Whole-school or strand immersion: what the research says

What is known about running immersion as a whole school versus a strand inside a neighborhood school

PPS staff and board members have called whole-school immersion the better model for dual language immersion (DLI). This page asks what evidence supports that view. In a whole-school program, every class in the building is an immersion class, and there is no neighborhood program. In a strand, an immersion program shares a building with a regular neighborhood program. The page covers:

the published research
the guidance used in the field
what PPS’s own experience at Lent shows.

Short answer

No study we could find compares student outcomes in whole-school and strand immersion programs. The strong research on immersion compares students in immersion with students not in it. Its positive results come from strand programs, whole-school programs and mixes of both.
PPS’s own October 13 slides agree that structure isn’t the main factor. They list conditions for strong immersion programs and say these “matter more than configuration (full-school vs. strand) alone”. None of the studies they cite compares the two models (details).
The case for whole schools rests on staffing and school culture. Field guides say strands can work when the whole school supports them. Some case studies describe tension in strand schools. Neither measures student outcomes.
PPS’s own conversion of Lent is the closest local evidence, and it is mixed. Students who stayed after Lent became a whole school improved. But they are 43% of the original students, and there is no comparison group. The audit says the results “do not prove causality”. Most of the school’s students left that year, disproportionately Black and Asian children. Many were neighborhood students PPS assigned to Marysville, and some were fifth-graders moving up to middle school as usual (details). Lent lost 63% of its staff in one year.
PPS’s stated goal is fewer single strands. A single strand has one immersion class per grade. Two immersion classes per grade in a shared school also meet that goal. PPS already runs strands that size.

Sources: Steele, J. L., Slater, R. O., Zamarro, G., Miller, T., Li, J., Burkhauser, S., & Bacon, M. (2017). Effects of dual-language immersion programs on student achievement: Evidence from lottery data. American Educational Research Journal, 54(1 suppl), 282S–306S (doi; open manuscript); RAND brief RB-9903; Bibler, A. J. Dual language education and student achievement. Education Finance and Policy (doi; quotes from the 2016 working paper, p. 6); Steele, J. L., Watzinger-Tharp, J., Slater, R. O., Roberts, G., & Bowman, K. (2021). Achievement effects of dual language immersion in one-way and two-way programs: Evidence from a state scale-up in Utah (working paper, University of Arkansas, PDF); National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the educational success of children and youth learning English: Promising futures, ch. 8, p. 300 (read online); Howard, E. R., Lindholm-Leary, K. J., Rogers, D., Olague, N., Medina, J., Kennedy, D., Sugarman, J., & Christian, D. (2018). Guiding principles for dual language education (3rd ed.). Center for Applied Linguistics (publisher); Freire, J. A., & Alemán, E., Jr. (2021). “Two schools within a school”: Elitism, divisiveness, and intra-racial gentrification in a dual language strand. Bilingual Research Journal, 44(2), 249–269 (doi); abstract; Green, T. L., & Hanson, H. (2026, June 29). Equity audit evaluation of Portland Public Schools’ Southeast Enrollment and Program Balancing initiative: School-by-school analysis (PPS SEGC documents); PPS Board of Education, regular meeting, October 6, 2026, staff presentation at 2:25:37 (video; automated transcript via Save PDX Schools); PPS, Enrollment by Grade and Program Type, October 2024 and October 2025 (PPS Data & Reporting); PPS, PPS Dual Language Immersion (DLI) Programs, presentation to the Board’s Teaching, Learning, and Enrollment Committee, October 13, 2026 (BoardBook). Method details: Sources & methods.

PPS says the problem is single strands

PPS staff point to two problems: the single strand, and the strain of running two programs in one building. At the October 6, 2026 board meeting, staff explained the plan to consolidate Spanish immersion in North Portland:

“Three of those programs have mostly single strands for both neighborhood and Spanish immersion. What that means is one teacher that does Spanish immersion at the grade level, one teacher that does the neighborhood side at each grade level. … The proposal here is to have less single strands and more areas where teachers can have planning, partnership, and others to work with. … Without making these DLI consolidation moves in the north, we are going to leave single strands across our system.”

A month earlier, staff spoke to the board’s Teaching, Learning and Enrollment Committee. They said that “grade-level teacher collaboration is challenging, particularly in the co-located DLI building where enrollment is uneven and curriculum is different across strands.”

