Sage Journals: Discover world-class research

Abstract

Using a longitudinal data set, this paper examines to what extent school-entry (N = 553; mean age 6.32 years) early mathematics and literacy skills predict students’ later achievement in a lower-middle-income country, Kenya. Controlling for socioeconomic status, intervention status, rural versus urban settings and parental literacy, the findings reveal that school-entry mathematics skills were significantly predictive of students’ end of Grade 2 mathematics and reading achievement in English and Kiswahili. Likewise, school-entry English early literacy skills predicted students’ end of Grade 2 mathematics and reading achievement in English and Kiswahili. This article extends earlier research on links between elements of school readiness and later achievement in high-income countries.

Keywords

early academic skills Kenya reading achievement mathematics

Introduction

Previous studies have produced strong evidence that early mathematics and literacy skills are significantly predictive of later mathematics and reading achievement (cf. Foster, 2010; Fyfe et al., 2019; Watts et al., 2014). A paper by Duncan and colleagues in 2007 on predictors of academic achievement in three high-income countries (Canada, Great Britain and the United States) reported on meta-analyses of six longitudinal data sets exploring the effects of school-entry mathematics and literacy skills, attention and socioemotional behaviours on third- and fifth-grade mathematics and reading achievement. The study offered multi-country evidence that early mathematics, literacy and attention skills were significantly predictive of later mathematics and reading achievement. Of particular note was the finding that early mathematics skills were as predictive of later reading as were early literacy skills, while literacy skills were much less predictive of later mathematics achievement than early mathematics skills. Although attention skills were a statistically significant predictor, they contributed considerably less to later achievement.

These studies have laid the groundwork for policy reforms, curricular recommendations and changes in instructional practices that emphasize the role and importance of early mathematics and literacy competencies. However, the generalizability of these findings is limited due to the context in which they were situated. Specifically, the outcomes of Duncan et al. (2007) and subsequent studies by Fyfe et al. (2019), Watts et al. (2014) and others (Foster, 2010) reflect research in high-income countries that have granted free and universal access to primary education in public schools for many decades. Given this context, it is not clear that early mathematics and literacy skills at school entry are universally and equally predictive across all geographical, cultural and socioeconomic settings.

Constructing standards, curricula and instruction based on this relatively narrow research – that is, without an extended range of cultural and socioeconomic contexts – may prove to be inappropriate. Of particular concern is the use of research from high-income countries to support interventions that include curricular changes in low-income and lower-middle-income countries (LMICs; low-income is defined as gross national annual income per capita of US$1045 or lower; lower-middle-income as US$1046 to US$4095; World Bank, 2021a). These countries vary not only in socioeconomic status (SES) from the high-income countries described in much of the research, but also in the range of cultural and educational norms in homes and schools within and between countries. These norms may differ greatly even within countries and depend on many factors, including gender, rural versus urban status, literacy levels, number of home languages spoken in a classroom, multilingual status of children and teachers, teacher education requirements, variables affecting age of school entry (e.g. distance to school, transportation availability or age of siblings) and size and grade configuration of school (Global Education Monitoring Report Team, 2017).

This paper considers several of these influential cultural and socioeconomic variables in the context of predicting students’ academic achievement in Kenya. Specifically, we examined to what extent school-entry mathematics and literacy skills (in English and Kiswahili) predicted later mathematics and reading achievement, controlling for SES, intervention status, rural versus urban settings and parental literacy. English and Kiswahili are the two predominant instructional languages in early primary school in Kenya. We used longitudinal data collected in 2012 and 2013 that were part of a larger data set collected for an impact evaluation. This study builds on findings from high-income contexts and may provide much needed evidence in future policy reforms, including the development and implementation of standards, curricula and instruction in low-income and LMIC.

The following section explores additional research on predictors of achievement in high-income contexts. We then describe the more limited body of research on predictors of achievement in LMICs.

Predictors of achievement in high-income contexts

Three years after the publication of the study by Duncan and colleagues (2007), additional research teams re-analysed these and related data sets, examining relationships among a range of variables including gender, race and ethnicity, fine motor skills and later academic achievement. These studies confirmed Duncan and colleagues’ (2007) earlier findings, but also painted a more nuanced picture of development, including differential outcomes by gender, the role of fine motor skills, effects of general knowledge on later achievement and relationships between early mathematics skills and later socioemotional behaviours (Foster, 2010).

Early mathematics skills as a predictor of later mathematics achievement

Later studies investigated specific early mathematics subdomains and more distal effects. Nguyen and colleagues (2016) determined that advanced counting and numeracy skills in preschool, when compared with early geometry, patterning and measurement skills, were most predictive of fifth-grade mathematics achievement. In a similar study investigating performance on fourth- to sixth-grade standardized mathematics assessments, Fyfe et al. (2019) found that end-of-preschool knowledge of nonsymbolic quantity and repeating patterns was predictive of later mathematics achievement, mediated by first-grade symbolic mapping and calculation knowledge. In a longitudinal study that controlled for intelligence, working memory, mathematical achievement and attention in school (Geary et al., 2013), school-entry number system knowledge predicted functional numeracy at age 13. Watts and colleagues (2014) found strong relationships between preschool mathematics abilities and mathematics achievement through age 15, even when controlling for attention, home environments and cognitive abilities. Taken together, these studies provide additional evidence confirming the predictive relationship between early mathematics skills and later mathematics achievement.

Early literacy skills as a predictor of later reading achievement

The effects of early literacy skills on later reading achievement have been investigated in equal, if not greater, measure. In a seminal longitudinal study of 626 children, Storch and Whitehurst (2002) found that reading accuracy and comprehension in early elementary school was directly related to a child’s kindergarten early literacy skills, such as naming letters, printing name, identifying letters and pronouncing letter sounds. In a meta-analysis of 299 studies examining the links between early language and literacy skills and later literacy achievement, the National Early Literacy Panel (NELP, 2008) concluded that phonological awareness; alphabet knowledge; rapid naming of letters, numbers, objects or colours; writing letters and name; and phonological memory had the strongest predictive relationships to later literacy. Subsequent longitudinal studies have found that school-entry literacy skills predicted reading comprehension at age 12 in a New Zealand population (Suggate et al., 2018) and that kindergarten early literacy skills predicted comprehension at fourth grade in the United States (Dickinson and Porche, 2011). Taken together, these studies provide evidence that school-entry early literacy skills play an important part in the development of later reading skills.

