Beyond weak ties in prison: An investigation of core support networks of incarcerated persons

Abstract

Tie formation has been a core interest of network researchers. However, most existing research on tie formation in prisons has focused on weak ties, assuming that a core network approach is less applicable within total institutions such as prisons. Drawing on egocentric network data from 389 incarcerated individuals and 1352 named network members, this study demonstrates that relationships between incarcerated persons can be examined using a core support network framework. We find that about 20 % of participants include at least one incarcerated peer in their core support networks, with these ties providing both practical and emotional support comparable to that offered by other core support network members. These findings underscore the critical role that peer relationships play in coping with the stresses of incarceration. This study also offers a rare opportunity to test tie formation theories in a total institution context. Specifically, our results support theories of contact opportunities, unmet social needs, and experiential homophily in shaping decisions to include at least one incarcerated peer in core support networks. By examining the dynamics of core support networks during incarceration, this study contributes to a deeper understanding of resilience and trust formation in highly constrained, high-stress environments.

Introduction

Social scientists have long been interested in understanding whom people turn to for support during challenging times, such as poverty, illness, and the COVID-19 pandemic (Böhnke and Link, 2017; Peng and Roth, 2021; Perry and Pescosolido, 2015). Social networks play a crucial role in emotional well-being, personal growth, and resilience, offering a sense of belonging, encouragement, and stability that helps individuals navigate difficult situations (Perry et al., 2018; Thoits, 2011). However, little is known about the support systems that incarcerated individuals rely on while in prison, which are state-run facilities designed for the long-term incarceration of people who have been convicted of felonies and sentenced to more than one year. Specifically, this study examines whether individuals who are incarcerated include incarcerated peers in their core support network. From a network perspective, prisons present a unique and stark contrast to other social environments. Incarceration represents one of the most extreme disruptions to an individual’s accumulated web of social affiliations (Rocque et al., 2011; Volker et al., 2016). Prisons, as total institutions (Foucault, 1979), forcibly constrain individuals’ access to their external social networks, severely limiting contact with family, friends, and broader community ties. Beyond the reductions in access to these established connections, incarcerated individuals are thrust into a high-risk environment where misplaced trust can lead to exploitation, victimization, or even fatal consequences (Trammell, 2009). This precarious setting intensifies stress and underscores the importance of social support for coping and survival (Schaefer et al., 2021). Given these conditions, incarceration provides a valuable yet understudied context for examining how individuals organize their support networks in response to extreme social disruption.

Using egocentric network data from 389 incarcerated persons with a history of opioid use disorder, encompassing 1352 network members, from prisons in a southeastern state in the U.S., we study networks of individuals during incarceration to answer the following questions: 1) Do people who are incarcerated include at least one incarcerated peer in their core support network? 2) Who includes incarcerated peers in their core support network, and what demographic, network, and prison-related factors predict this inclusion? 3) How do ties with incarcerated peers differ from other ties in the core support network? Our focus is on the core support network, which consists of the network members on whom one relies for support. This network represents a fundamental social support system that helps individuals navigate life’s challenges (Antonucci and Akiyama, 1995; Marsden, 1987). Theoretically, this study represents a unique opportunity to test tie formation theories in the context of total institutions of which there is limited existing research (Volker et al., 2016). Specifically, we investigate the role of contact opportunities (Feld, 1981), unmet social needs (Thibaut, 1959) and homophily (McPherson et al., 2001) in shaping incarcerated individuals’ decisions to include incarcerated peers in their core support network.

Background

Research on social networks during incarceration

Mass incarceration in the United States has become a defining feature of the criminal legal system, with the country maintaining one of the highest incarceration rates in the world (Sawyer and Wagner, 2024). The costs of incarceration are multifaceted, encompassing adverse physical and mental health outcomes, elevated recidivism rates, and significant collateral effects on families, communities, and societal inequalities (Alexander, 2012; Wildeman and Wang, 2017). While research has studied incarcerated persons’ contacts with family and friends who remain in the community (Brodsky, 1975; Duwe and Clark, 2013; Holt and Miller, 1972; Western, 2007), relatively little attention has been given to whether incarcerated persons rely on incarcerated peers as part of their core support network. Studies on relationships between people who are incarcerated (e.g., Crewe, 2007; De Viggiani, 2006; Schaefer et al., 2017; Schaefer and Kreager, 2020) have rarely used a core network framework. For example, some network researchers intentionally avoid asking about friendship, arguing that people who are incarcerated are often skeptical of the concept of “friendship” with incarcerated peers. Instead, they assess positive affiliations as a proxy for friendship, asking respondents which incarcerated peers they “get along with most” (Schaefer et al., 2017, 2021; Schaefer and Kreager, 2020). While we recognize that these network scholars are mindful of the high-risk nature of prison environments, where misplaced trust can have serious consequences (De Viggiani, 2006), the reluctance to study relationships while incarcerated through a core network framework may inadvertently reinforce the social disability theory (Hansell and Wiatrowski, 1981; Hirschi, 1969). This theory argues that the friendships of individuals involved with the criminal legal system are “cold and brittle” relative to those with no criminal legal system involvement.

