Background/Context: School readiness refers to the foundational skills and behaviors children leverage as they begin kindergarten, including a combination of early academics, executive function (EF), and social-emotional competencies. In both high-income and low- and middle-income countries, these early skills collectively enable children to achieve later school success (Duncan et al., 2007; La Paro & Pianta, 2000; Wolf & McCoy, 2019). This prediction likely reflects a combination of the causal effects of these early skills on later academic competencies, but also unmeasured, stable characteristics affecting learning throughout development (e.g., other domain-general cognitive skills, motivation, environmental factors; Bailey et al., 2018, 2020). School readiness skills are interconnected and co-develop throughout early and middle childhood, with well documented links between reading and math (e.g., HŸbner et al., 2022), and EF and academic skills (e.g., Spiegel et al., 2021). A smaller body of work demonstrates the relations between EF and social-emotional skills (e.g., Burchinal et al., 2020), although findings are mixed and less consistent between social skills and academic performance. Models like the Random Intercept Cross-Lagged Panel Model (RI-CLPM), which account for stable confounders, offer better prediction for intervention impact patterns over subsequent years, and sometimes lead to theoretically distinct conclusions (Bailey et al., 2020). Whether skill-building dynamics are universal across contexts remains an open question. On the one hand, they may operate similarly in Ghana as they do elsewhere. On the other hand, children's early skills and later development are influenced directly by their learning environments, cultural norms, and broader educational systems in which they are situated (Greenfield, 2016), which may differentially shape how children acquire skills over time. Purpose/Objective/Research Question: The current study draws on the longitudinal Quality Preschool for Ghana dataset that assesses children's early skills over six years. We aim to understand how individual differences in children's learning in late primary school co-develop, and how inferences about their reciprocal effects depend on modeling decisions, specifically whether models allow for the potential role of stable factors influencing children's skills throughout this developmental period. We test two primary questions: (1) What are the associations between early skills in preschool with later primary school outcomes? (2) How do the associations among the co-development of early skills change when accounting for stable confounding factors using a random intercept during early and middle childhood? We expect preschool academic skills to be the strongest predictors of middle childhood outcomes, but literacy would be more important than numeracy in the models that most strongly adjust for confounding. Setting: The current study takes place in Ghana, a lower-middle-income country in West Africa. At the first wave when schools and children were sampled, participating schools were in six of the sixteen districts in the Greater Accra region. Participants: Our sample is drawn from a 2015-2016 school-randomized trial evaluating an intervention of teacher training and parental awareness programs aimed at enhancing the quality of preprimary education and promoting children's school readiness (see Wolf et al., 2019). Data used in this study were collected from 2,012 children at five different time points: fall of pre-kindergarten and kindergarten (Mage=5.2), spring of pre-kindergarten and kindergarten (Mage=5.8), kindergarten and first grade (Mage=6.8), first and second grade (Mage=7.5), and fourth and fifth grade (Mage=10.9). Child assessments were administered by trained Ghanaian assessors at the child's school. Among participating children, by the final wave, 49.7% were female, 40.4% attended Primary 4 and 41.9% or Primary 5, and 55.6% attended public schools. Measures: All measures used in this study have been previously used in research in the Sub-Saharan African region. In kindergarten and first grade, early numeracy, early literacy, EF, social competence skills, and approaches to learning were assessed via the International Development and Early Learning Assessment (Pisani et al., 2018). In first and second grade, literacy and numeracy were assessed using the Early Grade Reading Assessment (EGRA; RTI International, 2015) and Early Grade Math Assessment (EGMA; RTI International, 2014), respectively. In fourth and fifth grade, literacy and numeracy/math were assessed using a combination of EGRA and EGMA and select items from the Young Lives School Survey (Cueto & Le--n, 2012). Research Design: The RI-CLPM employs a latent variable modeling approach to disaggregate stable and time-varying variance (Hamaker et al., 2015). We employ the RI-CLPM model to estimate time-varying effects (via autoregressive and cross-lagged paths) of early skills across the elementary school years. Results Table 1 presents the correlation between the four skills. All preschool skills were statistically significantly correlated with one another, with magnitudes ranging from .42 to .75. Table 2 presents RI-CLPM model results for four domains across five timepoints, including factor loadings, path estimates, latent factor relation estimates, and fit statistics indicating good fit (CFI = 0.98, RMSEA = 0.05). Auto-regressive paths in magnitude were larger for math and literacy, indicating these two domains had greater time-specific dependency between two consecutive timepoints. Cross-lagged paths were larger in size for math and literacy, suggesting a time-specific change in these domains at one timepoint would predict a larger difference in the other domain skills at the subsequent timepoint. The latent estimates among math, literacy and EF were strongly correlated, indicating substantial overlap in the factors contributing to these domains during this developmental period. We observed smaller but still substantial positive latent estimates between social competency and the other domains. Our study provided different findings between the RI-CLPM and regression models. In the regression analyses (not presented), early numeracy was observed as the strongest predictor of later outcomes. Conclusions: This study examined how the relative strength of early skills and stable characteristics shape children's learning over six years of pre-primary and primary school in peri-urban Ghana. Results from the RI-CLPM suggest that literacy and math have greater time-specific dependencies, relative to EF and social competence, and that literacy and math were more predictive of other school readiness than vice versa. Our findings have important implications for how intervention programs and policies can support children's academic learning, such as by targeting early cognitive skills.