Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

Are Low-Cost Private Schools Worth the Investment?: Evidence on Literacy and Mathematics Gains in Nairobi Primary Schools


by Stephanie Simmons Zuilkowski, Benjamin Piper & Salome Ong’ele - 2020

Background/Context: Low-cost private schools (LCPSs) represent a large and growing share of schools in many low- and middle-income countries, including Kenya. In some Nairobi neighborhoods, more than half of children attend LCPSs, despite policies providing free access to public education. Parents generally choose LCPSs because they believe they are higher quality, although there is little conclusive evidence supporting this belief.

Objective: In this study, we aim to add to the evidence available on the comparison between LCPSs and public schools by using student gains over time as our outcome, thus controlling for the initial level of achievement and instead examining improvement.

Participants: The randomly selected longitudinal sample was composed of 326 children attending 47 LCPSs and government schools in several of Nairobi’s geographic zones. These children’s literacy and numeracy outcomes were tracked over two academic years to determine their learning gains over time.

Research Design: We used residual gain scores—as opposed to cross-sectional measures—in English and Kiswahili literacy and mathematics, to compare the outcomes of students attending LCPSs and public primary schools across the first and second grades. We discuss how these schools impacted achievement over time.

Findings/Results: We found that these LCPSs did not produce significantly higher student growth than public schools under the status quo condition. However, among schools participating in an instructional improvement intervention supported by the United States Agency for International Development, LCPSs increased performance more than public schools in English, Kiswahili, and mathematics, with the largest differences in English.

Conclusions/Recommendations: These findings offer a cautionary note to the rapid expansion of LCPSs in low-resource settings. The fees paid to LCPSs by low-income households are often burdensome for families and, in some contexts, may not be worth the trade-offs that families make to afford them. On the other hand, the findings also suggest that low-cost private school teachers may respond more effectively than public school teachers to project-based support.



INTRODUCTION


The low-cost private school (LCPS) sector is growing rapidly in low- and middle-income countries (LMICs), particularly in South Asia and sub-Saharan Africa. In some areas, these schools are now the predominant form of primary education. The Partnership Schools for Liberia project turned hundreds of primary schools over to several LCPS chain operators, including Bridge International Academies and BRAC (Rosenberg, 2016), with positive impacts but complicated implementation realities (Romero, Sandefur, & Sandholtz, 2017). Funders including Bill Gates, Mark Zuckerberg, the UK Department for International Development (DFID), and the publisher Pearson have invested large amounts in this model of education (Stevis & Clark, 2015). In countries where enrollment increases over the past several decades have not been followed by significantly better student performance in public schools, LCPSs are often seen as the only high-quality education option available to some families.


In some urban areas in Kenya, more than half of students are now attending LCPSs (Ngware et al., 2013), with rapid growth in the number of schools since the implementation of free primary education in 2003 (Nishimura & Yamano, 2013). Previous work has shown that parents choose LCPSs in Kenya and other LMICs largely because of perceived quality differences: Parents see the overcrowded classrooms and disinterested teachers in public primary schools and choose to send their children elsewhere (Zuilkowski, Piper, Ong’ele, & Kiminza, 2018). But it is unclear whether LCPSs are actually better at producing learning than public schools. Previous studies’ comparisons between learning outcomes in LCPSs and in public schools likely were confounded by the relationships between family background characteristics—measurable and unmeasurable—and school type. Therefore, it is unclear whether investments in LCPSs—generally by parents or large-scale funders—will pay off over time.


With this study, we aimed to add to the evidence available on the comparison between LCPSs and public schools by using student gains over time as our outcome, thus controlling for the initial level of achievement and instead examining improvement. Drawing on data from the 2011–2014 Primary Math and Reading (PRIMR) Initiative in Kenya, we compared student gains in public primary schools to those of LCPSs in two conditions: control and intervention. This approach allowed us to compare the two school types in the status quo control condition as well as in a situation where the instructional approach was changing dramatically. We also examined the background characteristics of the two school types’ teacher corps, which has relevance to the interpretation of the results. We discuss the implications of our findings for public policy and education interventions in urban Kenya and similar contexts.


BACKGROUND AND CONTEXT


The precise definition of an LCPS in a specific context has been the topic of considerable debate (Srivastava, 2013). Srivastava (2013) defined low-cost private schools as being


independently funded through comparatively lower tuition fees (relative to elite or higher-fee private schools), financially sustained through direct payments from poorer or relatively disadvantaged households (though not necessarily the poorest or most disadvantaged), and independently managed and owned by a single owner or team, usually comprising family members. (Srivastava, 2013, pp. 11–12)


Defining LCPSs in Kenya is particularly difficult, as some LCPSs receive government funding to support instructional materials. Since 2011, LCPSs have sometimes been eligible for grants to subsidize the costs of textbooks and other basic instructional materials through the Alternative Provision of Basic Education and Training (APBET) policy, but schools must be registered with the government and comply with requirements, including annual visits from Ministry of Education (MoE) staff to ensure basic standards are met (Edwards Jr., Klees, & Wildish, 2017). In PRIMR, the population of LCPSs was defined as those registered with a Kenyan government agency, charging less than US$12 per month per pupil, and enrolling more than 10 pupils in both Grade 1 and Grade 2. Such schools have been increasingly popular options in Kenya in recent years. In 2018, 1,500 LCPSs were being supported by the Kenyan government’s national literacy program, called the Tusome Early Grade Reading Activity, with funding from USAID.


In some Nairobi neighborhoods, the locations of public schools have not matched population growth. For example, there are only three public primary schools in Kibera, a large informal settlement in Nairobi, to serve an estimated 55,000 primary-age children, with an additional six nearby that also serve the area. The majority of children in Kibera thus attend LCPSs (Tooley & Dixon, 2007). However, the evidence largely supports the hypothesis that parents choose an LCPS for their children not because public school spaces are unavailable, but because they perceive LCPSs to be of higher quality (Dixon & Tooley, 2012; Härmä, 2013; Heyneman & Stern, 2014).


Researchers and parents alike have identified a number of reasons that LCPSs may be of higher quality than public schools. Pupil–teacher ratios are often far lower in LCPSs than in nearby public schools (Mehrotra & Panchamukhi, 2007; Rose & Adelabu, 2007; Stern & Heyneman, 2013; Tooley, Dixon, & Gomathi, 2007). Class size is an important, highly visible marker to parents (Zuilkowski et al., 2018). LCPSs are also perceived as having stricter discipline (Oketch, Mutisya, Ngware, Ezeh, & Epari, 2010; Rose & Adelabu, 2007), which is preferred by many parents (Zuilkowski et al., 2018). Generally, LCPSs have less teacher absenteeism, as school administrators can easily fire their staff, who are not protected by unions or government regulations (Ngware et al., 2013; Tooley et al., 2007).