Cross-domain predictors of achievement

Researchers have also examined the predictive relationship between early language and literacy skills and later mathematics achievement. Purpura et al. (2011) found that performance on preschool vocabulary and print knowledge assessments was predictive of numeracy skills a year later when the analysis controlled for beginning numeracy skills and nonverbal cognitive skills. Pace et al. (2019) found that language at school entry predicted four of five domains (language, mathematics, reading and social skills) in first through fifth grade, early literacy skills predicted outcomes in three domains (language, mathematics and reading), and early mathematics skills predicted two (mathematics and reading). In the study by Fyfe and colleagues (2019), literacy skills as measured by the Woodcock–Johnson III Letter–Word Identification subtest also predicted performance on the standardized mathematics assessment in fourth grade, but not fifth or sixth grades; and on narrative recall in all three grades. Adding to the previous research documenting the predictive association between early and later skills within a domain, these findings point to the importance of cross-domain dependencies in child development within high-income countries.

Predictors of achievement in low-income and LMIC

Studies that have explored predictors of later achievement in LMICs have rarely examined early domain-specific predictors of later achievement. Many have concentrated their efforts on investigating the consequences of poverty and lack of access to resources (e.g. developmental delays due to malnutrition, inadequate cognitive stimulation, maternal depression and exposure to violence), parental education and attendance at an early childhood education program (cf. Engle et al., 2011; McCoy et al., 2016; Rao et al., 2019; Tanner et al., 2015; Wolf and McCoy, 2019a; to name a few). Others, like Hanushek and Woessmann (2008), have looked at relationships in adulthood between cognitive skills and economic well-being. In their meta-analysis of a range of countries, Hanushek and Woessmann concluded that cognitive skills, measured by literacy and numeracy levels, predicted annual income, but they found considerable disparity between low- and high-income countries. These disparities highlight the need for more research on the development of mathematics and reading skills from LMICs.

Many LMICs are investing in improving learning outcomes in lower primary school. The magnitude of these efforts is substantial, including national or nearly national programs under way in Kenya, Malawi, Mozambique, Rwanda, Tanzania, Uganda and Zambia in East and Southern Africa alone. Several studies in Kenya and Ghana have examined the results of these interventions and can inform our understanding of the development of reading and mathematics achievement in LMICs.

In a series of studies of a large-scale Kenyan preschool intervention program that included student workbooks in language/literacy (Kiswahili and English) and mathematics and teacher guides in language/literacy (Kiswahili and English), mathematics and social and life skills, researchers found that the program had substantial impacts on school readiness (Piper et al., 2018a), including mathematics (Piper et al., 2018b). In a related study tracking a subset of these students into primary school, the gains in mathematics outcomes at the end of pre-primary faded out as children entered Grades 1 and 2 (Kwayumba et al., 2019).

In a Ghanaian study on a teacher in-service training and parental awareness program, researchers examined the effects of early preschool academic (literacy and mathematics) and the effects of non-academic skills (executive functioning and socioemotional skills) on end-of-preschool skills. Wolf and McCoy (2019b) found that early literacy and mathematics skills at the beginning of the first year of preschool (mean age of children was 5.2 years) were significantly predictive of early literacy and early mathematics skills at the end of the school year – and, likewise, literacy and mathematics skills at the end of the first year of preschool were predictive of those at the end of the second year of preschool. However, this study did not follow children into the primary years.

These Kenyan and Ghanaian studies have contributed to the body of knowledge on achievement in LMICs, with the primary purpose of measuring intervention effects. A longitudinal study in such contexts that examined school-entry literacy and mathematics skills and their effects on later reading and mathematics achievement could inform our knowledge base on the relative contributions of these school-entry skills. The purpose of this paper is to examine early predictors of later reading and mathematics achievement in an LMIC, Kenya, and as such to contribute to and extend our understanding of the role of school-entry skills on children’s later development in the LMIC context.

Method and sample

The context

Kenya is a lower-middle-income country with a population of 54 million, with 28% of the population living in urban centres (World Bank, 2021b, 2021c). There are two official languages – Kiswahili and English – and 42 additional dialects or languages (Embassy of the Republic of Kenya, 2019). The teacher-to-student ratio in public primary schools is 1:50 (Kenya National Bureau of Statistics, 2015). The language-of-instruction policy for preschool and early primary is use of the local language, but Kiswahili and English predominate (56% versus 44% mother tongue; Begi, 2014: 40; Hungi et al., 2018). English is the de facto language used for instruction in primary school (Abuya et al., 2018).

Study design

We used longitudinal data collected from a set of randomized controlled trials on an instructional program (the treatment group) designed to inform Kenya’s Ministry of Education (MOE) on how they could most effectively increase access to quality education in Grades 1 and 2, in both reading and mathematics. Key components of the instructional program were professional development, instructional guides and coaching for classroom teachers, as well as new materials and texts for students.

The instructional program was designed to be implemented in public schools and low-cost private schools (LCPSs). Schools in Kenya are clustered into zones based on geographic proximity, with around 12 to 30 schools in a typical zone (University of Kenya, Nairobi and Makerere University, Uganda, 2019). Each of these zones was supported by a Curriculum Support Officer (CSO; previously titled Teachers’ Advisory Centre Tutors, or TACs). Guided by the MOE, eligible zones in four Kenyan counties and LCPSs in Nairobi were randomly assigned to treatment and control groups. Students in the treatment group received the following interventions: (a) students received English and Kiswahili textbooks and a mathematics workbook in English; (b) teachers were given structured teachers’ guides containing lessons and guidance on how to use the student textbooks in a carefully paced program; (c) teachers received 10 days of CSO-led hands-on training on effective use of the instructional program materials in their classrooms; and (d) CSOs provided classroom support to teachers, visiting the teachers monthly and offering collaborative feedback on the quality of their instruction and their use of the materials.

The longitudinal study, with its three subjects (English, Kiswahili and mathematics), produced a data set uniquely equipped to allow us to examine the following research questions (controlling for SES, urban/rural, mother’s literacy, father’s literacy and treatment/control status):

1. To what extent do school-entry early mathematics and English and Kiswahili early literacy skills predict mathematics achievement at end of Grade 2?

2. To what extent do school-entry early mathematics and English and Kiswahili early literacy skills predict reading achievement in English at end of Grade 2?

3. To what extent do school-entry early mathematics and English and Kiswahili early literacy skills predict reading achievement in Kiswahili at end of Grade 2?

Participants

The data set contains records from 628 students with three waves of data, 312 (49.7%) of whom were girls. There were 357 treatment and 271 control students in public schools and LCPSs. We used data collected at school entry (January 2012), the end of Grade 1 (October 2012) and the end of Grade 2 (October 2013). Students were randomly selected from treatment and control schools using simple random sampling at the baseline data collection period in January 2012. At that time period, the mean age of the students was 6.32 (standard deviation [SD] = 0.98) years old, and sampled students ranged from 4 to 10 years old. Attrition was 37.0% over the 3 years, similar to a prior 2-year longitudinal study in Kenya (28–41%; Alderman et al., 2001) and in a six-study analysis (minimum 28, median 34.4, maximum 70; Molina Millán and Macours, 2017), but higher than desirable. The initial longitudinal sample was powered to identify a minimum detectable effect size of 0.2 SD with 10 students per school in 50 treatment and 50 control schools, resulting in a sample of 996 children. There were small differences in background characteristics of those who persisted to the end of Grade 2 data collection on four measures; students retained in the longitudinal study were 6% more likely to have a phone in the household (p = .04), 9% less likely to have electricity in their house (p < .01), 11% more likely to have a bicycle in the house (p < .01) and 0.1 years younger (p = .04) than those who did not persist in the study. The average magnitude of the differences by attrition status was d = 0.00 SD.

