Comeback kids: School suspension and high school graduation

Students applaud as U.S. President Obama arrives to deliver the commencement address at the Worcester Technical High School graduation ceremony in Worcester

People on track to success tend, obviously enough, to succeed. They complete their education, develop essential life skills, stay out of trouble, and plan for their futures. But not everybody succeeds at every stage. How can those who go ‘off-track’ get back on? What distinguishes them from those who go off-track and struggle to get back on?

Suspended from school, but still graduating

Take children or adolescents who get suspended from school. Other things equal, suspension is a clear setback, with an impact on children’s futures, including subsequent educational achievement and, by extension, adult income.

Our analysis of a self-reported measure of suspension, available in the NLSY79 dataset, confirms this picture. High school graduation rates are markedly lower among the almost one in five respondents (19.8 percent) who said they had been suspended at some point in primary or secondary school. They were also less likely to achieve middle-income status by middle age:

Chart 1 - the "suspension penalty"

As a retrospective, self-reported measure, the NLSY suspension variable is inevitably at risk of many kinds of bias. As a binary question, it also offers no information on frequency or severity of suspension. But it is clearly associated with worse outcomes, an indication that someone is ‘off-track’ or in danger of being so.

Low-income and black students at highest risk of suspension

The risk of suspension varies by income and race. The reported suspension rate for kids from low-income families (below 200 percent of the federal poverty line, or FPL) is 26 percent, compared to 16 percent for those from families with higher incomes (above 200 percent FPL). This finding is consistent with studies documenting that low-income students are more often subject to disciplinary action.

There is a well-known race gap in school suspensions, reinforced in recent months by The Ferguson Commission report. In our sample, 19 percent of whites report a school suspension, compared to 38 percent of black respondents. School disciplinary policies and their unequal application have been linked to the school-to-prison pipeline. Any comprehensive evaluation of suspensions and their effectiveness has to address these issues of race. We are not attempting such an evaluation here.

It is difficult in our analysis to tease apart the influence of income and race, in part because of sample size issues among the group of suspendees. We can say a little, however. Even among low-income families (less than 200 percent FPL), the race gap is stark: black children are almost twice as likely as whites to be suspended (41 percent vs. 24 percent). But given our data constraints, and our attempt to focus on the factors connected with the chances of recovering from suspension, we restrict our further analysis to white respondents.

Getting back on track: Finishing high school

Suspendees who do worst in life are, unsurprisingly, those who end up with low educational qualifications. Those who make it to the middle class by middle age, despite a suspension, are much more likely to have finished high school. Indeed, 12 percent went on to get a bachelor’s degree:

Chart 2 - suspendees who make it to middle class by middle age


Chart 3 - suspendees who don't make it to middle class by middle age


Finishing high school is a vital step for suspendees and non-suspendees alike. So who manages to get a diploma? Are there differences between suspendees who get a diploma and those who do not? We look at three factors: family income, family stability, and maternal education.

The association between family income and educational attainment is well-documented. So it is no surprise to find that family income during childhood is strongly associated with the odds of achieving a high school diploma, for both suspended and non-suspended respondents:

Chart 4 - suspended, but graduating high school: family income

What about family stability? Stability and resources, often associated with family structure, seem to be critical factors. Recent research suggests thatchanges in family structure have a separate and discernable impact on child cognitive, socio-emotional, and health outcomes. We find the chances of graduating high school for all students were greater for those raised in high-stability families. In particular, suspendees who lived continuously with their biological parents until the age of 18 were 25 percentage points more likely to graduate high school:

Chart 5 - high school graduation, by family stability

Last, we examine the relationship with maternal education. Parental educational achievement influences a child’s development and later-life economic success. High school graduation rates were substantially higher for both suspendees (around 21 percentage points) and non-suspendees (around 19 percentage points) for children with a mother who had herself completed high school:

Chart 6 - high school graduation, by maternal education

A high school diploma might seem a low bar for maternal education, so it is worth noting that in 1986, when the majority of the respondents’ mothers were between the ages of 25-42, the proportion of white females above the age of 25 who had completed high school was around 75 percent.


When children and adolescents go off-track, the goal is to help them back on. We have looked at various potential factors. Our descriptive data is limited, and in particular throws precious little light on the specific experience of black Americans, who experience suspension in such disproportionate numbers. But we are able to show that certain factors are associated with a greater chance of staying on the education track after a suspension: family income, family stability, and maternal education. Note that all of these factors also have a strong association among non-suspendees. There is a general effect here, rather than one specific to the population who were suspended.

…certain factors are associated with a greater chance of staying on the education track after a suspension: family income, family stability, and maternal education.

Individuals who received a suspension are more likely to be struggling to stay on track at school. Their risk of failing to finish high school is not surprisingly elevated. But among this vulnerable group, the chances of getting back or staying on track are nonetheless greater when families provide more stability and have higher incomes, and when mothers are more educated. There are of course much stronger arguments for boosting incomes for poorer families, promoting family stability, and raising maternal education. But we might add at least the possibility that we will also be helping to ensure that children who stumble do not fall.


Technical Note on Data

We use the NLSY79, a cross-sectional sample of 6,111 nationally representative youths ages 14 to 21 as of December 31, 1978 and merge it with the 1980 survey that asks respondents whether they had experienced a suspension in their schooling history. All observations for which sufficient data was not collected to adequately measure adolescent (age 13-14) income or later-life adult income (age 40-41) were dropped. Where possible, proximity imputations were executed to provide more comprehensive data. Income is measured at the family level and given as a percentage of the federal poverty level. Ninety observations for which suspension history was unclear were dropped. There was no indication that non-responses yielded markedly different distributions of family income at adulthood or childhood or family structure compared to the global population of the full dataset. However, the suspension non-responses did record slightly lower levels of high school completion and maternal education. That said, given the relatively small number of suspension non-responses we can be encouraged that our results are not substantially influenced by such attrition. In the end we were left with a sample of 3,979 persons. We apply the NLSY79 custom weights in all of our analysis in order to help control for potential survey sampling error.