Although LCPSs may have lower student–teacher ratios and different disciplinary practices, their effects on academic performance are less clear. In their study of LCPSs and government schools in India, Ghana, Nigeria, and Kenya, Tooley and Dixon (2007) found, on average, better student performance in LCPSs compared to government schools, a finding echoed by other studies conducted in Kenya (Alcott & Rose, 2016; Ejakait, Mutisya, Ezeh, Oketch, & Ngware, 2011; Ngware et al., 2013; Piper, King, & Mugenda, 2016; Piper & Mugenda, 2012). However, there is a lack of causal evidence of the effects of LCPS attendance on student performance. Families that choose LCPSs for their children likely differ in a number of ways from families that do not. Some of these differences may be measurable and straightforward to control for, like household income; while others may be difficult to measure, such as the extent to which parents value formal education. A study from India found that LCPS effects were no longer observed when propensity score matching was used to address selection issues (Chudgar & Quin, 2012), a finding that should add a cautionary note to cross-sectional comparisons of outcomes in LCPSs and public schools.


The potential benefits of LCPSs have a corresponding set of drawbacks. Teachers may be less likely to be absent in LCPSs, but they are also less likely to be trained (Srivastava, 2007). For example, one study found that in Bihar, India, just 1% of “private unaided” school teachers had any pre-service training, compared to 68% of government school teachers (Mehrotra & Panchamukhi, 2007, p. 139). Teacher salaries are generally far lower in LCPSs than in government schools (Dixon & Tooley, 2012; Srivastava, 2007; Stern & Heyneman, 2013; Tooley & Dixon, 2007); one study found a salary ratio of three to one in Nairobi (Tooley & Dixon, 2007). Low salaries lead to teacher turnover (Srivastava, 2007), which has implications for student–teacher relationships and school climate, as well as loss of instructional time during transitions.


Some scholars have used a human rights framework to critique LCPSs, noting that the right to education is enshrined in national and international law (Aubry & Dorsi, 2016). Aubry and Dorsi (2016) concluded that “Options where private fee-charging schools become, or threaten to become, the only options available for some people, are therefore clearly in violation of human rights law” (p. 620). The privatization of education leads to serious questions about equity. The most disadvantaged children are still excluded from LCPSs due to inability to pay (Härmä, 2009, 2016), although their parents often express preferences for private schooling generally (Härmä, 2011). LCPSs can deepen inequality not only by income level but also by geography or gender (Alderman, Kim, & Orazem, 2003; Andrabi, Das, & Khwaja, 2002, 2008; Härmä, 2013). In placing the responsibility for identifying the best available school for children on their parents, who may have little formal education themselves, this system can leave the poorest even further behind. The information that parents have to use in making schooling decisions for their children is often incomplete (Heyneman & Stern, 2014) and based on input factors that are clearly visible—such as class size, the availability of instructional materials, and end-of-primary examination scores—as well as anecdotal information from community members.


Taken as a whole, this literature indicates that the appropriate role of LCPSs in low- and middle-income countries remains unresolved. One of the major arguments supporting their expansion is that they are more efficient than government school systems (Srivastava, 2010). However, as Edwards, Klees, & Wildish  (2017) stated, LCPSs do not necessarily operate more efficiently, from a large-scale perspective, in terms of moving students through the system in a low-cost manner. Instead, small-scale operators are concerned with their school’s profitability, which may or may not align with efficiency at scale. We still do not know whether the higher learning outcomes for students who attend LCPSs are due to the higher quality of the schools or to family background or other selection effects.


RESEARCH QUESTIONS


The aim of this study is to examine how school types—LCPSs and public primary schools—compare in their ability to improve student outcomes over time, in both control and intervention conditions. Specifically, we address the following research questions in this analysis:


RQ1: Did children attending LCPSs improve more over time, on average, than public school students on outcomes in mathematics and in English and Kiswahili literacy?

RQ2: Did children attending LCPSs participating in PRIMR gain more over time, on average, than children attending public schools participating in PRIMR?

RQ3: How did LCPS teachers differ from public school teachers?


RESEARCH DESIGN


OVERALL DESIGN OF PRIMR


As noted, the data used in the analyses presented here were drawn from the Primary Mathematics and Reading Initiative, a medium-scale intervention funded by USAID and DFID.1 The project included a series of research subprojects, investigating topics such as the use of mother-tongue instruction (Piper, Zuilkowski, & Ong’ele, 2016), the costs and benefits of instructional approaches that used information and communication technology (ICT) (Piper, Zuilkowski, Kwayumba, & Strigel, 2016), and technical issues surrounding the measurement of early grade reading (Piper & Zuilkowski, 2015, 2016).


The PRIMR intervention is described in detail elsewhere (Piper, King, & Mugenda, 2016; Piper, Zuilkowski, & Mugenda, 2014), and is outlined briefly here. The goal of the project was to improve instructional practices in reading and mathematics classes in the early grades, using research-based approaches. In the portion of PRIMR reviewed here, the intervention covered the school subjects of English, Kiswahili, and mathematics. The project consisted of four main activities. First, project staff, in collaboration with the MoE, developed teachers’ guides for the three subjects. These structured teachers’ guides scaffolded teachers, in a systemic way, in moving away from whole-class, lecture-based approaches to more student-centered instruction. Second, the program designed and produced books that aligned with the teachers’ guides and provided structured literacy skill practice. The books were longer than the standard government textbooks and were distributed at a 1:1 student-to-book ratio, an improvement over the typical 3:1 ratio. Third, all first- and second-grade teachers in intervention schools received 10 days of training per academic year on the research behind the intervention, how to use the teachers’ guides and student books, and how to utilize basic instructional materials such as pocket charts and flashcards. Fourth, instructional coaches visited classrooms to support participating teachers in implementing PRIMR. This coaching role was filled by MoE Curriculum Support Officers (CSOs), who at the time were called Teachers’ Advisory Centre tutors, and who supported the public schools; and program-hired instructional coaches who supported the LCPSs. The CSOs and coaches received 15 days of training in the first year of the project. Given a delay in the design of the mathematics materials and training, the mathematics intervention had approximately four months less instructional intervention time than the English and Kiswahili intervention (Piper, Ralaingita, Akach, & King, 2016).


The PRIMR Initiative was designed to estimate the impact of the intervention on public schools in several counties, as well as LCPSs in Nairobi. For public schools, all zones (groups of 12 to 20 schools) were randomly assigned into treatment and control groups. For LCPSs, prior to random assignment, the PRIMR technical team worked with the Ministry of Education to collect data on the more than 1,000 LCPSs in Nairobi. The team investigated each LCPS’s interest in participating in the PRIMR study, as well as the fees charged to children, the registration status of each LCPS with the Kenyan government, and the enrollment numbers by grade. After determining eligibility according to the MoE’s minimum standards for registration as a LCPS, called an APBET institution (Ministry of Education, Science and Technology, 2015), the PRIMR technical team grouped the LCPSs into geographic clusters of 10 and 15 schools. These clusters were then randomly assigned to the treatment or control groups.


SITE


The USAID-funded portion of PRIMR was implemented in 547 public primary schools and LCPSs in four Kenyan counties between 2011 and 2014. The data presented here were drawn from the 12 zones (for public schools) and coach clusters (for LCPSs) in Nairobi County (see Figure 1). Nairobi was chosen because it was the only county that had both LCPSs and public schools available in the sample. Figure 1 shows the proximity of the randomly selected clusters and zones within Nairobi. The majority of Nairobi’s parents had both public schools and LCPSs available relatively near their homes (Piper et al., 2014).


The LCPS sites were quite diverse. Just over half were schools owned by an individual, with smaller percentages run by community-based organizations (30%), religious groups (9%), or for-profit companies (9%). Across the ownership categories, more than 70% of the LCPSs were nonprofit schools. Forty-seven percent were registered with the Ministry of Gender and Social Services and 19% with the MoE.