Because this study focuses on students’ development over time, we used listwise deletion and retained only participants with complete data at the two time points of interest: school entry and end of Grade 2. As such, 553 participants were retained for the analyses and were distributed relatively equally among treatment and control conditions (55% within treatment condition) and genders (50% girls). A larger proportion of students in our sample were from peri-urban or urban areas (73%) compared with rural. Students in the sample had access to formal and/or informal pre-primary education; 96% of students confirmed attendance in such programs (‘Did you go to a nursery or preschool before Class 1’? in separate questionnaire described below).

Ethical considerations

In accordance with legal and ethical requirements, approval to collect data from students in the sampled schools was requested and obtained from the National Council of Science and Technology, Kenya Medical Research Institute and the RTI Institutional Review Board. The reviews included the research protocol, instruments, data collection procedures, mechanisms for confidentiality of data and parental/guardian consent.

Measures

The data in this study were collected using three instruments: the Early Grade Mathematics Assessment (EGMA), the Early Grade Reading Assessment (EGRA) and a separate questionnaire used to interview the students on pre-primary access, previous year enrolment, home language and home assets (see Table 1; Dubeck and Gove, 2015; Platas et al., 2014; RTI International, 2016). Student assessments and interviews were implemented at entry to Grade 1, end of Grade 1 and end of Grade 2. This study reports only on school-entry and end of Grade 2 measures.

Table 1.

Descriptive statistics for subtests and household wealth and parental literacy items.

Subtests or items	Mean	SD.	Min	Max
School entry
Object counting	45.53	25.43	1	100.00
Number identification^a	10.67	6.71	0	52.94
Number discrimination	2.86	2.21	0	9.00
Missing number	1.63	0.97	0	9.00
Letter sounds – English^a	18.15	17.17	0	72.00
Decoding – English^a	5.92	9.15	0	61.25
Letter sounds – Kiswahili^a	13.00	9.81	0	53.00
Decoding – Kiswahili^a	2.98	6.42	0	35.00
End of grade 2
Number identification^a	26.31	11.63	0	80.00
Number discrimination	6.64	2.61	0	10.00
Missing number	4.61	2.19	0	10.00
Word problems	2.16	1.33	0	5.00
Addition level 1^a	11.09	4.98	0	32.43
Subtraction level^a	7.85	4.62	0	46.15
Addition level 2	1.98	1.64	0	5.00
Subtraction level 2	1.21	1.41	0	5.00
Oral reading fluency – English (words/minute)	46.52	36.04	0	166.96
Oral reading fluency – Kiswahili (words/minute)	29.79	21.61	0	120.00
Student questionnaire: SES composite measure
Radio	1.88	0.40	1	2
Telephone or mobile phone	1.96	0.19	1	2
Electricity	1.64	0.48	1	2
Television	1.65	0.48	1	2
Refrigerator	1.15	0.35	1	2
Inside toilet	1.14	0.35	1	2
Bicycle	1.44	0.50	1	2
Motorcycle	1.19	0.39	1	2
Car, truck, 4 by 4, tractor or engine boat	1.18	0.38	1	2
Student questionnaire: Parental literacy
Mother literacy	1.97	0.18	1	2
Father literacy	1.94	0.23	1	2

^aTimed to 1 minute.

At baseline, an instrument adaptation workshop was held in October 2011 with the MOE and officers from several Semi-Autonomous Government Agencies. The purpose of the workshop was to adapt and validate the instruments in the Kenyan context and to engage and inform stakeholders. Those who attended the workshop included officers from the Kenya Institute of Education, Kenya National Examinations Council and Directorates of Education (Quality Assurance, Basic Education, Standards, Policy and Planning). Other organizations represented included the University of Nairobi, SIL International and teachers.

Before the end of the adaptation workshop, workshop participants conducted a mini-pilot of the instruments among Grade 1 and 2 students in several LCPSs in Nairobi. The workshop participants shared their experiences in the field and recommended changes to the instruments as well as more efficient procedures for tool administration. Adaptations were made to assessment terms and phrasing to accommodate local dialect variations, and administration procedures were fine-tuned to allow for more efficient engagement with a variety of school settings.

Data from the EGRA, EGMA and questionnaire were obtained by data collectors who had been trained by experienced staff who had participated in administration of previous Kenyan versions of the English and Kiswahili EGRA and EGMA. The assessors attended a 1-week training during January 2012 and September 2013 for the baseline and endline, respectively. As part of the training, these assessors underwent observer reliability tests for both the EGRA and EGMA at the baseline and at the endline. At baseline, the average scores for Kiswahili reliability were 95.3%. For English, the average reliability scores were 96.1% and for mathematics, 96.0%. At the endline, two sets of interrater reliability tests were conducted. The average score for English was 95.5%, Kiswahili was 95.0% and mathematics was 98.0%.

The 68 assessors were grouped into 17 teams of four people each. The most effective (as measured by EGRA and EGMA reliability tests) assessor in each team was assigned as the team supervisor. A fieldwork manual was employed during a subsequent 2-day supervisor training. The manual explained the supervisors’ role in validating data at the field level and ensuring that consistency in procedures and adherence to ethical principles were maintained. Each student was assessed one-on-one with the EGRA–Kiswahili, EGRA–English and EGMA instruments at the school. The assessment order was randomized to avoid a ‘fatigue effect’ across the whole sample.

Early grade reading assessment

The English and Kiswahili EGRAs were administered to measure early literacy skills at school entry and reading achievement at the end of Grade 2. The EGRA included subtests of letter sounds, nonwords per minute, 1-min oral reading fluency, 1-min oral reading comprehension, 3-min timed oral reading score and 3-min oral reading comprehension. For school entry, only the letter sounds and nonword reading per minute skills were appropriate expectations. Supporting this assertion, at school entry, all of the subtests except letter sounds and nonword reading had zero scores of 70% or more in the English version and 76% or more in the Kiswahili version. In the letter sounds subtest, children were asked to identify the correct English or Kiswahili letter sounds of an array of 100 letters. The score was a fluency measure: number correct per minute. In the decoding (nonword reading) subtest, children were asked to decode short words (with consonant–vowel patterns of CVCV for Kiswahili and CVC for English) that were structured like English/Kiswahili words but were not real words, in an array of 50 nonwords. This score also was a fluency measure, scored as the number correct per minute.

