ABSTRACT
What causes the literacy gap and can schools compensate for it? The authors investigated 3 drivers of the gap: preliteracy knowledge, schooling, and the summer vacation. Longitudinal literacy data over 5 time points were collected on 126 five-year-olds attending higher or lower socioeconomic status (SES) schools during their first 15 months of school. There were several noteworthy findings: (a) gaps in preliteracy knowledge at school entry favor higher SES schools, (b) preliteracy knowledge predicted later progress over and above SES and gender, (c) during the school year there was a widening of the gap between higher SES schools and lower SES schools in reading and spelling skills, and (d) children attending lower SES schools exhibited losses during summer whereas children attending higher SES schools nearly always gained. Contrary to previous studies, the present results indicated that when there are concentrations of children from higher and lower SES in schools located in the children's respective SES areas, the achievement gap widens.
Acknowledgments
The manuscript is based on the Shanthi Tiruchittampalam's doctoral dissertation at Massey University, New Zealand. Special thanks to the schools, parents, and children who enabled the study to take place.
Notes
1. As can be seen in , on most measures the two SES groups' standard deviations are quite different, which may render suspect statistical conclusions based on a standard (Student) two-sample t test that assumes variance homogeneity. However, because the two SES groups' sample sizes are large (n = 63) and equal, here and throughout the manuscript the statistical results based on both the standard t test and a separate-variance Welch-Aspin t test are identical. In addition, (a) because the overwhelming majority of scores (96%) on the WRAT-Spelling test were 0, an additional chi-square test of homogeneity was conducted and corroborated the statistically nonsignificant conclusion of the t test; and (b) because all of the lower SES children received scores of 0 on the test of basic decoding skills (with a resulting standard deviation of 0), an analogous nonparametric Mann-Whitney rank test was conducted instead and produced a result favoring the higher SES children (p = .007).
2. For a discussion of Matthew effects in an academic achievement context, see Pforst, Hattie, Dorfler, & Artelt, (Citation2014) and Stanovich (Citation1986).
3. Note that despite this descriptively longer T1–T2 in school testing interval for the lower SES students (and thereby benefitting those students) all seven β_6j values in are negative (with four of them being statistically significant), which in turn indicates a widening of the SES achievement gap with respect to schooling effects.
4. It should be kept in mind that these measures would have been too difficult for many of the children at earlier points in the study.