Figure 1. Map of selected public-school zones and LCPS clusters in Nairobi

[39_22953.htm_g/00002.jpg]

Note. LCPS = low-cost private school.



SAMPLE


The PRIMR longitudinal data subsample consisted of 326 children living in Nairobi who were in first grade at baseline. The subsample was drawn in January 2012 using simple random sampling, stratified by sex, within the first-grade classrooms of sampled schools. The longitudinal subsample of children was followed from January 2012 at the beginning of Grade 1 and reassessed in October 2012 at the end of Grade 1 and October 2013 at the end of Grade 2. The overall longitudinal data set had a 37.2% attrition rate, higher than desirable but lower than the 41% attrition rate from a recent survey (Alderman, Behrman, Kohler, Maluccio, & Watkins, 2001). The analysis of the Kenyan household survey results showed few differences in demographics, even with the attrition rates (Alderman et al., 2001). The larger PRIMR impact evaluation research compared the results from a model that used the gains from a difference-in-differences analysis from a repeated cross-sectional sample with another model that used the same longitudinal subsample of children that we utilized in this study. The results from these two models were similar in their direction and magnitude across a wide range of regression results (Piper, King, & Mugenda, 2016).


The children attended 47 schools across the 12 zones and clusters in Nairobi; 166 children (51%) were male and 160 (49%) female. The modal age among the children at baseline was 6 years (mean 6.2 years), with a range from 4 to 10 years. In terms of type of school, 66% attended LCPSs and 34% attended public schools. Regarding program engagement, 182 (56%) were attending PRIMR-participant schools at baseline, either LCPSs or public schools. Nearly all (93%) had attended some type of preschool program before starting Grade 1. PRIMR also assessed the English, Kiswahili, and mathematics teachers of the students in the sample. Our analysis is based on data from 82 teachers teaching in the LCPSs and public schools in Nairobi at the time of the October 2013 final assessment.


MEASURES


To assess children’s literacy and mathematics skills, PRIMR used versions of the one-on-one Early Grade Reading Assessment (EGRA) and the Early Grade Mathematics Assessment (EGMA) that had been developed in Kenya during the early stages of the PRIMR project (Piper & Mugenda, 2012). Table 1 briefly describes the instruments and the skills that were assessed, as well as the Cronbach’s alpha reliability scores for each of the instruments (Piper, King, & Mugenda, 2016).



Table 1. Instruments and measure descriptions for English, Kiswahili, and mathematics

Instrument

Subcomponent

Description

Timed/

untimed

Measure

Cronbach’s alpha reliability

English Early Grade Reading Assessment (EGRA)

 

 

 

Letter fluency

Letters read correctly per minute

Timed

(1 minute)

Correct letters per minute

0.85

Nonword fluency

Nonsense words read correctly per minute

Timed

(1 minute)

Correct nonwords per minute

0.82

Oral reading fluency

Connected text words read correctly per minute

Timed

(1 minute)

Correct words per minute

0.81

Reading comprehension

Percentage of comprehension questions correct

Untimed

Percentage correct

0.83

Kiswahili EGRA

Letter fluency

Letters read correctly per minute

Timed

(1 minute)

Correct letters per minute

0.89

 

Nonword fluency

Nonsense words read correctly per minute

Timed

(1 minute)

Correct nonwords per minute

0.86

 

Oral reading fluency

Connected-text words read correctly per minute

Timed

(1 minute)

Correct words per minute

0.86

 

Reading comprehension

Percentage of comprehension questions correct

Untimed

Percentage correct

0.87

Early Grade Mathematics Assessment (EGMA)

Number identification

Numbers correctly identified

Timed

Correct numbers per minute

0.87

Missing number

Determining what is the missing number from a pattern

Untimed

Percentage correct

0.86

Addition fluency

Addition problems answered correctly per minute

Timed
(1 minute)

Correct addition problems per minute

0.86

Subtraction fluency

Subtraction problems answered correctly per minute

Timed
(1 minute)

Correct subtraction problems per minute

0.86




Additionally, we collected data on gender and on family socioeconomic status as control variables. Socioeconomic status was measured by a household possessions index self-reported by the students. This index included having a range of household items such as a radio, a telephone, electricity, a television, a refrigerator, a toilet inside the house, a bicycle, a motorcycle, or a vehicle.


PROCEDURES


As discussed above, the participants were assessed at three time points—January 2012, October 2012, and October 2013. The assessments were conducted by an experienced team of Kenyan research assistants, who had been involved with reading-related research projects for several years. All assessors received 5 days of training before fieldwork began. The assessors used tablets with Tangerine® open-source survey software to collect data in October 2012 and October 2013, after having used paper-based assessments in January 2012. Assessor supervisors collected the teacher interview data at each school in a one-on-one oral interview, with data entered into the tablets via Tangerine in October 2013.


ANALYTICAL APPROACH


To evaluate the relative performance of public schools and LCPSs in two conditions—control and intervention—we used residual gain scores for students between January 2012 and October 2013. Using the residual gain scores allowed us to control for minor differences in baseline scores for students in LCPSs and public schools (see Table 2 and Appendix A) and for the differences in baseline scores for students in treatment and control schools, which we have described in depth elsewhere (Zuilkowski et al., 2018). Finally, given that we used residual gain scores, we have focused on the interactions between school type and treatment status and compared changes over two academic years. We therefore are comparing not overall outcomes but increases in student performance associated with each school type. Residual gain scores are more stable than simple gain scores, making them useful for comparison. In addition to allowing us to adjust for the baseline differences, gains in reading fluency are useful because they have been shown to predict outcomes over the course of the first-grade year in the US (Smith et al., 2014). In fitting our models, we adjusted for clustering by zone using svy analysis methods in Stata version 15. The svy suite of commands employs the weights produced by the complex sampling procedures utilized in the PRIMR study. As the PRIMR intervention worked through the government’s CSOs, who were assigned by zone, this was the appropriate clustering level. LCPSs were grouped into geographic clusters of approximately the same number of schools as in the public-school zones. All models also controlled for gender and family socioeconomic status, measured by a household possessions index.


To answer our two research questions, we divided schools into four groups for comparison: control public, control LCPS, PRIMR public, and PRIMR LCPS. The regression models we present below allowed us to make the comparisons between the public control and public LCPS schools and the PRIMR control and control LCPS schools in a single model, using an interaction variable between PRIMR and LCPS to estimate the additional gains that PRIMR children also in LCPS schools experienced. Given that the basic analytic method in the regression was a difference-in-differences model, using the interaction term allowed us to isolate the gains attributed to children in treatment schools as well as LCPSs, and compare them to children who were in treatment public schools and those in control LCPSs.



Table 2. Baseline comparison of average pupil performance scores, public school and LCPS samples, in English, Kiswahili, and mathematics

Subject

Assessment

Public school

(= 112)

LCPS

(= 214)

T

p

English

Letter fluency

24.8

26.7

–0.95

.34

 

Nonword fluency

7.5

9.0

–1.30

.19

 

Oral reading fluency

5.6

7.8

–1.39

.16

 

Reading comprehension

0.03

0.04

–0.49

.63

Kiswahili

Letter fluency

16.0

18.0

–1.85

.06

 

Nonword fluency

3.4

4.2

–0.91

.37

 

Oral reading fluency

4.0

4.7

–0.70

.49

 

Reading comprehension

0.04

0.07

–1.89

.06

Mathematics

Number identification

12.0

13.6

–2.31

.02

 

Missing number

0.17

0.19

–1.21

.23

 

Addition fluency

3.3

3.6

–0.60

.55

 

Subtraction fluency

1.6

1.7

–0.32

.75

Note. LCPS = low-cost private school.