Reading achievement at the end of Grade 2 was measured using the 1-min oral reading fluency measure. Children were asked to read a short story of ∼60 words fluently, timed to 1 min. Scores were generated by counting the correct words read per minute. Oral reading fluency has a high correlation with reading comprehension and is theoretically supported as an indicator of overall reading competence (Fuchs et al., 2001).

Early grade mathematics assessment

The EGMA was administered to measure early mathematics skills at school entry and mathematics achievement at the end of Grade 2. Early Grade Mathematics Assessment subtests (administered in the child’s preferred language, English or Kiswahili) included object counting, number identification, number discrimination, missing number, addition level 1, subtraction level 1, addition level 2, subtraction level 2 and word problems. At school entry, object counting, number identification, number discrimination and missing number were appropriate expectations (like the EGRA, the EGMA was administered at all time points to allow for analyses of gains across time in the larger study; see Table 1). These subtests exist in other preschool or school-entry assessments, exist in preschool curricula and are developmentally appropriate. In the object counting subtest, the child was presented with 100 circles (10 rows of 10 circles) and requested to count from left to right. In the number identification subtest, the child was presented with 20 numbers and asked to give the names of the numbers (English or Kiswahili answers were accepted). The child received a score of numbers correctly identified per minute. In the number discrimination subtest, the child was asked to state the larger-value number from a pair of numbers. The objective of this subtest was to test the children’s number fluency, number sense and place value. Their responses were scored as a percentage correct out of total attempted. In the missing number subtest, children were asked to identify a missing number in an array of four (one of which was represented by an empty box). Their final score was a percentage correct out of 10 problems. The objective of this subtest was to test children’s ability to detect number patterns, which is a foundation for algebraic skills (Sarama and Clements, 2009).

To measure mathematics achievement at the end of Grade 2, all EGMA subtests were administered, with the exception of object counting (see Table 1). The addition and subtraction level 1 subtests were timed tests (60 s) consisting of 20 items each that increased in difficulty. No addends were greater than 10, and no sums were greater than 19. The subtraction problems were the inverse of the addition problems. The addition and subtraction level 2 subtests were untimed tests consisting of five items each that increased in difficulty, with two-digit addends/subtrahends/minuends, and with a stop rule of four successive errors. Addition level 2 was not given to students who received a score of zero for addition level 1, and subtraction level 2 was not given to students who received a score of zero for subtraction level 1. No sums were greater than 70. The subtraction problems were the inverse of the addition problems. The word problems subtest was an untimed test consisting of six items that increased in difficulty, with a stop rule of four successive errors. Three of these items used numbers that matched three items from the addition and subtraction level 1 subtest.

Scoring of the EGRA and EGMA

To overcome a variety of concerns in relation to creating composite scores from these two assessments, we used exploratory factor analysis (EFA) with the respective EGRA and EGMA subtests for school-entry mathematics, early literacy in English and Kiswahili and mathematics achievement at the end of Grade 2. This method allowed us to create a single-factor score for each child, for each construct and time point. The EFAs were conducted in R (R Core Team, 2016) with an oblimin rotation. Each single-factor EFA had loadings of 0.53 or higher. Additionally, fit indices were satisfactory (root mean square error of approximation [RMSEA] ≤ 0.06, Tucker–Lewis index [TLI] ≥ 0.95, root mean square residual [RMSR] ≤ 0.08 according to Hu and Bentler, 1999) for each of the EFAs, except for the one-factor solution for end of Grade 2 mathematics achievement (RMSEA = 0.13, TLI = 0.88, RMSR = 0.06). Because the subtests used for the mathematics achievement are theoretically related, the one-factor solution was retained. For the end of Grade 2 scores for English and Kiswahili reading achievement, EGRA oral reading fluency scores were standardized to allow for increased comparability with the factor scores.

Covariates

Like Duncan et al. (2007) and others, we attempted to isolate the effects of these school-entry skills by controlling for contextual influences that might have been related to children’s later achievement. Thus, we included five covariates in the analysis to account for student characteristics that might be related to achievement: a SES composite measure from a list of household wealth items commonly used in sub-Saharan Africa (see Table 1), urban/rural, mother’s literacy, father’s literacy and treatment/control status. Urban/rural, intervention/control, mother’s literacy and father’s literacy were dichotomous variables, with ‘1’ representing urban, treatment, a literate mother and a literate father, respectively. Urban/rural and intervention/control status were collected at the school level. Socioeconomic status data and mother’s literacy and father’s literacy (‘Can your mother/father read and write’?) were obtained through student self-report at end of Grade 2.

The SES measure was created from the set of variables intended to proxy living standards that was included in the student questionnaire. A principal component analysis (PCA) was conducted in R (R Core Team, 2016) to create a single-factor score for SES based on the loadings for the assets. Factor scores for SES have been successfully estimated using PCAs in a wide range of studies (Vyas and Kumaranayake, 2006). See Table 2 for assets and loadings.

Table 2.

Principal component analysis for socioeconomic status measure.

Items	Principal component scores
At your house, do you have
a radio?	0.10
a telephone or mobile phone?	0.23
electricity?	0.74
a television?	0.75
a refrigerator?	0.60
a toilet inside the house?	0.53
a bicycle?	0.19
a motorcycle?	0.26
a car, truck, 4 by 4 vehicle, tractor or engine boat?	0.51
Eigenvalue	2.17
Variance explained (%)	0.24

For all of these covariates, we used responses from the end of Grade 2 questionnaire. There was agreement between questionnaire responses across data collections at school-entry and end of Grade 2 ranging from 68%–89%, and we hypothesized that older children would have had a better understanding of the resources in their homes and their parents’ literacy skills.

Analytic plan

We used multiple regression for each of the research questions to examine how early mathematics skills and early literacy in English and Kiswahili were related to later mathematics and reading achievement, controlling for SES, urban/rural, mother’s literacy, father’s literacy and treatment/control status. All analyses were conducted in R (R Core Team, 2016).

Results

Table 1 contains the descriptive statistics for the items and subtests. Each of the regressions revealed statistically significant predictors of later mathematics and literacy achievement; see Table 3.

Table 3.

Coefficients in regressions of mathematics and reading achievement at end of Grade 2.

Model description	End of grade 2
Model description	Reading – English	Reading – Kiswahili	Mathematics
Intercept	−0.11 (0.21)	−0.10 (0.23)	−0.40 (0.22)
Early literacy – English	0.33*** (0.06)	0.21*** (0.06)	0.19** (0.06)
Early literacy – Kiswahili	0.05 (0.05)	0.11 (0.06)	−0.01 (0.06)
Early mathematics	0.20*** (0.04)	0.22*** (0.05)	0.31*** (0.05)
SES	0.11** (0.04)	0.06 (0.04)	0.03 (0.04)
Urban/rural	0.18 (0.09)	0.06 (0.10)	−0.13 (0.10)
Mother – literacy	−0.28 (0.20)	−0.17 (0.22)	0.32 (0.22)
Father – literacy	0.10 (0.16)	0.08 (0.18)	0.08 (0.17)
Treatment/control	0.27*** (0.07)	0.25*** (0.07)	0.18* (0.07)
R ²	0.40	0.28	0.24

SES: socioeconomic status.

