrowded. We the divided our sample of studets ito those who switched from overcrowded schools, ad those who switched from ot-overcrowded schools. While the lists of schools were ot idetical, the effect sizes, directios, ad sigificace levels were early exactly the same as whe usig our classificatio. 3 Note that this sample of studets is ot the same as the samples of studets discussed above, ad a proper subset of the sample discussed i the last sectio. Refereces Bowers, A., & Urick, A. (2011). Does high school facility quality affect studet achievemet? A two-level hierarchical liear model. Joural of Educatio Fiace, 37(1), Cotto, K. (2001). New small learig commuities: Fidigs from recet literature. Portlad, OR: Northwest Regioal Educatio Laboratory. Retrieved 03/03/04 from: scpd/sirs/slc.pdf Fuller, B.; Dauter, L.; Hosek, A.; Kirschebaum, G.; McKoy, D.; Rigby, J.; & Vicet, J. M. (2009). Buildig schools, rethikig quality? Early lessos from Los Ageles. Joural of Educatioal Admiistratio, 47, Fuller, B.; Recios, A.; & Scholl, B. (2008). Voter support for Los Ageles school facility bods, Berkeley; Los Ageles School Ifrastructure Project, Graduate School of Educatio. Retrieved from laschoolreform/ Hill, C. J., Bloom, H., Black, A. R., & Lipsey, M. W. (2008). Empirical bechmarks for iterpretig effect sizes i research. Child Developmet Perspectives, 2(3), Kercher, C. T., Meefee-Libey, D. J., Mulfiger, L. S., & Clayto, S. E. (2008). Learig from L.A.: Istitutioal chage i America public educatio. Cambridge, MA: Harvard Educatio Press. Lackey, J. A. (1999). Assessig school facilities for learig/assessig the impact of the physical eviromet o the educatioal process. Mississippi State, MS: Educatioal Desig Istitute. Lemasters, L. K. (1997). A sythesis of studies pertaiig to facilities, studet achievemet, ad studet behavior (Upublished doctoral dissertatio ED447687). Virgiia Polytechic ad State Uiversity. Los Ageles Uified School District, Facilities Services Divisio. (2010). New costructio: Strategic executio pla Jauary Available at about_fsd/sep/2010_sep.pdf Nielso, C., & Zimmerma, S. (2011). The effect of school costructio o test scores, school erollmet, ad home prices. The Istitute for the Study of Labor, IZA Discussio Paper No Oakes, J. (2002). Cocept 6 ad busig to relieve overcrowdig: Structural iequality i Califoria schools. UCLA s Istitute for Democracy, Educatio, ad Access. Retrieved from escholarship.org/uc/item/30q9d8x Rathbu, A., & West, J. (2004). From Kidergarte Through Third Grade: Childre s Begiig School Experieces (NCES ). U.S. Departmet of Educatio, Natioal Ceter for Educatio Statistics. Washigto, DC: U.S. Govermet Pritig Office. Scheider, M. (2002). Do school facilities affect academic outcomes? Washigto, D.C.: Natioal Clearighouse for Educatioal Facilities. Retrieved from pdf Ulie, C., & Tschae-Mora, M. (2008). The walls speak: The iterplay of quality facilities, school climate, ad studet achievemet. Joural of Educatioal Admiistratio, 46, NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES

11 Methodological Appedix All LAUSD studets who switched from old to ew facilities betwee 2002 ad 2009 are icluded i the model. I the process of determiig which studets were switchers, we excluded may ew facilities from our list of ew schools whose achievemet effects we could ot estimate. First, because o test-score data exists for their studets, early-educatio ew facilities (schools oly servig pre-k, kidergarte ad first-grade studets) were excluded. Secod, because we wished to focus o elemetary ad high school studets, we did ot examie ew juior high facilities (servig grades 6 8 oly). Fially, we did ot iclude ew facilities that housed cotiuatio high schools, special educatio facilities, opportuity schools, ewcomer ceters, or commuity day schools. Out of 73 total ew facilities ope durig , 44 met our criteria. To assess effects o achievemet, we first had to create a commo scale for each of our two depedet variables, the CST math ad laguage arts scores. Because the tests i each grade are ot vertically equated (the items measure differet cocepts), we stadardized all scores, covertig them to a commo z-score metric. For Laguage Arts CST tests, which are the same for each studet i a give grade ad a give year, we stadardized by grade-year: i each caledar year, we coverted each grade s test scores to a distributio with a mea of zero ad stadard deviatio of oe. For Mathematics CST tests, we stadardized by grade-year-test; that is, i each caledar year, ad i each grade, we coverted the scores of studets takig each test to a distributio with a mea of zero ad stadard deviatio of oe. I this way, we miimize distortio i our data caused by differeces betwee tests. For example, a studet who raked highly i 10th grade i 2003 takig Geometry, ad who raked eve higher i 11th grade i 2004 takig Algebra II, will show the appropriate icrease, eve if her raw CST scale score is lower i the secod year. I additio, for our model of high school math achievemet, we icluded cotrols for studet s differet progressio DDDDDD = YY!" YY!"!! = ββ! oooooo + ββ! ffffffffffffffffff + (ββ!! oooooo ssss oooooooooooo! )!!"#!! + (ββ!! gggggggggg! ) + ββ! ss oooooooooooo + ββ! (ss oooooooooooo! ss oooooooooooo!!! )!!!!"