Therefore, the general format of the models was as follows:


[39_22953.htm_g/00004.jpg]

where Yiz refers to the residual gain score for an outcome for student i in zone z; LCPS and PRIMR are dichotomous variables indicating, respectively, whether a student was enrolled in an LCPS or a PRIMR school; LCPSxPRIMR is an interaction term; Z is a vector of background characteristics for student i that included student gender, socioeconomic status, and a dichotomous variable indicating the child had access to the appropriate book in English, Kiswahili or mathematics; and ɛ is an error term, with clustering at the zone level.


We used ordinary least squares (OLS) regression to analyze the relationship between school type and the teacher characteristics for research question 3.


FINDINGS


RQ1: Did children attending LCPSs improve more over time, on average, than public school students on outcomes in mathematics and in English and Kiswahili literacy?


To address our first research question, we fit a series of regression models examining the relationships between school category—control public schools and control LCPSs—and 12 academic outcomes in mathematics, Kiswahili, and English. The models also controlled for gender, age, socioeconomic status, and whether the child had access to the appropriate pupil textbook for the subject being taught. Overall, while the mean gains were on average larger for students attending LCPSs, the differences were generally not statistically significant. Significance can be determined by examining whether the parameter estimates associated with the control LCPS variable are statistically significant in each model in Tables 3, 4, and 5. For example, we found that the difference between the control public and control LCPS schools in English letter sound fluency was –0.33 correct letters per minute, which is not a statistically significant difference (see Table 3). Exceptions were the English oral reading fluency gains shown in Table 3, where children in LCPSs gained 6.25 more correct words per minute (p-value <.10); and the missing number score in Table 5, where children in control LCPS schools gained 9 percentage points more than those in control public schools (p < .01). There were no statistically significant differences in the average gain for any of the Kiswahili outcomes (Table 4). In the absence of an intervention, therefore, we found that the LCPSs did not increase students’ scores significantly more than public primary schools.



Table 3. English literacy regression results indicating gains in student outcomes

Variable

Correct letters per minute

Correct nonwords per minute

Correct words per minute (ORF)

Reading comprehension (% gain)

Constant (Control public)

–7.54

3.94

17.27

0.27

 

(17.32)

(6.81)

(14.88)

(0.15)

Control LCPS

–0.33

1.62

6.25~

0.01

 

(0.96)

(1.27)

(3.31)

(0.07)

PRIMR public

17.81***

1.15

–1.63

0.02

 

(1.04)

(1.34)

(4.17)

(0.01)

PRIMR LCPS

23.70*

10.70**

23.84*

0.25*

 

(8.66)

(2.84)

(7.84)

(0.11)

Female

–0.06

4.06*

4.36

0.01

 

(1.76)

(1.59)

(2.51)

(0.04)

Age

1.37

0.97

1.35

–0.02

 

(2.78)

(0.98)

(1.88)

(0.02)

Socioeconomic status

1.42

1.69**

2.71*

0.03~

 

(0.96)

(0.43)

(1.07)

(0.01)

English pupil book

3.61

0.78

–2.72

0.06

 

(3.99)

(2.36)

(3.16)

(0.04)

R2

.28

.16

.21

.13

Note. ~ p < .10, * p < .05, ** p < .01, *** p < .001. Linearized standard errors in parentheses. LCPS = low-cost private school; ORF = oral reading fluency; PRIMR = Primary Math and Reading Initiative.




Table 4. Kiswahili literacy regression results indicating gains in student outcomes

Variable

Correct letters per minute

Correct nonwords per minute

Correct words per minute (ORF)

Reading comprehension (% gain)

Constant (Control public)

8.39

7.43

12.44

0.27

 

(15.89)

(7.50)

(13.88)

(0.16)

Control LCPS

–3.23

4.015

5.61

–0.04

 

(2.72)

(3.09)

(3.29)

(0.04)

PRIMR public

12.59*

1.14

0.44

–0.06

 

(4.44)

(2.92)

(3.93)

(0.05)

PRIMR LCPS

17.94~

4.81

11.81~

0.27**

 

(9.72)

(4.74)

(6.52)

(0.08)

Female

3.39

1.41

2.21

0.02

 

(2.37)

(1.19)

(2.07)

(0.02)

Age

1.75

0.80

1.28

0.01

 

(2.43)

(0.97)

(1.80)

(0.02)

Socioeconomic status

1.23

0.95

1.49~

0.01

 

(1.29)

(0.61)

(0.81)

(0.01)

Kiswahili pupil book

–0.41

–0.40

–1.84

–0.06

 

(3.22)

(1.46)

(2.42)

(0.05)

R2

.17

.07

0.14

.12

Note. ~ p < .10, * p < .05. ** p < .01, *** p < .001. Linearized standard errors in parentheses. LCPS = low-cost private school; ORF = oral reading fluency; PRIMR = Primary Math and Reading Initiative.




Table 5. Mathematics regression results indicating gains in student outcomes

Variable

Numbers identified per minute

Missing number score (% gain)

Addition fluency

Subtraction fluency

Constant (Control public)

11.61

0.04

5.61*

3.59*

 

(4.99)

(0.13)

(2.45)

(1.55)

Control LCPS

0.35

0.09**

0.95

0.87

 

(1.53)

(0.03)

(0.63)

(0.52)

PRIMR public

–0.92

0.07***

0.52

0.26

 

(1.39)

(0.01)

(0.80)

(0.19)

PRIMR LCPS

9.07**

0.05

2.03~

0.94

 

(2.80)

(0.04)

(1.12)

(0.93)

Female

1.44

–0.01

0.76

0.82

 

(0.89)

(0.02)

(0.44)

(0.57)

Age

–0.29

0.03

–0.05

0.23

 

(0.73)

(0.02)

(0.30)

(0.22)

Socioeconomic status

0.79~

0.00

0.23~

0.23~

 

(0.43)

(0.01)

(0.13)

(0.13)

Mathematics pupil book

–1.02

–0.04

–0.14

–1.11*

 

(1.14)

(0.03)

(0.62)

(0.38)

R2

.13

.12

.09

.04

Note. ~ p < .10, * p < .05. ** p < .01, *** p < .001. Linearized standard errors in parentheses. LCPS = low-cost private school; PRIMR = Primary Math and Reading Initiative.




RQ2: Did children attending LCPSs participating in PRIMR gain more over time, on average, than children attending public schools participating in PRIMR?


The comparison of LCPSs and public schools participating in a high-quality, structured literacy and mathematics intervention produced different patterns than in the control condition examined above. Turning first to the English literacy outcomes (see Table 3), we found that PRIMR LCPS students gained statistically significantly more than PRIMR public school students on all four English literacy outcomes, from basic (letter knowledge) to advanced (reading comprehension). Many of these differences were quite large in magnitude. In PRIMR LCPSs, the average student gained 23.7 more correct letters per minute (p value < .05), 10.7 more correct nonwords per minute (p value < .01), 23.84 more correct words per minute on the oral reading fluency task (p value < .05), and 25 percentage points more on reading comprehension (p value < .05). Each of these differences between the PRIMR public and PRIMR LCPS outcomes was statistically significant as well as substantively meaningful, with the average child in PRIMR LCPS closing most of the distance to being identified as a reader, according to the government’s English fluency benchmark of 65 words per minute.