***p < .001; **p < .01; *p < .05.

Research Question 1: To what extent do school-entry early mathematics and English and Kiswahili early literacy skills predict mathematics achievement at end of Grade 2? Early English literacy skills (β = 0.19, p < .01) and early mathematics skills (β = 0.31, p < .001) were statistically significant predictors of mathematics achievement at the end of Grade 2 (F[8, 543] = 20.91, p < .001). While these were significant predictors, linear hypothesis testing indicated that there was no statistically significant difference (F[1, 543] = 2.02, p = 0.16) between the regression coefficients for early English skills (β = 0.19, p < .01) and early mathematics skills (β = 0.31, p < .001). Early Kiswahili literacy skills were not predictive of mathematics achievement at end of Grade 2. Of the control variables, only treatment status was significant (β = 0.18, p < .05). The predictors in the full model were responsible for 24% of the variance in later mathematics achievement. Early mathematics skills uniquely explained 6% of the variance, and early English literacy skills uniquely explained 2% of the variance. Treatment uniquely explained 1% of the variance, while the other independent variables uniquely explained less than 1% of the variance each. The remaining variance was explained by combinations of the independent variables.

Research Question 2: To what extent do school-entry early mathematics and English and Kiswahili early literacy skills predict reading achievement in English at end of Grade 2? Early mathematics skills (β = 0.20, p < .001) and early English literacy skills (β = 0.33, p < .001) were statistically significant predictors of English reading achievement at the end of Grade 2 (F[8, 543 = 45.69], p < .001). While these were significant predictors, linear hypothesis testing indicated that there was no statistically significant difference (F[1, 543] = 2.64, p = 0.11) between the regression coefficients for early English literacy skills (β = 0.33, p < .001) and early mathematics skills (β = 0.20, p < .001). Kiswahili early literacy skills were not a predictor of reading achievement in English at end of Grade 2. Of the control variables, only SES (β = 0.11, p < .01) and treatment (β = 0.27, p < .001) were statistically significant predictors. The predictors in the full model were responsible for 40% of the variance in later English literacy scores. Early mathematics skills uniquely explained 2% of the variance, and early English literacy skills uniquely explained 4% of the variance. Treatment uniquely explained 2% of the variance, SES uniquely explained 1% and the other independent variables uniquely explained less than 1% of the variance each. The remaining variance was explained by combinations of the independent variables.

Research Question 3: To what extent do school-entry early mathematics and English and Kiswahili early literacy skills predict reading achievement in Kiswahili at end of Grade 2? Early English literacy skills (β = 0.21, p < .001) and early mathematics skills (β = 0.22, p < .001) were statistically significant predictors of Kiswahili reading achievement at the end of Grade 2 (F[8, 543] = 26.35, p < .001). While these were significant predictors, linear hypothesis testing indicated no statistically significant difference (F[1, 543] = .02, p = 0.89) between the regression coefficients for early English literacy skills (β = 0.21, p < .001) and early mathematics skills (β = 0.22, p < .001). Kiswahili early literacy skills were not a predictor of Kiswahili reading achievement at end of Grade 2. Of the control variables, only treatment status (β = 0.25, p < .001) was statistically significant. The predictors in the full model were responsible for 28% of the variance in later Kiswahili literacy scores. Early mathematics skills uniquely explained 3% of the variance, and early English literacy skills uniquely explained 2% of the variance. Treatment uniquely explained 2% of the variance, and the other independent variables uniquely explained less than 1% of the variance each. As with the other two regressions, the remaining variance was explained by combinations of the independent variables.

Discussion

Our results indicate that, in models that controlled for key variables, school-entry early mathematics and early literacy skills in English were statistically significant predictors of Grade 2 mathematics achievement, Grade 2 English reading achievement and Grade 2 Kiswahili reading achievement. School-entry early mathematics was the strongest predictor of Grade 2 mathematics achievement (β = 0.31), and school-entry early literacy in English was the strongest predictor of end of Grade 2 English reading achievement (β = 0.33). School-entry early literacy in English was predictive of Grade 2 Kiswahili reading (β = 0.21) and mathematics achievement (β = 0.19), and school-entry mathematics skills were predictive of Grade 2 Kiswahili (β = 0.22) and English reading achievement (β = 0.20). School-entry Kiswahili literacy skills were not predictive of reading or mathematics achievement. While school-entry early mathematics and school-entry early literacy skills in English were statistically significant predictors in each of the models, there were no statistically significant differences between the regression coefficients for school-entry early mathematics and school-entry English early literacy in any of the models.

Two of the findings from this LMIC study in Kenya replicated earlier studies that showed strong relationships between early foundational literacy and mathematics skills and later achievement (Duncan et al., 2007; Wolf and McCoy, 2019b). In this and previous studies, early mathematics skills were strongly related to later mathematics achievement and early English literacy skills were strongly related to later reading achievement (in this study, respectively .31 and .33 SD). However, we found that early Kiswahili literacy skills were not predictive of any of the later outcomes. Additionally, while the effect of early English literacy skills was larger than the effect of early mathematics skills on later reading achievement, linear hypothesis testing showed no significant differences between these effects.

Early mathematics skills as a predictor of later mathematics achievement

Similar to the outcomes in many earlier studies (Duncan et al., 2007; Geary et al., 2013; Nguyen et al., 2016), early mathematics skills predicted later mathematics achievement in this study. Mathematics skills build upon one another, and in particular rely on an understanding of counting and quantity (Sarama and Clements, 2009), the subdomains that were measured at school entry in the subtests involving counting objects, identifying numerals, comparing numbers and determining missing numbers. These findings corroborate findings from high-income contexts, supporting the notion that these early mathematics skills form the foundation of later mathematics achievement, regardless of country-level income status.

Early literacy as a predictor of later reading achievement

This study measured early literacy and later reading achievement for all children in two languages (Kiswahili and English). Early literacy skills in English were predictive of later reading achievement in English and Kiswahili, with greater coefficients for English. Like mathematics, early skills create the foundation for later proficiencies in literacy. In particular, in studies of English language development, phonological awareness; alphabet knowledge; rapid naming of letters, numbers, objects or colours; writing letters and name; and phonological memory have had the strongest predictive relationships with later literacy (cf. NELP, 2008, for an overview). The present study used two subtests of school-entry skills – letter sounds and nonword reading – to predict later reading achievement (as measured by oral reading fluency). These two subtests directly or indirectly measured the skills of phonological awareness and alphabet knowledge. These skills are also known to contribute to reading acquisition in a wide variety of alphabetic languages (Ibrahim et al., 2007; Share 2008).

