Our treatment cohorts, which we sometimes refer to as “entering cohorts,” arise within our baseline sample of high school graduates (as described above) when high school graduates in our baseline sample enroll in public institutions of higher education in Texas.

Like our baseline sample of high school graduates, our treatment cohorts are subject to certain unavoidable limitations because of deficits in the relevant data files and subject to certain choices by us to exclude students.

As is the case when we make decisions to exclude high school graduates from our baseline sample, when we make subsequent decisions to exclude students from our treatment cohorts, we judiciously balance our interest in producing estimates that are as accurate as possible for the students we study (i.e., maximizing “internal validity”) with our interest in studying as many students as possible so that our findings generalize (i.e., maximizing “external validity”).

Below we describe the limitations of, and the restrictions we impose on, our treatment cohorts of entering students.
 

Rule 1: Exclude Entering Students Under 20 if They Have No Test Scores

We exclude from our treatment cohorts entering students who are under 20 years of age at the time of entry and for whom we do not have high school test score data. Entering students who are under 20 years of age at the time of entry do not have reliable earnings histories so we require them to have high school test scores as part of the data that we use to match them to comparison group individuals.

As we explain in the Comparison Groups section of this website, we also apply this rule to individuals in our comparison groups who are matched with entering students under 20 years of age.
 

Rule 2: Exclude Entering Students with No Reported Earnings Late in the Relevant Follow-up Period

From our treatment cohorts, we remove entering students with zero reported earnings in the latter portion of their follow-up period, after we generally expect them to have exited postsecondary education and (re)entered the workforce.

We define “latter portion of the follow-up period” as:

• Years 7–15 post-entry for bachelor’s degree-seeking students
• Years 4–10 post-entry for associate’s degree-seeking students
• Years 2–5 post-entry for certificate-seeking students

It is our view that, if entering students have zero reported earnings in all quarters in the latter portion of their follow-up period (when they typically have exited the institution), then they have a heightened likelihood of having moved out of Texas and of having unobserved earnings out of state (in which case including them in our sample would create error).

As we explain in the Comparison Groups section of this website, we also apply this rule to individuals in our comparison groups.

We recognize that excluding entering students with zero reported earnings in the latter portion of their follow-up period from our treatment cohorts might also lead us to exclude entering students who remain in Texas or are simply unemployed for the full follow-up period and should therefore be included in our analysis.
 

Rule 3: Exclude Entering Students with No High-quality Matches to Comparison Group Individuals

From our treatment cohorts, we judiciously exclude entering students for whom we cannot locate suitable, adequately similar comparison group individuals.

For example, we sometimes exclude recent high school graduates with high standardized test scores from treatment cohorts at selective four-year institutions because we cannot identify comparison group individuals who, among other traits, had high test scores in high school but who did not seek postsecondary education after high school.
 

Rule 4: Exclude Entering Students if They Enrolled Previously in the Same Institution

We exclude entering students from treatment cohorts if they previously enrolled in the institution in question.

We make this decision because we want to avoid conflating the effects on cumulative net VAE of sequential enrollment decisions in the same institution.

We allow students to be members of treatment cohorts at different institutions.
 

Rule 5: Exclude Entire Cohorts of Entering Students if Cohort Size is <50

For methodological reasons and to avoid measurement problems that arise with small samples, we exclude from our study entire treatment cohorts if they have fewer than 50 students.

Number of Institutions

We study 86 distinct public colleges and universities in Texas, including:

• 29 institutions with cohorts of bachelor’s degree-seeking students
• 57 institutions with cohorts of associate’s degree-seeking students
• 57 institutions with cohorts of certificate-seeking students

Number of Cohorts

We study 8,495 discrete entering cohorts, including: 

• 907 institutional cohorts
• 2,825 programmatic cohorts
• 4,763 demographic cohorts

We often study pooled cohorts that are composed of students from multiple institutions and/or from multiple entry years. We never pool students who pursue different degree types (i.e., certificate-seekers are only ever pooled with other certificate-seekers, associate’s degree-seekers are only ever pooled with other associate’s degree-seekers).

Data on all 8,495 cohorts is available for download on this site.

Number of Entry Years

We study cohorts of students who entered public colleges and universities in Texas from 2008-09 through 2018-19, including:

• Cohorts of bachelor’s degree-seeking students who entered in 2008-09, for whom we have findings for the full 15-year follow-up period.
• Cohorts of bachelor’s degree-seeking students who entered from 2009-10 through 2013-14, for whom we have results for the first 10-14 years of the full 15-year follow-up period. We sometimes refer to these cohorts as “immature” cohorts.
• Cohorts of associate’s degree-seeking students who entered from 2008-09 through 2013-14.
• Cohorts of certificate-seeking students who entered from 2008-09 through 2018-19.

Number of Students

We study 935,767 students, including:

• 309,213 bachelor’s degree-seeking students (28,614 of whom are in the 2008-09 entering cohort on which we have data for the full 15-year follow-up period and 280,599 of whom are in the 2009-10 through 2013-14 cohorts on which we have 10-14 years of data)
• 559,068 associate’s degree-seeking students
• 67,486 certificate-seeking students