The differences between LCPSs and public schools were less stark for Kiswahili literacy outcomes (Table 4), although PRIMR LCPS students gained more than PRIMR public school students. The difference in gains was statistically significant for Kiswahili reading comprehension, as children in PRIMR LCPS schools gained 27 percentage points more than children in PRIMR public schools (p value < .01). There were outcome gains for PRIMR LCPS schools on Kiswahili letter sound fluency, at 17.94 letters more per minute than PRIMR public schools (p value < .10); and for Kiswahili oral reading fluency by 11.81 correct words per minute more than in PRIMR public schools (p value < .10). These differences are quite large, if not statistically significant at the .05 level. In mathematics (Table 5), PRIMR LCPSs statistically significantly outperformed PRIMR public schools by 9.07 percentage points on the number identification task (p value < .01) as well as the addition fluency task, where children in PRIMR LCPSs outperformed PRIMR public schools by 2.03 addition problems per minute at the .10 significance level.



RQ3: How do LCPS teachers differ from public school teachers?


Given the gains observed at LCPSs that participated in the PRIMR program, as described above, we next examined the characteristics of teachers in the two school types to determine whether statistically significant differences might exist that could explain the differential performance. We present the results of simple OLS regression models in Table 6. The results showed that 100% of the public-school teachers and 92.9% of the LCPS teachers were female, which was not statistically significantly different (p = .17). The two groups of teachers attended similar numbers of PRIMR training sessions in the intervention year on average, at 3.6 for public school teachers and 3.4 for LCPS teachers (p = .33). The average age for public school teachers was 47.6 years old, with LCPS teachers significantly younger at 29.3 years old on average (< .001). Accordingly, public teachers had many more years of experience (21.8) than did LCPS teachers (5.7) (p < .001).


 

Table 6. Characteristics of public and LCPS teachers in Nairobi PRIMR treatment schools

Measure

n

Public

LCPS

Teacher is female

82

100

(4.2)

92.9

(5.1)

Number of PRIMR training sessions attended

79

3.6

(0.2)

3.4

(0.2)

Teacher age

82

47.6***

(1.2)

29.3

(1.5)

Years of experience

82

21.8***

(1.1)

5.7

(1.3)

Note. ~ p < .10, * p < .05. ** p < .01, *** p < .001. Standard errors in parentheses. LCPS = low-cost private school; PRIMR = Primary Math and Reading Initiative.




We also compared the highest qualification levels of teachers in the two school types. Figure 2 presents the results. The public teachers primarily were holders of P1 certificates (26.9%), diplomas/S1 certificates (diploma in education) (42.3%), and Bachelor’s of Education (BEd) degrees (23.1%). This shows that the Nairobi teachers surveyed on this item were significantly better trained than the majority of Kenya’s teachers, who are far less likely to have a BEd (Japan International Cooperation Agency, 2012). On the other hand, 26.8% of LCPS teachers had only a secondary school education or had a secondary education but were untrained as teachers, and 23.2% had an Early Childhood Development and Education (ECD) certificate only, leaving 50% of teachers who had either a P1 or diploma/S1 certificate. The results show that the LCPS teachers were far less well trained on average than the public-school teachers.



Figure 2. Highest qualification levels of PRIMR teachers in public schools and LCPSs

[39_22953.htm_g/00006.jpg]


Note. ECD = Early Childhood Diploma; LCPS = low-cost private school; P1 =  Certificate in Education (Primary); PRIMR = Primary Math and Reading Initiative; S1 = Certificate in Education (Secondary)


 


DISCUSSION


The debate over LCPSs in Kenya, and in LMICs more broadly, has been drawing toward the consensus that students in LCPSs generally perform better than children who attend public schools, and that parents perceive LCPSs to be of higher quality. However, comparisons between students attending the two types of schools that assume the two groups are equal in expectation are often questionable, even with controls for observable background characteristics. Research in the US has shown that individual charter schools often have populations that are very different from those of the nearest public school, even when population-level averages appear to be similar (Malkus, 2016). To our knowledge, to date results are available from just one study in sub-Saharan Africa that randomly assigned students to LCPSs or public schools.2 Conducted in the context of the Partnership Schools for Liberia program, the study found positive results on assessments in reading and mathematics, but also noted that the private schools forced weak students and teachers back into the public schools (Romero, Sandefur, & Sandholtz, 2017). As pointed out above, a study in India found that the apparent “effects” of LCPSs disappeared when a propensity score approach was used to better account for selection bias (Chudgar & Quin, 2012). It may be that the apparent benefit of LCPSs is driven by selection bias, and our aim was to advance the discussion by examining student-level residual learning gains rather than basic levels of achievement, which may be linked to unobserved factors such as parental values regarding education.


We found that in Nairobi, control LCPSs did not produce statistically significantly larger residual gains in student performance than control public schools. The answer to the question we posed in the title to this paper—are LCPSs worth the investment by parents, donors, and governments—therefore seems to be “maybe not.” In the absence of effective interventions focused on improving learning, such as PRIMR, there appear to be no additional impacts on learning if a child is sent to an LCPS, despite parental perceptions of higher quality in those schools (Zuilkowski et al., 2018). Our analytic approach, using residual student gains over time rather than cross-sectional scores, led us to overall findings that contrasted with much of the recent literature on LCPSs in Kenya, which has suggested that LCPSs are intrinsically higher-quality schools (Alcott & Rose, 2016; Ejakait et al., 2011) rather than schools that have higher outcomes because of selection bias. It also contrasts with the recent push in England to create a set of LCPSs (Tooley, 2018). Parental background, education, wealth, and willingness to invest in education for children may be the primary factors driving the higher LCPS outcomes in Kenya, at least in primary education. This finding is similar to much of the literature in the United States and UK and suggests that educational outcome differences are in no small part due to “sorting” decisions that families make in the educational sphere (Putnam, 2016).


In contrast to the discussion above, we found that LCPSs improved students’ scores more than public schools in the context of the PRIMR intervention focused on improving reading and mathematics outcomes. In some cases, the differences in the gains were quite large, particularly for the English literacy outcomes. This could have been due to the prestige and importance of English—national examinations are in English, and students’ success on these examinations is the best advertisement for the services offered at LCPSs. The differences for Kiswahili and mathematics outcomes were smaller and mixed in terms of statistical significance, but none of the results favored public schools.


This leads to the question of why the LCPSs were so much more responsive to the intervention than were the public schools. One possible explanation is that pre-service teacher education has a strong impact on teachers’ pedagogy and educational philosophy, a hypothesis that is backed by research on teacher education elsewhere in sub-Saharan Africa (Akyeampong, Lussier, Pryor, & Westbrook, 2013). Given that public schools in Nairobi are more likely to have trained teachers than are LCPSs (Ngware, Oketch, & Ezeh, 2011), their teaching corps may be less open to learning new approaches. In PRIMR, untrained teachers in LCPSs were eager to receive training and assistance in improving their methods. In addition, given that LCPS results are related to profits for LCPS proprietors, inclusion in PRIMR might have been seen as a mechanism for improved quality, which could increase enrollment and potential profits. It may be that LCPSs are able to capitalize on even modest forms of external support, and benefits to students can ensue.