Early literacy skills in Kiswahili were not predictive of reading achievement in Kiswahili or English. It is unlikely that children entered primary school with equal fluency in both languages (children’s speaking fluency in these languages was not measured in this study). This may, in part, be due to limited literacy instruction in Kiswahili before primary school, resulting in greater early literacy skills in English. Most children in Kenya use three languages in their everyday interactions (one of the 42 mother tongues, Kiswahili and English). Kiswahili is not the mother tongue for a large percentage of young children in Kenya, but it is used to communicate more broadly. In many parts of Kenya, English skills are prioritized because of the perception of English as a tool for upward mobility (Begi, 2014; Dubeck et al., 2012) and there is empirical evidence that the majority of instruction is in English (Piper and Miksic, 2011). The lack of impact from early literacy skills in Kiswahili on later reading achievement should not be construed as evidence for a particular language-of-instruction model, but rather as an argument for more research on how literacy and numeracy skills interact in complex multilingual contexts.

Cross-domain predictors of achievement

Similar to earlier studies, early literacy skills in English were predictive of later achievement mathematics. Pace and colleagues (2019) found that performance on the Woodcock–Johnson Letter–Word Identification Scale, which emphasizes letter and early decoding tasks (similar to the EGRA school-entry subtests), predicted performance on later mathematics skills. Hypotheses about these positive relationships include the influence of language on mathematics skills (Purpura et al., 2017) and the use of symbols in both domains (Yamagata, 2007), School-entry Kiswahili literacy skills were not predictive of Grade 2 mathematics achievement. Initial Kiswahili literacy skills may be less predictive of later mathematics outcomes than early English literacy skills, given the limited utilization of Kiswahili in mathematics instruction.

School-entry mathematics skills were predictive of end of Grade 2 Kiswahili and English reading achievement. This may be for similar reasons as the school-entry literacy to mathematics achievement relationships, specifically the importance of language and symbol in both of these domains. An additional hypothesis that supports the reading and mathematics cross-domain relationships is the notion that domain-general skills are at work in both domains, such as attentive behaviour, reasoning, rapid autonomic naming (objects, colours, letters and numerals) and visuospatial memory (cf. Fuchs et al., 2016, for a review).

Limitations

Limitations to this study include generalizability, attrition and aspects of the data set that limited broader examination of achievement. These are discussed below.

As in many studies with data from high-income countries, this study focused on one particular LMIC context. Many aspects of the Kenyan environment, from the language policy to the rural/urban population distribution to the design of the education system, limit the generalizability of the study to other LMIC contexts. On the other hand, the Kenyan education system is closer to the realities facing the majority of children worldwide who are in environments that do not support the acquisition of basic skills (Global Education Monitoring Report Team, 2017) than are systems in the United States or in other high-income countries. In addition, Kenya’s language environment is similar to that in many sub-Saharan countries, with multiple languages of instruction and many different home languages. Therefore, these findings may be more applicable to other LMIC contexts than previous research conducted in high-income countries.

As noted earlier, at 37%, the level of attrition in study was higher than desirable. Longitudinal studies, by nature of their design, frequently feature significant data point loss. In a study examining similar constructs, Claessens et al. (2009) began with 17,622 reading scores in kindergarten and finished the study with 11,265 in fifth grade, a loss of 36%. Even though the magnitude of attrition in our study may have been an improvement over other similar studies (Alderman et al., 2001; Molina Millán and Macours, 2017), there may be factors affecting this attrition that have not been measured and that affect the outcomes in question.

We worked with a data set that was not explicitly designed to measure the breadth of early literacy and mathematics skills and later mathematics and reading achievement that we know are important (Fyfe et al., 2019; Pace et al., 2019; Purpura et al., 2011). Oral reading fluency is one of the most important components of reading achievement in primary school and is predictive of other measures of achievement such as comprehension (Fuchs et al., 2001), but is not the only one (cf. Dowd and Bartlett, 2019). Likewise, skills involving numbers and operations are significant predictors of mathematics achievement in primary school (Nguyen et al., 2016), but arguably are not the only ones (Fyfe et al., 2019). Including other domains (cf. NELP, 2008, and Purpura et al., 2019 for reviews) that have been shown to influence later achievement could provide guidance on the development of curriculum, standards and interventions, which was limited here by the subdomains included. The end of Grade 2 mathematics achievement score, although theoretically derived, had less than optimal fit indices. Future studies could consider additional subdomains and investigate the use of alternative factor models.

Future directions

An obvious implication of our findings – that early mathematics and literacy skills are significant predictors of later mathematics and reading achievement – is that both of these academic domains should be supported in the pre-primary years. Many interventions internationally (as in the United States) emphasize the support of early literacy skills over mathematics (Bethell, 2016; Lutfeali et al., 2021; Piper et al., 2016). As Purpura et al. (2019) suggested, the early childhood field needs to move away from siloed approaches, in both research and interventions, that target only mathematics or only literacy. Intervention and education policies that rely solely on one subject domain or the other may short-change children.

Although this research into the influence of early mathematics and literacy skills on later mathematics and reading achievement in LMICs is nascent, combined with similar outcomes from research in high-income countries, it suggests that both of these domains are important before school entry, at home and in school. Although Purpura and colleagues (2019) argued for their inclusion in high-income settings such as the United States, their advice is well worth considering in LMICs. As LMICs engage in policy reforms, curricular development and changes in instructional practices, considering both early mathematics and literacy as important contributors to later academic success may set the stage for optimally supporting both.

More research is needed to understand how early skills in these academic domains interact in LMICs. This study supports the notion that increases in early mathematics and literacy skills positively influence later achievement in both domains. However, the lack of predictive power for literacy skills in Kiswahili in this study points to the need for research in settings where multiple languages are both spoken and used as languages of instruction.

Finally, implications of our study can also inform non-academic early interventions. Improving children’s early nutrition and psychosocial well-being results in pre-primary improvements in academic domains (Tanner et al., 2015). This study showed that early mathematics and literacy skills at school entry are as important for later success in LMICs as they are in high-income environments. Ensuring that children’s brains are well equipped and ready to engage in pre-primary classroom activities in these highly predictive domains is essential. Funding interventions when children cannot fully engage in learning activities wastes money, time and lives.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

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).

Data availability

Benjamin Piper was the Chief of Party (Project Director) of the Primary Math and Reading (PRIMR) Initiative from which the data were derived. He does not have any financial interests in any of the findings of the study.

ORCID iD

Linda M Platas

References

Abuya

Hungi

Kitsao-Wekulo

, et al. (2018) Advancing Learning Outcomes and Leadership Skills. Briefing Paper. Nairobi, Kenya: African Population and Health Research Center. Available at http://aphrc.org/wp-content/uploads/2018/10/ALOTChange-Midline-Briefing-Paper-2018_Final.pdf (accessed 7 November 2021).

