For instance, Conger, Conger, and Elder (1997) found that greater levels of family income were associated with higher GPA's in grade 10. Further, the findings indicated that adolescents' experiences with poverty here also predictive of their grades in school in this study. For instance, Conger, et al. (1997) found that adolescents who were poor during some point in the four year study period had lower GPA's in 10th grade than those who were not poor during any point of the study. More recent poverty was associated with 10th grade GPA. However, being in poverty early in the four year study or throughout the entire four year study period was not associated with 10th grade GPA. Having a higher family income was predictive of having a higher GPA. Income-to-needs ratios greater than three were also positively associated with GPA at grade 10. This means that children in families whose incomes were at least three times as great as the poverty line (averaged over the four years of the study) had higher levels of achievement. Additional longitudinal evidence on the importance of recent poverty on adolescent achievement was found by in a national data set with a racially diverse sample, the National Longitudinal Survey of Youth (NLSY) and NLSY child sample. Using a mixed linear model, Guo (1998) examined the influences of timing of poverty on adolescents' achievement outcomes for approximately 922 children were given reading recognition, comprehension, and math tests between ages 5 and 8, and again when they were early adolescents, between the ages of 11 and 14. Early poverty was defined as being in poverty during the four years before turning six, and late poverty was defined as being in poverty during the four years prior to the early adolescent follow-up. Early cumulative poverty was defined as being in poverty during their first six years and the late cumulative poverty was defined as being in poverty throughout their life until the early adolescent follow-up. Other measures of income included income-to-needs ratio and total family income. Guo (1998) found that early cumulative poverty, late cumulative poverty, and late poverty predict lower scores on adolescents' achievement tests. Long-term average income-to-needs ratio and long-term family income also predicted lower achievement test scores. However, later poverty was found to be more detrimental to adolescent's achievement, predicting even lower scores, than childhood poverty. These findings remained after accounting for child characteristics and family background factors, including prior achievement, race, age, sex, maternal cognitive ability, maternal education, birth weight, birth order, number of siblings, maternal age, family structure, region, and urban residence.