While we found that LCPSs did not statistically significantly outperform public schools in the control condition, we note that in no case—across all of the 12 outcome models presented in this analysis—did public school students gain statistically significantly more than LCPS students. This should be of concern to policy makers, as public schools spend far more per student than do LCPSs. On the other hand, although we may categorize LCPSs as “low cost,” tuition and fees can be burdensome to the poor families that they serve. In a related study, parents whose children attended LCPSs reported spending 7,395 Kenyan shillings (approximately US$71) per child per year—a significant percentage of the average household income (Zuilkowski et al., 2018). It is important to note that in many countries with official free public education policies, schooling is not actually free to households (Edwards Jr. et al., 2017; Heyneman & Stern, 2014; Ohba, 2013). In fact, parents of public-school children reported spending an average of 3,244 Kenyan shillings (approximately US$31) annually on tuition, extra fees, and school meals (Zuilkowski et al., 2018).


While the overall effects of the PRIMR intervention compared to control schools have been reported elsewhere (Piper, King, & Mugenda, 2016), it is important to note that there were interactions between the PRIMR intervention and school type. For all 12 outcomes reported in Tables 3 through 5, LCPS students in PRIMR treatment schools gained more on average than public school students in control schools. PRIMR public school students performed statistically significantly better than control public school students on only three outcomes (English and Kiswahili letters per minute and missing number score). Thus, in Nairobi, the overall impacts of the PRIMR project were driven largely by the impacts observed in LCPSs.


Our results showed that teachers in the LCPS setting were significantly different from teachers in public schools in Nairobi in several key ways. Nairobi public school teachers were on average 18 years older, and nearly all of that time had been spent teaching, as the public-school teachers had 16 years more teaching experience than LCPS teachers. They were far more educated as well, with 92% having at least the basic qualifications, compared with only 50% of LCPS teachers. Understanding why these less well trained, younger, and less experienced teachers were more likely to respond to the PRIMR intervention effectively than were the public-school teachers is beyond the scope of our research design. However, it may be that the teachers in LCPSs without formal qualifications were simply more open to a new intervention than were the public school teachers with their experience and formal training, potentially because of the perceived benefit to their educational career, or to the expectations of the LCPS for improved learning as a pathway to increased profits. The starkness of the contrast leads to questioning of the current teacher career ladder in Kenya, which rewards additional training courses and years of experience with increased salary. If those characteristics are actually commensurate with worse learning outcomes, then it is worth reconsidering the mechanisms by which formal pre-service training and the career ladder evaluate and reward teachers with resources.


Governments have a responsibility to ensure that basic standards of quality are met, even as LCPSs are serving increasing proportions of children in many low- and middle-income countries. It is possible that enforcing such standards—such as having at least one qualified teacher in each school or ensuring adequate facilities—might increase costs, with implications for the families who choose to send their children to these schools. However, it is also possible that some of these costs could be offset through government subsidies—for example, payments made to LCPSs on a per-student basis for materials, as has been done previously in Kenya. Governments could also invest in providing professional development for LCPS teachers. If schools operating in areas like Nairobi’s informal settlements raised their tuition significantly as a way to pay for teacher training or other investments in quality, they would likely be forced to close, due to the sensitivity of poor parents to cost, and the competition for students in these parts of Nairobi is intense.


We acknowledge several threats to validity in this study. First, we had access to a relatively small longitudinal sample of 326 Grade 1 students in Nairobi, and the longitudinal data set suffered from attrition, as described above. We also had only limited background information on the students, and given that the students provided the data when they were quite young, we hesitated to include more control variables that might increase the precision of the analysis, but not be entirely reliable. While we believe our focus on residual learning gains from the longitudinal data set, as opposed to repeated cross-sectional comparisons, is an improvement over past studies, we cannot discount the possibility that there may have been an interaction between an unmeasured background variable related to school type selection and gains over time. Future studies designed to collect more parent and household data could examine not only gains, but also differences in gains according to factors associated with school type. For example, parents who believe more strongly in the value of education may be more likely to enroll their children in LCPSs and may also make investments in their children—helping them with homework or buying additional books—that result in greater gains.


A final threat to validity is that, while the PRIMR identification strategy allowed for a causal estimate of the impact of PRIMR, LCPS or public-school attendance was not able to be randomly assigned to students. Our estimates are therefore not a causal estimate of the impact of LCPSs, but the availability of residual gain scores from the PRIMR longitudinal data set is nevertheless a contribution to our ability to estimate the association between LCPS attendance and learning, in both treatment and control conditions.


CONCLUSION


Our findings emanate from Nairobi, a large city with a high percentage of children who have LCPSs available in their geographic area. While generalizations should be drawn cautiously, as settings differ, our findings support the hypothesis that some of the observed effects of LCPSs in Kenya may in fact have been due to selection effects rather than higher-quality instruction, similar to what was documented in India (Chudgar & Quin, 2012). This should give pause to governments and funders that are funneling large sums of money into LCPSs in LMICs . The outcomes that have been documented for LCPS students in Kenya and elsewhere may depend more on the measurable or unmeasurable characteristics of their families that result in sorting than to the higher quality of education provided at LCPSs. More rigorous research allowing for stronger causal inference is urgently needed on this topic.


On the other hand, our analyses showed that LCPSs were much more responsive to instructional interventions than public schools were. Given the rapid spread of LCPSs in Kenya and in other LMICs, there may be meaningful potential benefits in providing modest resources to improve the instructional support available to LCPS teachers, many of whom have little formal training. With even small investments, LCPSs may produce significantly larger gains than public schools, which spend far more money per student. Finally, this study suggests that there may be lessons for the public-school system in how LCPSs and their teachers respond to new instructional programs, and consideration should be given to a reform of the teacher career ladder and the incentives to improved teaching.



Acknowledgment


The Primary Math and Reading (PRIMR) Initiative was funded by the United States Agency for International Development (USAID/Kenya), under the Education Data for Decision Making (EdData II) Blanket Purchase Agreement, Task Order No. AID-623-M-11-00001 (RTI International Task 13). One of the authors served as the Chief of Party on the PRIMR Initiative and another served as the Deputy Chief of Party. The contents of the article are the sole responsibility of the authors. The authors’ views expressed in this publication do not necessarily reflect the views of USAID, the United States Government, the Kenyan Ministry of Education, or RTI International.


Notes


1. The second author of this article was the Chief of Party for PRIMR and the third author was Deputy Chief of Party. The first author had no affiliation with the program.


2. A World Bank-funded evaluation of Bridge International Academies in Kenya was in process in 2018. This evaluation will utilize a lottery system to offer randomly selected students enrollment into Bridge schools.


