The concept of school readiness is multi-faceted, encompassing the physical health, social-emotional, cognitive and linguistic status of children. There is abundant evidence that poor children lag behind their more advantaged counterparts on most if not all aspects of readiness. The effects of poverty manifest themselves early in a childs life. Children in low-income families have higher rates of asthma and dental disease, are more vulnerable to measles and other preventable illnesses, are more likely to have behavioral or developmental problems and less likely to see a pediatrician on a regular basis or live in a safe home environment that nurtures their development (Garbarino, 1990; Dryfoos, 1987; Rosenbaum, 1992; Gelles, 1992). At the end of preschool, they score up to one standard deviation below the norms on measures of language and early math (Lee & Burkham, 2002; U.S. DHHS, 2003). These problems are not readily overcome when children enter school; low-income children are more likely to perform poorly once in school (Dryfoos, 1987), to repeat grades and have frequent absences (Ravitch & Finn, 1987), fail to complete high school and lack basic literacy and numeracy skills (Gardner, 1990).
Ideally then, any consideration of the size of the gap in school readiness, its implications and the results of efforts to close it would consider all facets of the problem. In reality, there are difficulties in attempting this for several reasons: we have not specified and quantified the extent of the gap in several of these domains, in part because we lack good measures; we do not fully understand the implications of some early problems for school success; and there have been limited efforts to address problems in some domains. This paper confines discussion to the gap in cognitive (i.e., language) achievement for several reasons: early cognitive skills are foundational for later school performance and success (Whitehurst & Lonigan, 2001; Duncan et al., 2007) making the discussion substantively important; characteristically (though not always) they are measured using standardized tests with norms that allow us to examine and describe the data in different ways; and, to date, most evaluations of early childhood interventions have measured effects in these areas.
Using several large national data sets, researchers have both identified the gap in school readiness in terms of pre-literacy skills (as well as math and cognitive skills) and quantified it. Lee and Burkham, in their analysis of data from the Early Childhood Longitudinal Study Kindergarten cohort (ECLS-K), found that low-SES children scored almost half a standard deviation (.47) lower on the reading skills test at the start of kindergarten than their middle-class counterparts, and 1.17 standard deviations below high-SES children (Lee & Burkham, 2002). Analysis of data from the Family and Child Experiences Survey (FACES), administered when a child enters Head Start, found a gap of a standard deviation (1.0) in the pre-reading and math skills of Head Start children compared with national norms (U.S. DHHS, 2003).
Why should we be concerned about this gap? As the earlier discussion showed, a number of researchers have demonstrated the implications of the gap for future school achievement. Recently, Duncan and his colleagues, analyzing data from more than 35,000 preschoolers in Canada and the United States, found that language and literacy skills at the start of school along with early math skills were the most important predictors of later school success (Duncan et al., 2007). Children who enter school with many fewer of the necessary skills than their more advantaged counterparts do not catch up.
For this paper we used data from the ECLS-K to graphically illustrate the different trajectories traveled by the general population of children and those whose families lived in poverty at the time their children entered kindergarten and were in poverty at one or more measurement points before the end of fifth grade. Exhibit 1 shows, from the start of kindergarten through the spring of grade five, children demonstrate steady growth in their reading abilities. Children experiencing repeated poverty (i.e., children whose families are below the federal poverty line at baseline and at least one subsequent measurement point), however, follow a growth trajectory that is somewhat below the average trajectory for all children. That is, the difference in the reading scores increases over time, with children living in continued poverty falling farther and farther behind. These analyses suggest, consistent with previous studies, children living in continued poverty have a slightly slower rate of achievement in reading scores than the average of all children.
Growth Trajectories of Children in ECLS-K Full Sample and Children in Continuing Poverty*
Mean at Each Measurement Point for Full Sample (top black) and
Children in Repeat Poverty (bottom orange)
* Children whose families are in repeated poverty
(poverty at the time of the Fall K test and poverty at one or more subsequent measurement points).
Children in continuing poverty are defined as children whose families are in poverty at the time of the fall kindergarten test and in poverty at one or more subsequent measurement points.
Exhibit 1 shows the growth in children's average reading IRT scores from the fall of kindergarten through the spring of grade 5. As shown in the X-axis, reading IRT scores were measured in the fall of kindergarten (baseline), spring of kindergarten (9 months later), spring of grade 1, (21 months after baseline), spring of grade 3 (45 months after baseline), and spring of grade 5 (69 months after baseline). The range of possible reading IRT scores shown on the Y-axis is 0 points to 150 points. From baseline through the spring of grade 5, average reading IRT scores for the full sample begin at approximately 25 points at baseline and increase to approximately 40 points in the spring of kindergarten, to approximately 75 points in the spring of grade 1, to approximately 120 points in the spring of grade 3, and to approximately 140 points in the spring of grade 5. The average trajectory for children in continuing poverty follows a similar pattern, but is somewhat below the average trajectory for the full sample and has a slightly slower rate of growth. Specifically, the average reading IRT scores for children in continuing poverty begin at approximately 20 points at baseline and increase to approximately 28 points in the spring of kindergarten, to approximately 60 points in the spring of grade 1, to approximately 100 points in the spring of grade 3, and to approximately 120 points in the spring of grade 5.