Alderman

Behrman

Kohler

H-P

, et al. (2001) Attrition in longitudinal household survey data: Some tests from three developing-country samples. Demographic Research 5(4): 79–124. DOI: 10.4054/DemRes.2001.5.4.

Begi

(2014) Use of mother tongue as a language of instruction in early years of school to preserve the Kenyan culture. Journal of Education and Practice 5(3): 37–49. DOI: 10.7176/JEP.

Bethell

(2016) Mathematics Education in Sub-saharan Africa: Status, Challenges, and Opportunities. World Bank. Available at https://openknowledge.worldbank.org/handle/10986/25289 (accessed on 7 November 2021).

Claessens

Duncan

Engel

(2009) Kindergarten skills and fifth-grade achievement: evidence from the ECLS-K [Early Childhood Longitudinal Study, Kindergarten Class of 1998–1999]. Economics of Education Review 28(4): 415–427.

Dickinson

Porche

(2011) Relation between language experiences in preschool classrooms and children’s kindergarten and fourth-grade language and reading abilities. Child Development 82(3): 870–886. DOI: 10.1111/j.1467-8624.2011.01576.x.

Dowd

Bartlett

(2019) The need for speed: Interrogating the dominance of oral reading fluency in international reading efforts. Comparative Education Review 63(2): 189–212. DOI: 10.1086/702612.

Dubeck

Gove

(2015) The early grade reading assessment (EGRA): its theoretical foundation, purpose, and limitations. International Journal of Educational Development 40: 315–322. DOI: 10.1016/j.ijedudev.2014.11.004.

Dubeck

Jukes

Okello

(2012) Early primary literacy instruction in Kenya. Comparative Education Review 56(1): 48–68. DOI: 10.1086/660693

10.

Duncan

Dowsett

Claessens

, et al. (2007) School readiness and later achievement. Developmental Psychology 43(6): 1428–1446. DOI: 10.1037/0012-1649.43.6.1428

11.

Embassy of the Republic of Kenya (2019) Culture [Web page]. Washington, DC. Available at https://kenyaembassydc.org/aboutkenyaculture/(accessed 7 November 2021).

12.

Engle

Fernald

Alderman

, et al. (2011) Strategies for reducing inequalities and improving developmental outcomes for young children in low-income and middle-income countries. Lancet 378(9799): 1339–1353. DOI: 10.1016/S0140-6736(11)60889-1

13.

Foster

(ed), (2010) The value of reanalysis and replication: a consideration of school readiness [Special section]. Developmental Psychology 46(5): 973–1029. DOI: 10.1037/a0020183

14.

Fuchs

Hosp

, et al. (2001) Oral reading fluency as an indicator of reading competence: a theoretical, empirical, and historical analysis. Scientific Studies of Reading 5(3): 239–256. DOI: 10.1207/S1532799XSSR0503_3

15.

Fuchs

Geary

Fuchs

, et al. (2016) Pathways to third-grade calculation versus word-reading competence: are they more alike or different? Child Development 87(2): 558–567. DOI: 10.1111/cdev.12474

16.

Fyfe

Rittle-Johnson

Farran

(2019) Predicting success on high-stakes math tests from preschool math measures among children from low-income homes. Journal of Educational Psychology 111(3): 402–413. DOI: 10.1037/edu0000298

17.

Geary

Hoard

Nugent

, et al. (2013) Adolescents’ functional numeracy is predicted by their school entry number system knowledge. Plos One 8(1): e54651. DOI: 10.1371/journal.pone.0054651

18.

Global Education Monitoring Report Team (2017) Accountability in Education: Meeting Our Commitments. Paris: United Nations Educational, Scientific and Cultural Organization (UNESCO). Available at https://unesdoc.unesco.org/ark:/48223/pf0000259338 (accessed on 7 November 2021).

19.

Hanushek

Woessmann

(2008) The role of cognitive skills in economic development. Journal of Economic Literature 46(3): 607–668. DOI: 10.1257/jel.46.3.607

20.

Bentler

(1999) Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling 6(1): 1–55. DOI: 10.1080/10705519909540118

21.

Hungi

Njagi

Wekulo

, et al. (2018) Effect of language of instruction on learning of literacy skills among pre-primary school children from low-income urban communities in Kenya. Early Childhood Education Journal 46(2): 187–199. DOI: 10.1007/s10643-017-0850-1

22.

Ibrahim

Eviatar

Aharon-Peretz

(2007) Metalinguistic awareness and reading performance: a cross language comparison. Journal of Psycholinguistic Research 36(4): 297–317. DOI: 10.1007/s10936-006-9046-3

23.

Kenya National Bureau of Statistics (2015) Facts and Figures 2015. Nairobi: Kenya National Bureau of Statistics. Available at https://www.knbs.or.ke/?wpdmpro=kenya-facts-2015 (accessed 7 November 2021).

24.

Kwayumba

Piper

Oyanga

, et al. (2019) The Impact of the Tayari Early Childhood Development and Education Programme on the Fourth Wave of Learning Outcomes Data in a Longitudinal Tracer Study in Kenya. Prepared for the Children’s Investment Fund Foundation. Research Triangle Park, NC: RTI International.

25.

Lutfeali

Sitabkhan

Ralaingita

, et al. (2021) Designing effective numeracy programs in low- and middle-income countries. Science of Teaching. Available at https://scienceofteaching.s3.eu-west-3.amazonaws.com/index.html#/lessons/bAK8mo7rBunMbxgIGkkZHMpIaBE3uzEf (accessed on 7 November 2021).

26.

Molina Millán

Macours

(2017) Attrition in Randomized Control Trials: Using Tracking Information to Correct Ias. CEPR Discussion Paper, No. 11962.

27.

McCoy

Peet

Ezzati

, et al. (2016) Early childhood developmental status in low- and middle-income countries: national, regional, and global prevalence estimates using predictive modeling. Plos Medicine 13(6): e1002034. DOI: 10.1371/journal.pmed.1002034

28.

National Early Literacy Panel (NELP) (2008) Developing Early Literacy: Report of the National Early Literacy Panel. Washington, DC: National Institute for Literacy. Available at https://www.nichd.nih.gov/publications/product/346 (accessed 31 May 2021).

29.

Nguyen

Watts

Duncan

, et al. (2016) Which preschool mathematics competencies are most predictive of fifth grade achievement? Early Childhood Research Quarterly 36: 550–560. DOI: 10.1016/j.ecresq.2016.02.003

30.

Pace

Alper

Burchinal

, et al. (2019) Measuring success: within and cross-domain predictors of academic and social trajectories in elementary school. Early Childhood Research Quarterly 46(1): 112–125. DOI: 10.1016/j.ecresq.2018.04.001

31.

