Research indicates that socio-emotional skills affect important long-term outcomes.  However, despite widespread worries about possible detrimental effects, few rigorous quantitative studies examine the effects of No Child Left Behind on socio-emotional outcomes.  This paper aims to address this gap in the literature.  Using quasi-experimental methods and a nationally representative dataset, I explore the following research questions:

• What is the effect of NCLB on children’s socio-emotional outcomes (including internalizing/test anxiety, attention/behavior, and interest/self-competence in math and reading)?
• Are there different effects across student subgroups as defined by race/ethnicity, gender, socio-economic status, and initial achievement quartile in reading and math?

Critics of high-stakes testing anticipate negative effects on children’s socio-emotional outcomes.  For example, many worry that schools “teach to the test” in a rote manner, leading to student disengagement, or that students feel more test anxiety. On the other hand, high-stakes testing could lead to improvements such as increased student self-competence in tested subjects due to more intensive or better instruction in these subjects. One rigorous causal study examines the effects of test pressure due to NCLB on socio-emotional outcomes (using the same dataset as this paper, but a different methodology and fewer socio-emotional outcomes).  It finds a positive effect on math interest, with no significant effects on reading interest or test anxiety (Reback et al., 2011). 

Methodology

This paper exploits the fact that some states already had accountability with strong consequences for schools before No Child Left Behind went into effect.  Thus, we assume that schools in states without prior consequential accountability experienced a much greater increase in test pressure after NCLB.  These are the “treatment” states, while the states with prior accountability are the “control” states.  With data from the Early Childhood Longitudinal Survey – Kindergarten Cohort, I use a difference-in-differences model with state fixed effects, across the years 2001-02 and 2003-04, to identify the effect of the introduction of high-stakes testing on socio-emotional outcomes.  I also run the model with additional interaction terms to detect differential effects for student subgroups. 

Results/Implications

I find no statistically significant effects for children overall. At the same time, I find substantial differences between student subgroups.  For example, students in the lowest reading quartile in third grade report a positive effect on reading interest/self-competence (.19 standard deviations) not shared by other students.  Hispanic students report greater math interest (.25), lower attention and behavior problems (.25 and .21, respectively) and lower internalizing behaviors (.32) such as worrying about school and tests as a result of NCLB.  However, black students experience particularly adverse effects of NCLB, such as worse effects on externalizing behavior than other students as well as negative effects on interest/self-competence in math and school in general.

Parents, schools, and policy-makers can take comfort from the fact that students are not showing signs of harm on average.  However, more research is needed on possible mechanisms leading to differences in subgroup effects in order to help schools and policy-makers better address the socio-emotional needs of all their students in the face of high-stakes testing.