In a single strand, one immersion teacher works alone at each grade. No colleague teaches the same grade in the same language. Until October 13, PPS cited no research for the choice of model. We checked the transcripts of the September and October 2026 board and committee meetings, and the per-school guides.

On October 13, staff brought research slides to the same committee. Slide 3 lists four conditions for strong immersion programs, such as a shared vision and enough bilingual staff. It then says: “These conditions matter more than configuration (full-school vs. strand) alone.” Slide 9 says a whole-school model “is structurally best positioned to deliver” these conditions. The sources listed on slide 14 are practice guides, PPS web pages and studies of PPS’s own immersion programs. None of them compares whole-school and strand programs (see Outcome research).

Sources: PPS Board of Education, regular meeting, October 6, 2026, staff presentation at 2:25:37 (video; automated transcript via Save PDX Schools); PPS Board of Education Teaching, Learning, and Enrollment Committee, September 8, 2026, at 13:15 (video; automated transcript via Save PDX Schools); PPS, How Could Your School Be Impacted? school guides (“PPS Family-Friendly School Guide”, per-school PDFs dated October 8, 2026), page 4, projected enrollment 2027-28 to 2031-32 (page); PPS, PPS Dual Language Immersion (DLI) Programs, presentation to the Board’s Teaching, Learning, and Enrollment Committee, October 13, 2026 (BoardBook). Method details: Sources & methods.

Outcome research compares immersion with no immersion

The outcome studies we found ask whether immersion helps students. They don’t test whether a whole school works better than a strand.

The best evidence on immersion comes from Portland. RAND’s study used PPS’s admission lotteries. Chance decided who won a seat, so winners and losers were alike in other ways. Winners outperformed losers in English reading “by about seven months of learning in grade 5, and nine months of learning in grade 8”. For math and science, the authors found “no statistically significant immersion benefit, but also no detriment”. English learners who won a seat were 3 to 4 percentage points less likely to still be classified as English learners by grades 6 and 7.

The study checked whether results differed by:

partner language
one-way versus two-way programs (a two-way program mixes native English speakers with native speakers of the partner language; a one-way program serves mainly one group)
students’ home language.

It did not compare school structures. We searched its full text for “strand”, “whole-school”, “school-wide” and “magnet” and found no matches. The authors note one reason structure could matter even though they didn’t measure it. Students “who win immersion slots may change not only their classroom placement but the school they attend, and it is possible that features of immersion schools differ in key ways (such as academic culture or parent involvement)” (p. 24).

The other lottery and statewide studies also compare immersion with no immersion. Some report results in standard deviations (SD). A standard deviation measures how widely scores spread across students. Researchers use it to compare results from different tests.

Table 1: Outcome studies of dual language immersion, and whether each compares whole-school with strand programs.

Study	Setting	School structure	What it compares	Main finding
Steele et al. (2017), RAND	Portland Public Schools; admission lotteries	Strands, as Bibler (2016) describes Portland's programs	Immersion vs. not	Reading +7 months by grade 5, +9 months by grade 8; no math or science penalty
Bibler (working paper 2016; published in Education Finance and Policy)	Charlotte-Mecklenburg, NC; lotteries	Whole-school magnets ("every classroom in the school is a dual language classroom")	Immersion vs. not	Gains in math and reading per year of attendance
Steele et al. (2021)	Utah statewide scale-up	Strands ("virtually all schools in Utah that offer DLI programs also offer parallel non-DLI strands")	Immersion vs. not; one-way vs. two-way	Null overall in math and reading in grades 3-6; gains of 0.10-0.11 SD in two-way programs
Watzinger-Tharp et al. (2018)	Utah	Strands (50/50 two-teacher model)	Immersion vs. not	Same third-grade math scores; more math growth
Morales (2025)	10 oversubscribed programs, lotteries (abstract only)	Not stated in abstract	Immersion vs. not	Grades 1-4: +0.12 SD reading, +0.14 SD math for native English speakers
None of these studies compares whole-school with strand programs. Steele et al. (2017) don’t describe their programs’ school structure; Bibler (2016, p. 6) describes them as “strand programs, meaning that they only make up a portion of the school”. Sources: Steele, J. L., Slater, R. O., Zamarro, G., Miller, T., Li, J., Burkhauser, S., & Bacon, M. (2017). Effects of dual-language immersion programs on student achievement: Evidence from lottery data. American Educational Research Journal, 54(1 suppl), 282S–306S (doi; open manuscript); RAND brief RB-9903; Bibler, A. J. Dual language education and student achievement. Education Finance and Policy (doi; quotes from the 2016 working paper, p. 6); Steele, J. L., Watzinger-Tharp, J., Slater, R. O., Roberts, G., & Bowman, K. (2021). Achievement effects of dual language immersion in one-way and two-way programs: Evidence from a state scale-up in Utah (working paper, University of Arkansas, PDF); Watzinger-Tharp, J., Swenson, K., & Mayne, Z. (2018). Academic achievement of students in dual language immersion. International Journal of Bilingual Education and Bilingualism, 21(8), 913–928 (doi); Morales, C. (2025). Dual language immersion programs and student achievement in early elementary grades. Educational Evaluation and Policy Analysis, 47(2), 616–623 (doi); abstract only. Method details: Sources & methods.

Immersion has produced gains as a strand (Utah, most Portland programs) and as a whole school (Charlotte). But those studies differ in place, language, students and method. So comparing them says nothing about which structure works better. The National Academies’ 2017 review looked at research on English learners. It found “virtually no research related to the features of school-wide programs that lead to better student outcomes” (p. 300).

Sources: National Academies of Sciences, Engineering, and Medicine. (2017). Promoting the educational success of children and youth learning English: Promising futures, ch. 8, p. 300 (read online). Method details: Sources & methods.

Field guides accept both models but ask more of strands

The main guides in the field present both models as options. The standard reference asks more of strand schools.

The standard reference is the Center for Applied Linguistics’ Guiding Principles for Dual Language Education. It uses “program” to mean any school with dual language instruction, “regardless of whether the program functions as a strand within a school or as a whole school”. It asks more of strand schools: “If the dual language program is a strand within the school, the program planning should be school-wide and include teachers from all programs in the school”. It also warns that where immersion “exists alongside other programs, the principal may be too busy with the needs of the whole school to provide the necessary instructional leadership specifically for the dual language program”. It cites no study comparing outcomes.

The California Association for Bilingual Education has a planning guide. It lists the choice between “a whole school or a strand within a school program” as the first decision a district makes. It names a cost on each side:

Strand: “a separate program operating within the school … can lead to a sense of the DLI education program feeling disconnected from the school as a whole”.
Whole school: “students living in the site’s attendance area who do not want to participate in the DLI program would have to attend a different school”. Conversion may also mean “fewer positions for non-DLI teachers at the site”. This is what happened at Lent.

Sources: Howard, E. R., Lindholm-Leary, K. J., Rogers, D., Olague, N., Medina, J., Kennedy, D., Sugarman, J., & Christian, D. (2018). Guiding principles for dual language education (3rd ed.). Center for Applied Linguistics (publisher); California Association for Bilingual Education, Dual Language Immersion Planning Guide, “Implementation model” (page), accessed October 9, 2026. Method details: Sources & methods.

Case studies describe strain in strand schools

Two case studies describe the strain a strand can put on a school. Each one looks closely at a single school instead of measuring student outcomes.

Freire and Alemán (2021) followed one urban elementary school with a Spanish-English two-way strand. They write that “with most elementary dual language bilingual education (DLBE) programs being implemented as strands within schools, tensions related to inequities, unequal distribution of resources, and academic quality and demographic differences often arise”. Teachers described the strand in terms of “elitism and divisiveness”.
de Jong and Bearse (2014) studied a two-way strand in one middle school. They found “a basic conflict between effective TWI practices and the middle school structure”. They concluded that “integrated, interdisciplinary approaches that are systemically supported within the school are necessary”. This case is from a middle school, not an elementary school.

These studies show what can go wrong in a strand. They also show what a strand needs: schoolwide support and shared planning. They don’t measure whether students learn more in a whole-school program.

RAND’s classroom observations in PPS found that teachers “consistently implemented key instructional practices targeted by the school district”.

Sources: Freire, J. A., & Alemán, E., Jr. (2021). “Two schools within a school”: Elitism, divisiveness, and intra-racial gentrification in a dual language strand. Bilingual Research Journal, 44(2), 249–269 (doi); abstract; de Jong, E. J., & Bearse, C. I. (2014). Dual language programs as a strand within a secondary school: Dilemmas of school organization and the TWI mission. International Journal of Bilingual Education and Bilingualism, 17(1), 15–31 (doi); abstract; Li, J. J., Steele, J. L., Slater, R., Bacon, M., & Miller, T. (2016). Implementing two-way dual-language immersion programs: Classroom insights from an urban district (RAND Research Brief RB-9921). Method details: Sources & methods.