References


Akyeampong, K., Lussier, K., Pryor, J., & Westbrook, J. (2013). Improving teaching and learning of basic maths and reading in Africa: Does teacher preparation count? International Journal of Educational Development, 33, 272–282. https://doi.org/10.1016/j.ijedudev.2012.09.006


Alcott, B., & Rose, P. (2016). Does private schooling narrow wealth inequalities in learning outcomes? Evidence from East Africa. Oxford Review of Education, 42(5), 495–510. https://doi.org/10.1080/03054985.2016.1215611


Alderman, H., Behrman, J., Kohler, H.-P., Maluccio, J. A., & Watkins, S. (2001). Attrition in longitudinal household survey data: Some tests from three developing countries. Demographic Research, 5(4), 79–124. https://doi.org/10.4054/DemRes.2001.5.4


Alderman, H., Kim, J., & Orazem, P. F. (2003). Design, evaluation, and sustainability of private schools for the poor: The Pakistan urban and rural fellowship school experiments. Economics of Education Review, 22(3), 265–274. https://doi.org/10.1016/S0272-7757(02)00051-1


Andrabi, T., Das, J., & Khwaja, I. (2002). The rise of private schooling in Pakistan: Catering to the urban elite or educating the rural poor? Cambridge, MA: Harvard University. Retrieved from https://www.innovations.harvard.edu/sites/default/files/2608043.pdf


Andrabi, T., Das, J., & Khwaja, A. I. (2008). A dime a day: The possibilities and limits of private schooling in Pakistan. Comparative Education Review, 52(3), 329–355. https://doi.org/10.2307/30218827


Aubry, S., & Dorsi, D. (2016). Towards a human rights framework to advance the debate on the role of private actors in education. Oxford Review of Education, 42(5), 612–628. https://doi.org/10.1080/03054985.2016.1224301


Chudgar, A., & Quin, E. (2012). Relationship between private schooling and achievement: Results from rural and urban India. Economics of Education Review, 31(4), 376–390. https://doi.org/10.1016/j.econedurev.2011.12.003


Dixon, P., & Tooley, J. (2012). A case study of private schools in Kibera: An update. Educational Management Administration & Leadership, 40(6), 690–706. https://doi.org/10.1177/1741143212456908


Edwards Jr., D. B., Klees, S. J., & Wildish, J. L. (2017). Dynamics of low-fee private schools in Kenya: Governmental legitimation, school-community dependence, and resource uncertainty. Teachers College Record, 119(7), 1–42.


Ejakait, E., Mutisya, M., Ezeh, A., Oketch, M., & Ngware, M. (2011). Factors associated with low achievement among students from Nairobi’s urban informal neighborhoods. Urban Education, 46(5), 1056–1077. https://doi.org/10.1177/0042085911400323


Härmä, J. (2009). Can choice promote Education for All? Evidence from growth in private primary schooling in India. Compare: A Journal of Comparative and International Education, 39(2), 151–165. https://doi.org/10.1080/03057920902750400


Härmä, J. (2011). Low cost private schooling in India: Is it pro poor and equitable? International Journal of Educational Development, 31, 350–356. https://doi.org/10.1016/j.ijedudev.2011.01.003


Härmä, J. (2013). Access or quality? Why do families living in slums choose low-cost private schools in Lagos, Nigeria? Oxford Review of Education, 39(4), 548–566. https://doi.org/10.1080/03054985.2013.825984


Härmä, J. (2016). School choice in rural Nigeria? The limits of low-fee private schooling in Kwara State. Comparative Education, 52(2), 246–266. https://doi.org/10.1080/03050068.2016.1142737


Heyneman, S. P., & Stern, J. M. B. (2014). Low cost private schools for the poor: What public policy is appropriate? International Journal of Educational Development, 35, 3–15. https://doi.org/10.1016/j.ijedudev.2013.01.002


Japan International Cooperation Agency. (2012). Basic education sector analysis report Kenya. Retrieved from http://open_jicareport.jica.go.jp/pdf/12083176.pdf


Malkus, N. (2016). Seeing charters differently: A new approach to national comparisons of charter and traditional public schools. Journal of School Choice, 10(4), 479–494. https://doi.org/10.1080/15582159.2016.1238737


Mehrotra, S., & Panchamukhi, P. R. (2007). Universalizing elementary education in India: Is the private sector the answer? In P. Srivastava & G. Walford (Eds.), Private schooling in less economically developed countries: Asian and African perspectives (pp. 129–151). Oxford: Symposium Books.


Ministry of Education, Science and Technology. (2015). Registration guidelines for Alternative Provision of Basic Education and Training (APBET). Nairobi: Ministry of Education, Science and Technology, Republic of Kenya. Retrieved from http://globalinitiative-escr.org/wp-content/uploads/2016/04/ALTERNATIVE-PROVISION-OF-BASIC-EDUCATION-AND-TRAINING-APBET-B5-size-low-resolution.pdf


Ngware, M., Abuya, B., Admassu, K., Mutisya, M., Musyoka, P., & Oketch, M. (2013). Quality and access to education in urban informal settlements in Kenya. Nairobi: African Population and Health Research Center.


Ngware, M., Oketch, M., & Ezeh, A. C. (2011). Quality of primary education inputs in urban schools: Evidence from Nairobi. Education and Urban Society, 43(1), 91–116. https://doi.org/10.1177/0013124510379131


Nishimura, M., & Yamano, T. (2013). Emerging private education in Africa: Determinants of school choice in rural Kenya. World Development, 43, 266–275. https://doi.org/10.1016/j.worlddev.2012.10.001


Ohba, A. (2013). Do low-cost private school leavers in the informal settlement have a good chance of admission to a government secondary school? A study from Kibera in Kenya. Compare: A Journal of Comparative and International Education, 43(6), 763–782. https://doi.org/10.1080/03057925.2012.733536


Oketch, M., Mutisya, M., Ngware, M., Ezeh, A. C., & Epari, C. (2010). Free primary education policy and pupil school mobility in urban Kenya. International Journal of Educational Research, 49(6), 173–183. https://doi.org/10.1016/j.ijer.2011.01.002


Piper, B., & King, S., & Mugenda, A. (2016, September; revised from 2014). The Primary Math and Reading (PRIMR) Initiative: Endline impact evaluation – revised edition. Prepared under the USAID Education Data for Decision Making (EdData II) project, Task Order No. AID-623-M-11-00001 (RTI Task 13). Research Triangle Park, NC: RTI International. Retrieved from https://globalreadingnetwork.net/eddata/usaidkenya-primary-math-and-reading-primr-initiative-endline-impact-evaluation-revised


Piper, B., & Mugenda, A. (2012). The Primary Math and Reading (PRIMR) Initiative: Baseline report. Prepared for USAID/Kenya under the Education Data for Decision Making (EdData II) project, Task Order No. AID-623-M-11-00001 (RTI Task 13). Research Triangle Park, NC: RTI International. https://pdf.usaid.gov/pdf_docs/pa00hx75.pdf


Piper, B., Ralaingita, W., Akach, L., & King, S. (2016). Improving procedural and conceptual mathematics outcomes: Evidence from a randomised controlled trial in Kenya. Journal of Development Effectiveness, 8(3), 404–422. https://doi.org/10.1080/19439342.2016.1149502


Piper, B., & Zuilkowski, S. S. (2015). Assessing reading fluency in Kenya: Oral or silent assessment? International Review of Education, 61(2), 153–171. https://doi.org/10.1007/s11159-015-9470-4


Piper, B., & Zuilkowski, S. S. (2016). The role of timing in assessing oral reading fluency and comprehension in Kenya. Language Testing, 33(1), 75–98. https://doi.org/10.1177/0265532215579529


Piper, B., Zuilkowski, S. S., Kwayumba, D., & Strigel, C. (2016). Does technology improve reading outcomes? Comparing the effectiveness and cost-effectiveness of ICT interventions for early grade reading in Kenya. International Journal of Educational Development, 49, 204–214. https://doi.org/10.1016/j.ijedudev.2016.03.006


Piper, B., Zuilkowski, S. S., & Mugenda, A. (2014). Improving reading outcomes in Kenya: First-year effects of the PRIMR Initiative. International Journal of Educational Development, 37, 11–21. https://doi.org/10.1016/j.ijedudev.2014.02.006


Piper, B., Zuilkowski, S. S., & Ong’ele, S. (2016). Implementing mother tongue instruction in the real world: Results from a medium scale randomized controlled trial in Kenya. Comparative Education Review, 60(4), 776–807. https://doi.org/10.1086/688493


Putnam, R. D. (2016). Our kids: The American dream in crisis. New York: Simon and Schuster.