#! + (ββ! tttttttt eeeeee aaaa ) + uu! + εε!"!! through the mathematics sequece. Fially, the depedet variable i our model is achievemet growth, the differece betwee the curret year s test score ad the score o the last test that the studet took. To fid the effect o achievemet of switchig to a ew school, we used a fixedeffects model of achievemet growth. A fixed-effects model takes advatage of the fact that we have multiple observatios for each studet to cotrol for all uobserved, time-ivariat characteristics (UTICs) that might have a effect o the depedet variable (achievemet). The model cotrols for UTICs by subtractig studets idividual meas for the depedet variable from each of their time-specific observatios. The same is doe for idepedet variables. As a result, the effects of UTICs (or fixed effects ) that led a give studet to have a idividual mea that is above or below the grad mea o ay variable are dropped from the aalysis. This leaves oly the effects of chages i the idepedet variables o chages i the depedet variable. I the preset paper, the idepedet variable of iterest is a chage i studet erollmet from a old facility to a ew facility. Thus, to isolate the effect of this switch, a fixed-effect model is most appropriate. To specify our model, we used the Stata statistical package to estimate coefficiet values for the followig equatio: Where Y it is the z-scored test score (either math or laguage arts, stadardized as per above) for studet i i time t; ewschool is a dummy variable coded 1 if the studet is erolled at a ew facility ad 0 otherwise; firstyear is a dummy variable coded 1 if the observatio was made durig the studet s first year at a give school ad 0 otherwise; schoolage j is a dummy variable!!!! coded 1 if the observatio was made i the school s jth year of operatio ad 0 otherwise, (the referece category thus beig ew schools third or greater year of operatio); grade k is a dummy variable coded 1 if the studet s curret grade is k ad 0 otherwise; g mi is 2 for elemetary studets ad 9 for high school studets; g max is 5 for elemetary studets ad 11 for high school studets; shortcal is a dummy variable coded 1 if the studet was o a shorter, 163-day caledar i that year ad 0 if the studet was o a 180-day caledar; teacherchar h are a list of teacher characteristics (educatio ad experiece, omitted i the dark bars of Figure 2); u i is a studet-level fixed effect (the studet s mea stadardized chage from year to year); ad ε it is the observatio-level error. Thus, β 1, the coefficiet of iterest for this brief, reports the average chage i studets relative stadig amog district peers upo movig to a ew school if the school has bee i operatio for more tha two years. The statistic is adjusted for the dowward shock of it beig the first year at a ew school, studets curret grade level, the caledar track of the school, ad all time-ivariat uobserved characteristics of the studet. Mea estimated values of β 1 are what is reported i all bar graphs i this brief. Table 1 shows sample sizes for each of the models reported i Figure 4. NEW SCHOOLS, OVERCROWDING RELIEF, AND ACHIEVEMENT GAINS IN LOS ANGELES 11

12 School of Educatio, Staford Uiversity 520 Galvez Mall CERAS Rm. 401 Staford, CA (650) We would like to thak the Califoria Educatio Policy Fud (a sposored project of Rockefeller Philathropy Advisors), the Dirk ad Charlee Kabceell Foudatio, the James Irvie Foudatio, ad the Stuart Foudatio for fiacial support for the publicatio of this policy brief. The views expressed are those of the authors, ad do ot ecessarily reflect the views of PACE or its fuders. Recet PACE Publicatios Robert Liquati ad Keji Hakuta. How Next-Geeratio Stadards ad Assessmets Ca Foster Success for Califoria s Eglish Learers. Policy Brief 12-1, July W. Charles Wiseley. Effective Basic Skills Istructio: The Case for Cotextualized Developmetal Math. Policy Brief 11-1, Jauary Mary Perry. School Fiace Reform A Weighted Pupil Formula for Califoria. Report 1, May Gettig Dow to Facts: Five Years Later. May Hilary McLea. Califoria s Early Assessmet Program: Its Effectiveess ad the Obstacles to Successful Program Implemetatio. March Michal Kurlaeder, Eric Grodsky, Samuel J. Agroow, Catherie L. Hor. State Stadards, the SAT, ad Admissio to the Uiversity of Califoria. Policy Brief 11-3, November The Road Ahead for State Assessmets. Research Report, May PACE ad Reie Ceter for Educatio Research & Policy. William S. Koski ad Aaro Tag. Teacher Employmet ad Collective Bargaiig Laws i Califoria: Structurig School District Discretio over Teacher Employmet. Policy Brief 11-2, February Douglas N. Harris. Value-Added Measures of Educatio Performace: Clearig Away the Smoke ad Mirrors. Policy Brief 10-4, October Jeifer L. Steele, Richard J. Murae, ad Joh B. Willett. Do Fiacial Icetives Draw Promisig Teachers to Low- Performig Schools? Policy Brief 10-3, May Marti Caroy. Califoria s Impedig College Graduate Crisis ad What Needs To Be Doe About It. Policy Brief 10-2, April Reformig Educatio i Califoria: A Guide for Citizes ad Cadidates. April Charles Taylor Kercher. There s Lots to Lear from L.A.: Policy Levers for Istitutioal Chage. Policy Brief 10-1, Jauary 2010.