Single strands and whole schools are different problems

PPS’s argument is about the lone immersion teacher at each grade. A school with two immersion classes per grade gives each teacher a partner who teaches the same grade in the same language. The neighborhood program can stay. In October 2025, 3 of the district’s 15 K-5 immersion sites were whole schools (Lent, Richmond, Rigler). The rest are strands of different sizes:

Table 2: K-5 immersion sites, October 2025: immersion students per grade (K-5 average) and students in the school’s neighborhood program.

School	Language	Model	Immersion students per grade	Neighborhood program, K-5
Ainsworth	Spanish	Strand	48	273
Scott	Spanish	Strand	37	254
Rose City Park	Vietnamese	Strand	34	210
César Chávez	Spanish	Strand	32	78
MLK Jr	Mandarin	Strand	32	126
Woodstock	Mandarin	Strand	32	177
Atkinson	Spanish	Strand	30	188
Beach	Spanish	Strand	26	133
Kelly	Russian	Strand	24	268
Sitton	Spanish	Strand	23	207
James John	Spanish	Strand	22	232
Clark	Mandarin	Strand	15	273
Richmond	Japanese	Whole school	85	–
Rigler	Spanish	Whole school	48	–
Lent	Spanish	Whole school	43	–
PPS doesn’t publish the number of classes per program; students per grade is a rough guide to it. Staff said in October 2026 that César Chávez already has two Spanish classes in most K-5 grades. Sources: PPS, Enrollment by Grade and Program Type, October 2024 and October 2025 (PPS Data & Reporting); PPS Board of Education, regular meeting, October 6, 2026, staff presentation at 2:25:37 (video; automated transcript via Save PDX Schools). Method details: Sources & methods.

PPS already runs larger strands. Ainsworth has about 48 Spanish immersion students per grade, and its neighborhood program is about the same size. Ainsworth draws the most Spanish kindergarten lottery applications in the district: 130 a year on average over 5 lotteries. Many things drive Ainsworth’s demand, including its neighborhood. So this shows that a two-class strand can be popular. It doesn’t show that strands outperform whole schools. In the Southeast, Atkinson has room for a second Spanish class per grade without moving its neighborhood students (Equity §8).

Sources: PPS, Enrollment by Grade and Program Type, October 2024 and October 2025 (PPS Data & Reporting); PPS Enrollment & Transfer, Elementary School / Kindergarten Lottery Results Summary, lotteries for 2022-23 through 2026-27 (prior lottery results); PPS, School Profiles, 2025-26 (Power BI dashboard: functional capacity, utilization, enrollment, direct certification, capture rate; page), retrieved October 9, 2026. Method details: Sources & methods.

Lent, PPS’s own whole-school conversion, brought gains and costs

Lent became a whole-school Spanish immersion school in 2023-24. Bridger’s program moved in, and Lent’s neighborhood program moved to Marysville. The district’s equity audit gives the most detailed account of what a whole-school conversion did in PPS. It reports both gains and costs.

What went well. Participants described the whole-school model “as more coherent, intentional, culturally affirming, and professionally sustaining for many Latino/a/x students, families, bilingual educators, and staff of color”. Staff morale “appears to have improved over time”. The audit followed the immersion students who were already at Lent before the conversion and stayed. It compared them with their own results from the year before: “math growth reached +0.235 standard deviations and ELA +0.409 standard deviations by 2024–25. Chronic absenteeism fell from 42% to 24%”. (ELA means English language arts.) Exclusionary discipline, which removes students from class or school, fell to zero (school-by-school analysis, p. 47).

This is the strongest local evidence for the whole-school model. But it has limits:

The audit itself says “these results do not prove causality”.
There is no comparison group. Nobody measured how immersion students at strand schools changed over the same years. Those were the years when scores recovered after the pandemic.
It covers only the students who stayed: 113 of 265 (43%), “most of whom were Latino/a/x or white” (p. 45, 47).