Romero, M., Sandefur, J., & Sandholtz, W. A. (2017). Can outsourcing improve Liberia's schools? Working Paper 462. Washington, DC: Center for Global Development. Retrieved from https://www.cgdev.org/publication/partnership-schools-for-liberia


Rose, P., & Adelabu, M. (2007). Private sector contributions to Education for All in Nigeria. In P. Srivastava & G. Walford (Eds.), Private schooling in less economically developed countries: Asian and African perspectives (pp. 67–87). Oxford: Symposium Books.


Rosenberg, T. (2016, June 14). Liberia, desperate to educate, turns to charter schools. The New York Times.


Smith, J. L. M., Cummings, K. D., Nese, J. F. T., Alonzo, J., Fien, H., & Baker, S. K. (2014). The relation of word reading fluency initial level and gains with reading outcomes. School Psychology Review, 43(1), 30–40.


Srivastava, P. (2007). For philanthropy or profit? The management and operation of low-fee private schools in India. In P. Srivastava & G. Walford (Eds.), Private schooling in less economically developed countries: Asian and African perspectives (pp. 153–186). Oxford: Symposium Books.


Srivastava, P. (2010). Privatization and Education for All: Unravelling the mobilizing frames. Development, 53(4), 522–528. https://doi.org/10.1057/dev.2010.88


Srivastava, P. (2013). Low-fee private schooling: Issues and evidence. In P. Srivastava (Ed.), Low-fee private schooling: Aggravating equity or mitigating disadvantage? (pp. 7–35). Oxford: Symposium Books.


Stern, J. M. B., & Heyneman, S. P. (2013). Low-fee private schooling: The case of Kenya. In P. Srivastava (Ed.), Low-fee private schooling: Aggravating equity or mitigating disadvantage? (pp. 105–128). Oxford: Symposium Books.


Stevis, M., & Clark, S. (2015, March 13). Zuckerberg-backed startup seeks to shake up African education. The Wall Street Journal.


Tooley, J. (2018). A chain of low-cost private schools for England. Policy, 34(2), 13–17.


Tooley, J., & Dixon, P. (2007). Private education for low-income families: Results from a global research project. In P. Srivastava & G. Walford (Eds.), Private schooling in less economically developed countries: Asian and African perspectives (pp. 15–39). Oxford: Symposium Books.


Tooley, J., Dixon, P., & Gomathi, S. V. (2007). Private schools and the millennium development goal of universal primary education: A census and comparative survey in Hyderabad, India. Oxford Review of Education, 33(5), 539–560. https://doi.org/10.1080/03054980701425664


Zuilkowski, S. S., Piper, B., Ong’ele, S., & Kiminza, O. (2018). Parents, quality, and school choice: Why parents in Nairobi choose low-cost private schools over public schools in Kenya’s free primary education era. Oxford Review of Education, 44(2), 258–274. https://doi.org/10.1080/03054985.2017.1391084






APPENDIX A


Baseline comparisons across school type and intervention group

  

Control

Intervention

Male

Public

57.8

46.3

 

LCPS

46.5

49.6

Age

Public

6.3

6.2

 

LCPS

6.2

6.1

Wealth index

Public

4.2

4.9

 

LCPS

4.5

4.4

Kiswahili correct letter sounds per minute

Public

11.9

18.7

 

LCPS

16.7

19.1

English correct letter sounds per minute

Public

19.7

28.3

 

LCPS

27.3

26.1

Kiswahili correct nonwords per minute

Public

1.8

4.5

 

LCPS

5.0

3.4

English correct nonwords per minute

Public

5.4

8.9

 

LCPS

10.3

8

Kiswahili oral reading fluency

Public

1.9

5.3

 

LCPS

5.7

3.8

English oral reading fluency

Public

1.8

8.1

 

LCPS

9.3

6.4

Kiswahili reading comprehension

Public

0.4

6.6

 

LCPS

9.5

5.4

English reading comprehension

Public

0.9

5.1

 

LCPS

6.4

2.1

Number identification

Public

10.1

13.2

 

LCPS

13.7

13.5

Quantity discrimination

Public

25.6

29.6

 

LCPS

34.1

34.5

Missing number

Public

16.7

17.6

 

LCPS

17.2

19.7

Word problems

Public

12.4

10.1

 

LCPS

11.3

11.3

Addition fluency

Public

3.6

3.2

 

LCPS

3.9

3.3

Subtraction fluency

Public

1.3

1.8

 

LCPS

1.8

1.6


Note. LCPS = low-cost private school.




Cite This Article as: Teachers College Record Volume 122 Number 1, 2020, p. 1-30
https://www.tcrecord.org ID Number: 22953, Date Accessed: 10/22/2021 10:14:17 PM

Purchase Reprint Rights for this article or review
 
Article Tools
Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Stephanie Zuilkowski
    Learning Systems Institute
    E-mail Author
    STEPHANIE SIMMONS ZUILKOWSKI is an associate professor at Florida State University, with a joint appointment in Educational Leadership and Policy Studies and the Learning Systems Institute. Her research focuses on school quality and reading outcomes in sub-Saharan Africa, including Kenya, Zambia, and Ethiopia.
  • Benjamin Piper
    RTI International
    E-mail Author
    BENJAMIN PIPER is the senior director for Africa education for RTI International, based in Nairobi. He provides technical support for education projects across sub-Saharan Africa and supervises Tusome, the Kenyan national literacy program, funded by the United States Agency for International Development (USAID) and the UK Department for International Development (DFID); and Tayari, a large-scale early childhood program and randomized controlled trial. He led the Primary Math and Reading (PRIMR) Initiative in Kenya from 2011 to 2015. Additionally, he has worked with the World Bank, UNICEF, and Save the Children.
  • Salome Ong’ele
    RTI International
    E-mail Author
    SALOME ONG’ELE is the Chief of Party for the Tusome Early Grade Reading Activity in Kenya, a five-year reading program funded by USAID. Ms. Ong’ele’s international education experience also encompasses work for RTI International on the PRIMR Initiative, for the government of Kenya, and for several large international nongovernmental organizations. She served as the National Education Advisor for World Vision Kenya, for which she led integrated education sector programming. Earlier in her career, she taught humanities in Kenyan schools and worked as a Senior Quality Assurance and Standards Officer.
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS