ARC LAB · Data Explorer
Advancing Research in Corrections · UNO School of Criminology & Criminal Justice
Selection and Facilitation

Do kids with psychopathic traits select into gangs — or do gangs facilitate psychopathic traits?

A seven-year test of the gang membership–psychopathic traits link, following 1,315 serious adolescent offenders through the Pathways to Desistance study. The answer, it turns out, is not either/or.

3.30×
Higher odds of gang membership per 1‑SD increase in psychopathic traits (selection)
+0.12 SD
Higher psychopathic traits during active gang membership than non-active periods (facilitation)
22.1%
Of the sample reported gang membership at least once over the panel
1,315
Adolescents tracked across 10 follow-up waves in Philadelphia and Phoenix
FIGURE 1

Three ways to explain the link

Gang scholarship has long argued three different stories about psychopathic traits and gang membership. Toggle between them to see how each predicts psychopathic traits should move around the onset and end of gang membership.

Selection, facilitation, and enhancement trajectories Illustrative trajectories adapted from Tostlebe and Pyrooz (2022): selection predicts a flat, elevated line; facilitation predicts a rise at gang onset and a fall at discontinuation; enhancement predicts a rise at onset that only partially falls at discontinuation, settling above the original baseline. A draggable marker lets you scrub through time. Gang onset Gang discontinuation ← earlier later → psychopathic traits before
Drag the dot along the curve to see how each model predicts psychopathic traits move around gang onset and discontinuation.
Selection — pre-existing, stays flat Facilitation — rises with gang, falls back to baseline Enhancement — rises with gang, settles above baseline
THE DATA

Pathways to Desistance

A longitudinal study of 1,354 serious adolescent offenders in Philadelphia, PA and Phoenix, AZ, interviewed at baseline and up to ten follow-up waves across seven years. After requiring at least three valid waves per respondent, the analytic sample is 1,315 adolescents.

50 items
Youth Psychopathic Traits Inventory (YPI), spanning interpersonal, affective, and behavioral domains
10 waves
6-month interviews for 3 years, then annual interviews, over a 7-year follow-up window
63.4%
Cumulative retention rate at the 84-month interview point
Hybrid panel model
Fixed/random-effects models separate between-person differences (selection) from within-person change (facilitation)
H1 · SELECTION

Individuals who score higher on psychopathic traits are more likely to be gang members than individuals who score lower — a between-individual difference.

Supported
H2 · FACILITATION

Psychopathic traits are higher during active periods of gang membership compared to non-active periods — a within-individual change.

Supported

A note on scoring: the YPI figures here use prorated (person-mean imputed) scores, not Pathways' raw public summated scores — the same scoring approach examined and validated in a companion paper, Tostlebe (2026), Know Your (Secondary) Data Inside and Out, Crime & Delinquency, which found the prorated and summated approaches produce nearly identical conclusions.

FIGURE 2

Testing selection: who joins gangs?

Across every domain of psychopathic traits, adolescents who scored higher were more likely to become gang members the following wave — even after accounting for neighborhood disorder, victimization, living arrangement, and ethnoracial identity.

Odds ratio, per 1-SD increase

Likelihood of gang membership in the following wave — click a bar or a chip above

What that looks like in practice

Predicted probability of gang membership, overall psychopathic traits (as published)
3.2%
1 SD below mean psychopathic traits

~2.7×higher
8.6%
1 SD above mean psychopathic traits
View the domain-level regression coefficients
DomainCoefficient (b)Odds ratioSignificance
Overall psychopathic traits1.193.30p < .001
Interpersonal (grandiose–manipulative)0.782.19p < .01
Affective (callous–unemotional)1.424.14p < .001
Behavioral (impulsive–irresponsible)1.153.17p < .001
FIGURE 3

Testing facilitation: does the gang change you?

Holding each person's own baseline constant, psychopathic traits run higher during the waves someone is an active gang member than during the waves they are not — for three of four domains.

Within-person change (SD units), active vs. non-active gang periods

Same person, compared to their own average — linked to the domain picker in Figure 2 above
Statistically significant (p < .05) Not statistically significant

The one exception

Interpersonal traits — charm, manipulation, grandiosity — did not significantly increase during active gang membership (p = .096). The author suggests these traits may take longer to develop, or may concentrate more in gang leaders than in the typical member.

FIGURE 4

Putting it together: the enhancement model

Selection and facilitation are not competing explanations — the data support both at once. Adolescents with elevated psychopathic traits are more likely to join gangs in the first place, and gang membership then reinforces and intensifies those traits while it's active.

FINDING  Psychopathic traits function as both a source and a consequence of gang membership — consistent with the enhancement model adapted from Thornberry and colleagues' (1993) selection-facilitation-enhancement framework for gang offending.

A LIMITATION WORTH FLAGGING

The study rarely caught the moment of joining or leaving

Figure 1's before/after trajectories are the theoretical prediction — but the author is upfront that the panel itself observed few of those actual transitions. Most participants who ever reported gang membership had already joined before the study began, and few joined or left during the seven-year follow-up window. That's why the within-individual test compares active vs. non-active waves, rather than a clean pre-gang-to-post-gang window.

Wave-by-wave gang status, 291 respondents who ever reported gang membership

Illustrative reconstruction of the paper's Appendix C — approximate, for shape not exact counts
Wave-based patterns of gang membership Reconstruction of Appendix C: the number of respondents who are future gang members, who just joined, who continue prior membership, who just left, or who are former members, across ten follow-up waves. Continued membership starts high and declines; former membership starts low and rises to become the largest group by wave ten, while joining and leaving stay comparatively rare throughout. 200 100 0 Follow-up wave (1 → 10)

Why it matters

Because so few people transitioned into or out of gangs mid-study, the "facilitation" test in Figures 1 and 3 leans on comparing active-membership waves to non-active waves for the same person — a good proxy, but not a direct before/after comparison. The author flags this explicitly as a reason to treat the enhancement finding as preliminary, and calls for future panels that begin tracking youth earlier, around age 10, before most gang joining happens.

WHAT THIS MEANS

Implications for practice

Because psychopathic traits are malleable rather than fixed, the strongest response is a combination strategy that works at both the individual and group level.