What it cost. The audit’s main report calls the conversion an example of “simultaneous benefit and burden”:

“nearly 60% of students left the school after 2022–23, primarily to newly assigned elementary and middle schools” (p. 50). Of the 265 students at Lent in spring 2023, 152 left (school-by-school p. 45):
- 79 went to Marysville, the assigned school for neighborhood students.
- 39 moved up to middle school at Kellogg, as fifth-graders do every year.
- 13 went to other schools.
Most departures were by design or routine, not families choosing to leave.
“Black student enrollment dropping from 10% to 2%, while the Latino/a/x share increased from 49% to 69%” (p. 50).
Students who moved to Marysville and Kellogg “showed flat-to-negative growth in math and ELA scores and chronic absenteeism in the 39–43% range” (school-by-school p. 47).
63% of Lent staff left after 2022-23, “nearly double the SEGC-school average of 32%” (school-by-school p. 39).
Lent enrolls about half the 527 students PPS projected for it. See Lent & Chávez and Equity §5.

PPS’s October 13 slides call Lent “a PPS proof point”. They say Lent “produced better outcomes for students, particularly Multilingual Learners and English Learners, who are showing significant gains.” They add that its trajectory “is exactly what the research conditions and the Collier & Thomas data predict”. The slides don’t mention the audit’s limits or the costs above. Collier and Thomas’s best-known paper compares kinds of programs for English learners. It doesn’t compare whole-school and strand programs.

The audit concludes that Lent “demonstrates the promise of a strong, culturally affirming Spanish DLI model, but it also shows the racial equity risks of achieving that model through a process that displaced other students of color” (school-by-school p. 47).

Sources: Green, T. L., & Hanson, H. (2026, June 10). Equity audit evaluation of Portland Public Schools’ Southeast Enrollment and Program Balancing; Green, T. L., & Hanson, H. (2026, June 29). Equity audit evaluation of Portland Public Schools’ Southeast Enrollment and Program Balancing initiative: School-by-school analysis (PPS SEGC documents); PPS, PPS Dual Language Immersion (DLI) Programs, presentation to the Board’s Teaching, Learning, and Enrollment Committee, October 13, 2026 (BoardBook); Collier, V. P., & Thomas, W. P. (2004). The astounding effectiveness of dual language education for all. NABE Journal of Research and Practice, 2(1), 1–20 (PDF). Method details: Sources & methods.

PPS has data that could answer the question

PPS holds data that could compare its own whole-school and strand programs:

Persistence: the share of each kindergarten immersion class still enrolled in grade 5, by site and model. This site can only estimate it from grade-level counts. Those counts mix in students who join later.
Teacher retention: how often immersion teachers leave single-strand, two-strand and whole-school sites.
Achievement by program: Oregon publishes test results for the state, districts and schools. It breaks them out by student group but not by program. So at a strand school, immersion and neighborhood students are mixed together. PPS has student-level results that it could split by program.
A comparison group for Lent: growth for immersion students at strand schools over 2023-24 to 2024-25, measured the same way the audit measured Lent.

Sources: Oregon Department of Education, Assessment Group Reports (state, district and school results by student group; page), accessed October 9, 2026; Green, T. L., & Hanson, H. (2026, June 29). Equity audit evaluation of Portland Public Schools’ Southeast Enrollment and Program Balancing initiative: School-by-school analysis (PPS SEGC documents). Method details: Sources & methods.

Questions for the board

These are this site’s questions, based on the evidence above:

What research or district data supports whole-school immersion over strand immersion?
If the concern is single strands, why not two-class strands, which PPS already runs at Ainsworth and César Chávez?
Will PPS publish persistence, teacher retention and achievement by program for whole-school and strand sites before deciding?
Lent’s gains are for the students who stayed. How does the plan account for the students a conversion moves out?

Sources: PPS Board of Education, regular meeting, October 6, 2026, staff presentation at 2:25:37 (video; automated transcript via Save PDX Schools); PPS, Enrollment by Grade and Program Type, October 2024 and October 2025 (PPS Data & Reporting); Green, T. L., & Hanson, H. (2026, June 29). Equity audit evaluation of Portland Public Schools’ Southeast Enrollment and Program Balancing initiative: School-by-school analysis (PPS SEGC documents). Method details: Sources & methods.

Limits of this review

“No study found” means none found in our search. The search covered the papers listed above, the National Academies review and the searches described in Sources & methods. A study we missed could exist. Please send it.
We read some sources only in abstract (Morales; Freire and Alemán; de Jong and Bearse). Quotes from them are from the abstracts.
Students per grade (Table 2) is not a count of classes.
The Lent results are from PPS’s equity audit. We have not seen the student-level data behind them.