Piper

Merseth

Ngaruiya

(2018a) Scaling up early childhood development and education in a devolved setting: policy making, resource allocations, and impacts of the Tayari School Readiness Program in Kenya. Global Education Review 5(2): 47–68.

32.

Piper

Miksic

(2011) Mother tongue and reading: using early grade reading assessments to investigate language-of-instruction policy in East Africa. In: Gove

Wetterberg

(eds), The Early Grade Reading Assessment: Applications and Interventions to Improve Basic Literacy. Research Triangle Park, NC: RTI Press, pp. 139–182.

33.

Piper

Ralaingita

Akach

King

(2016) Improving procedural and conceptual mathematics outcomes: Evidence from a randomized controlled trial in Kenya. Journal of Development Effectiveness 8(3): 404–422. doi: 10.1080/19439342.2016.1149502.

34.

Piper

Sitabkhan

Nderu

(2018b) Mathematics from the beginning: evaluating the Tayari preprimary program’s impact on early mathematics skills. Global Education Review 5(3): 57–81.

35.

Platas

Ketterlin-Geller

Brombacher

, et al. (2014) Early Grade Mathematics Assessment (EGMA) Toolkit. Research Triangle Park, NC: RTI International. Available at https://shared.rti.org/content/early-grade-mathematics-assessment-egma-toolkit (accessed 31 May 2021).

36.

Purpura

Hume

Sims

, et al. (2011) Early literacy and early numeracy: the value of including early literacy skills in the prediction of numeracy development. Journal of Experimental Child Psychology 110(4): 647–658. DOI: 10.1016/j.jecp.2011.07.004

37.

Purpura

Litkowski

Knopik

(2019) Mathematics and reading develop together in young children: practical and policy considerations. Policy Insights from the Behavioral and Brain Sciences 6(1): 12–20. DOI: 10.1177/2372732218815923

38.

Purpura

Napoli

Wehrspann

, et al. (2017) Causal connections between mathematical language and mathematical knowledge: A dialogic reading intervention. Journal of Research on Educational Effectiveness 10(1): 116–137. DOI: 10.1080/19345747.2016.1204639

39.

R Core Team (2016) R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing.

40.

Rao

Richards

Sun

, et al. (2019) Early childhood education and child development in four countries in East Asia and the Pacific. Early Childhood Research Quarterly 47: 169–181. DOI: 10.1016/j.ecresq.2018.08.011

41.

RTI International (2016) Early Grade Reading Assessment (EGRA) Toolkit. Prepared for USAID under the Education Data for Decision Making (EdData II) project, Measurement and Research Support to Education Strategy Goal 1, Task Order No. AID-OAA-12-BC-00003 (RTI Task 20). 2nd edition. Research Triangle Park, NC: RTI International. Available at https://pdf.usaid.gov/pdf_docs/PA00M4TN.pdf (accessed 31 May 2021).

42.

Sarama

Clements

(2009) Early Childhood Mathematics Education Research. New York, NY: Routledge. DOI: 10.4324/9780203883785

43.

(2008) On the Anglocentricities of current reading research and practice: the perils of overreliance on an “outlier” orthography. Psychological Bulletin 134(4): 584–615. DOI: 10.1037/0033-2909.134.4.584

44.

Storch

Whitehurst

(2002) Oral language and code-related precursors to reading: Evidence from a longitudinal structural model. Developmental Psychology 38(6): 934–947.DOI: 10.1037/0012-1649.38.6.934

45.

Suggate

Schaughency

McAnally

, et al. (2018) From infancy to adolescence: the longitudinal links between vocabulary, early literacy skills, oral narrative, and reading comprehension. Cognitive Development 47: 82–95. DOI: 10.1016/j.cogdev.2018.04.005

46.

Tanner

Candland

Odden

(2015) Later Impacts of Early Childhood Interventions: A Systematic Review. IEG Working Paper No. 2015/3. Washington, DC: Independent Evaluation Group, World Bank. Available at http://ieg.worldbankgroup.org/sites/default/files/Data/Evaluation/files/ImpactsofInterventions_EarlyChildhoodonLaterOutcomes.pdf (accessed 7 November 2021).

47.

University of Nairobi, Kenya and Makerere University, Uganda (2019) Tusome and PRIMR: A Desk Review of Early Grade Reading Programs in Kenya from 2011-2019. Available at https://thedocs.worldbank.org/en/doc/364011589472853972-0090022020/render/TusomeDeskReviewFinal2.pdf (accessed 7 November 2021).

48.

Vyas

Kumaranayake

(2006) Constructing socio-economic status indices: how to use principal components analysis. Health Policy and Planning 21(6): 459–468. DOI: 10.1093/heapol/czl029

49.

Watts

Duncan

Siegler

, et al. (2014) What’s past is prologue: relations between early mathematics knowledge and high school achievement. Educational Researcher 43(7): 352–360. DOI: 10.3102/0013189X14553660

50.

Wolf

McCoy

(2019a) Household socioeconomic status and parental investments: direct and indirect relations with school readiness in Ghana. Child Development 90(1): 260–278. DOI: 10.1111/cdev.12899

51.

Wolf

McCoy

(2019b) The role of executive function and social-emotional skills in the development of literacy and numeracy during preschool: a cross-lagged longitudinal study. Developmental Science 22(4): e12800. DOI: 10.1111/desc.12800

52.

World Bank (2021a) World Bank country and lending groups [Web page]. Available at https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups (accessed 7 November 2021).

53.

World Bank (2021b) Population, total – Kenya [Web page]. Available at https://data.worldbank.org/indicator/SP.POP.TOTL?locations=KE (accessed 7 November 2021).

54.

World Bank (2021c) Urban population (% of total population) – Kenya [Web page]. Available at https://data.worldbank.org/indicator/SP.URB.TOTL.IN.ZS?locations=KE (accessed 7 November 2021).

55.

Yamagata

(2007) Differential emergence of representational systems: drawings, letters and numerals. Cognitive Development 22(2): 244–257. DOI: 10.1016/j.cogdev.2006.10.006

School-entry predictors of lower primary reading and mathematics achievement in Kenya

Abstract

Keywords

Introduction

Predictors of achievement in high-income contexts

Early mathematics skills as a predictor of later mathematics achievement

Early literacy skills as a predictor of later reading achievement

Cross-domain predictors of achievement

Predictors of achievement in low-income and LMIC

Method and sample

The context

Study design

Participants

Ethical considerations

Measures

Early grade reading assessment

Early grade mathematics assessment

Scoring of the EGRA and EGMA

Covariates

Analytic plan

Results

Discussion

Early mathematics skills as a predictor of later mathematics achievement

Early literacy as a predictor of later reading achievement

Cross-domain predictors of achievement

Limitations

Future directions

Footnotes

Declaration of conflicting interests

Funding

Data availability

ORCID iD

References