Impact of Family, School and Classroom Factors on Children's Achievement

Annie Georges, Columbia University

Objective Using the Early Childhood Longitudinal Study, Kindergarten Cohort (ECLS-K) data and multi-level fixed effects and random effects model, the study examines how parental income, parental investments and classroom instructional practices affect mathematics proficiencies during kindergarten. Background The literature indicates that family income matters for children’s school readiness and academic achievement after the transition to school. In addition, parents use their income to invest in after-school activities, and home educational resources. Previous research indicates that these investments have a positive effect on achievement. Besides the investments in material resources, there are other strategies that rely more on positive emotional rewards, such as expectation for school achievement. These non-monetary strategies also promote better learning skills, particularly among low-income children in high-risk neighborhoods. However, whether the effects of income vary as children progress through elementary school; or whether the school environment reinforces the effects of income is not well established empirically. Learning not only takes place in and is affected by the family, it takes place in the organizational settings of schools and classrooms. Classroom instructional time and method can most directly influence how children learn. The various classroom practices to convey information to students include: drill and practice, critical thinking, individualization of instruction, and collaborative learning. Drill and practice emphasize learning through memorization or through solving similar types of problems. Critical thinking skills emphasize application of concepts to problems or by providing examples of the concepts learned. Individualization of instruction incorporates students’ understanding and experience in the teachers’ approach to instruction. Collaborative learning promotes interaction, but de-emphasizes that all strategies evolve from the teacher. Qualitative analyses show that children whose teachers use instructional practice that builds on children’s own strategies – such as critical thinking, individualization of instruction, and collaborative learning – to solve math problems are more likely to recall number facts, and these children exceed in problem solving abilities compared to children whose teachers did not use this type of instructional practice. However, these analyses are based on small-scale population, and the results may not be generalizable to all children. Since parents’ choice may depend on their preferences for a set of school characteristics, such as smaller classes, more experienced teachers, better instructional practices, it is the more educated parents with high income that might have an advantage in selecting their preferred schools. Therefore, it is important to differentiate whether the source of the influences is more strongly associated with family and home factors or with the school and classroom factors in order to select the appropriate instruments to design effective educational policies. Hypotheses Mathematics proficiencies will be lower for low-income children during kindergarten compared to children in high-income families due to compounded socioeconomic disadvantages other than income. Given that investment is a function of income, we hypothesize that low-income children will be in homes with lower quality educational investment. It is the lower quality investment that reduces the level of mathematics proficiencies of low-income children. Low-income children who are exposed to instructional practices that promote critical thinking in the early stages of their kindergarten year will be more proficient in mathematics than other low-income children who are not exposed to these types of instructional practices. However, whether their math proficiencies will be similar to that of high-income children is ambiguous. Research Design The sample consists of 21,075 children from different socioeconomic backgrounds, and racial and ethnic groups, who are surveyed in the base year of the ECLS-K (fall and spring kindergarten). These children are in 752 public and 249 private kindergarten programs. The average age of children in the sample is 7 years; and 51% of the sample is male. Fifty-five percent of the sample is White, non-Hispanic; 15% is Black, non-Hispanic; 18.0% is Hispanic; 6.4% is Asian; 2.8% is Native Hawaiian, other Pacific Islander, American Indian, or Alaska Native; and 2.4% is more than one race. Eighteen percent of the children live in household with income below the poverty level. The outcome variables are the probability scores in addition and subtraction, and in multiplication and division. These scores are based on an Item Response Theory (IRT) model, which reflects the probability that a child would have passed a proficiency level. The scores are continuous and take on values between zero and one. Each math outcome is examined separately. Family income is measured as income-to-needs, which is a ratio that compares the family’s income to the federal poverty level for a family of a given size. Measures of socioeconomic disadvantages include adequacy of food, frequency of residential moves during the school year; adequacy of medical care indicated by frequency of visit to a doctor for routine health visit, and access to health insurance are included. Classroom instructional practices include measures for active learning instructional strategy, which is derived from the frequency with which teachers report that children explain how a math problem is solved, solve math problems in small group or with a partner, work on math problems for which there are several appropriate methods or solutions, work on math problems that reflect real-life situations. Conventional strategy is a scale score based on the frequency with which teachers report that children use math worksheets, do math problems from their textbooks, complete math problems on the chalkboard, do worksheet or workbook page emphasizing route practice or drill. Another measure indexes content difficulty. Teachers who spend more class time on difficult content topics (e.g., operations with fraction, solving word problem with an unknown) receive a higher score than teachers who spend more class time on less difficult content topics (e.g., measurement, rounding, solving problem with addition and subtraction). Parental investment is measured through purchase of educational resources for the home, and for after-school activities. We also include other type of investments reflected in the parent’s aspiration for the child and the parent’s belief of the child’s educational ability. Other variables include: child’s gender, parent’s education, race and ethnicity, employment, receipt of public assistance, family structure, siblings in the household, school demographics characteristics, teacher characteristics, and child’s health status. We use multi-level fixed effects and random effects as the estimation procedure. The fixed effects of the growth model isolate the influence of family income, parental investments and instructional practices. The random effects procedure adjusts for the presence of heterogeneity in the error variance-covariance matrix between students. Also, teacher selection of instructional practices might be endogenous to their expectations about children’s achievement and learning ability. Conventional statistical methods cannot generate consistent estimate of causal effect if this type of endogeneity exists. Test for both selection bias and un-measured covariates to evaluate the robustness of the results are performed.