Ways to quantify the gap in readiness
Typically, as we have seen, achievement differences within a population and across populations are described in terms of the portion of a standard deviation (a measure of the variation in scores within a group) that the difference represents. Increasingly, the impacts of interventions designed to reduce those differences are expressed in effect sizes. Effect sizes are standardized measures of the magnitude (size) of treatment effects. For each measure, the effect size is equal to the estimated impact of the treatment, divided by the control group standard deviation. The standardization makes possible a comparison of the size of treatment effects across studies and, within limits, across outcome measures. For example, if the effect sizes of a treatment on outcome measures A and B are 0.50, and 0.25, respectively, then the size of the treatment impact on A is considered to be twice the size of the impact on B. This strategy also allows us to estimate how much of a gap in achievement has been narrowed by an intervention. For example, if the initial gap is 0.5 of a standard deviation, then an intervention that has an impact of 0.25 (to use the example given above) has succeeded in halving the gap. This is an important aid to our understanding of the effectiveness of an intervention.
However, for policymakers and other members of the interested public, standard deviations and effect sizes are quite abstract measures and offer no guidance about the educational or developmental meaning of the gap (or of the impact of an intervention). To try to address this problem, we developed a different metric for the school readiness gap, one that might help policymakers understand both the magnitude of the gap and the effectiveness of efforts to close it.
To do this, we identified some large data sets that included low-income preschoolers assessed with nationally-normed language/pre-reading tests. The first data set came from two evaluations of the Comprehensive Child Development Program (CCDP), in which almost 8,000 low-income children below the age of five were assessed repeatedly using the Peabody Picture Vocabulary Test (PPVT-R) and its Spanish equivalent (TVIP). The second data set came from the first evaluation of Even Start, in which more than 6,000 low-income preschool children were assessed using the PPVT-R and TVIP at several points in time. Since both evaluations showed that the programs had no effects on any of the outcomes measured, we used scores from both treatment and control group children. The readily available data are in the form of standard score equivalents. These were converted to raw scores using Table 1 of the PPVT-R Manual. We then converted the raw scores into age-equivalent scores for both the study samples and the norming samples, and assessed the difference in months of development between the low-income samples and the norming sample (which, of course, included some low-income children). Exhibit 2 presents the findings of those assessments.
For the CCDP sample of children, whose family incomes were substantially below the Federal Poverty Level (FPL) when they entered the study and moved only slightly upward over time, three-year-olds in the study sample lagged behind the norms by almost nine months. The gap continued to widen with age: the gap at four years of age was almost 14 months and before six years of age, the gap was almost 15 months. Findings from the analysis of data from the Even Start evaluation showed a similar gap in achievement 8 months for three-year-olds, increasing to 12 months for four-year-olds. The income criteria for participation in Even Start were not as uniform as those used in CCDP, so the income range is probably slightly wider. However, when we looked at scores for children in the lowest income group (less than 50% of the FPL), these showed an even wider gap in achievement as the children prepared to enter kindergarten almost 18 months.
Size of the Language Achievement Gap
in Two Samples of Low-Income Preschool-age Children
|Age span of children
|3 years to 3 years, 11 months
|4 years to 4 years, 11 months
|5 years to 5 years, 9 months
|3 years to 3 years, 11 months
|4 years to 4 years, 11 months
While both of these were large samples of low-income children drawn from many different parts of the country, neither is representative of children from low-income families nationally. The same analysis could be performed with data from the ECLS-K, for a nationally-representative sample. However, because the ECLS-K chose to use a compendium of subtests drawn from several existing measures rather than a single standardized test, standardization of the measure and creation of age-based norms would be a time-consuming undertaking that is beyond the scope of this project.
These analyses and descriptions assume that the gap to be closed is between the achievement of children from low-income families and the average child as they enter kindergarten. Although it may not have been quantified in these terms before, the gap in school readiness associated with socioeconomic status has certainly been recognized by policymakers at all levels of government for many decades. Considering it here in terms of childrens months of development gives new perspective to the weight and scale of the problem, the need for effective solutions, and the tremendous challenge these disparities pose for schools and social programs serving low-income families.