Even the Prison Project—a longitudinal, national Dutch study on the effects of incarceration (Volker et al., 2016)—collected core discussion network data only before and after incarceration, without identifying whether network members were incarcerated peers. The network data gathered during the incarceration period was limited to naming alters that participants “get along with most” (Kreager et al., 2016). This study design reflects a broader lack of focus on strong relationships formed within the prison context, despite qualitative evidence that such friendships exist (Bronson, 2008; Severance, 2005). One exception is a study using data from 198 individuals in a large US county jail, which found that 46 % of respondents reported having at least one close friend and 53 % reported having someone to confide in within the jail (Lindquist, 2000). However, this study relied on single-item questions to assess networks, asking, “How many close friends do you have here in jail?” and “Is there someone here in jail that you feel close enough to confide in?” rather than utilizing an egocentric network approach.

In sum, little is known about relationships between incarcerated individuals using a core network framework. The present study fills this gap by examining core support networks, assessed with the question: “Who have you been able to count on for support during last 6 months in prison?”

Theories of tie formation and selection

We aim to test three well-established theories on tie formation and selection in the context of prisons as total institutions: contact opportunities (Feld, 1981), unmet social needs (Thibaut, 1959) and homophily (McPherson et al., 2001).

Contact opportunities

Contact opportunities are crucial not only for the formation of relationships but also for their maintenance over time (Blau, 1960; Feld, 1981; Fischer, 2011; Roth and Peng, 2024). According to the focus theory (Feld, 1981), individuals come into contact with others primarily through the social settings—which can be places such as neighborhoods, workplaces, and public spaces—in which they participate. The existence of these social settings offers frequent and regular opportunities for interaction, which is vital for forming and nurturing social bonds. However, when individuals leave a particular social setting—such as moving out of a neighborhood or changing workplaces—they face increased challenges in maintaining their existing relationships. Without the frequent interactions that once occurred within that setting, the effort required to sustain these connections grows significantly, leading to a higher risk of relationship decline (Mollenhorst et al., 2014; Small and Adler, 2019). Conversely, entering a new social setting, such as moving to a different neighborhood, joining a new workplace, or even entering a prison, offers an opportunity to form new social connections. According to Feld and William (1999), a change in social setting naturally leads to the possibility of forming new ties.

This theory is particularly relevant in the context of incarcerated individuals, where changes in their social networks are likely influenced by the length of their prison sentence (Volker et al., 2016). Longer sentences may limit an individual’s ability to maintain pre-existing ties, given that contact opportunities with the outside world become more restricted. Conversely, incarceration can provide access to new social settings within the prison environment, where individuals have the chance to form relationships with incarcerated peers. Thus, we anticipate that likelihood of adding incarcerated peers into the core support network will be strongly influenced by the length of their sentences, with longer durations leading to a greater potential for the inclusion of incarcerated peers.

In addition to the length of sentence, completion of the substance use disorder (SUD) treatment program within the prison may promote the inclusion of incarcerated peers into one’s core support network. The residential SUD treatment program is housed within separate units in each prison and designed to provide an immersive, highly structured, pro-social environment using a modified therapeutic community model (De Leon, 2000; Schaefer and Kreager, 2020). The physical layout and therapy modality of the SUD treatment program is purposely designed to increase residents’ experience of community within the unit. The isolation of the SUD treatment program unit from the rest of the prison and an intensive group treatment schedule should both produce high levels of interaction between SUD treatment residents and foster honesty and self-disclosure during group discussions (Schaefer et al., 2021).

Unmet social needs

A critical factor driving the inclusion of incarcerated peers in one’s core support network may be the presence of unmet social needs (Thibaut, 1959). This theory suggests that when individuals have few network members or when their existing relationships are unsatisfying, they are more likely to seek out new social connections (Volker, Flap, and Lindenberg, 2007). Such unmet needs can be especially pronounced in the prison context. Although external communication modes—such as visitation, phone calls, and mail—are available, they often fall short in addressing the day-to-day requirements for companionship and support. In contrast, regular in-person interactions within the prison environment enable incarcerated peers to serve as vital social ties that help manage the myriad stressors of incarceration and alleviate the “pains of imprisonment” (Sykes, 2007).

In the prison context, this implies that individuals with limited core support network are more inclined to include at least one incarcerated peer into their core support network. Specifically, we will examine whether individuals with smaller or less robust core support networks—characterized by smaller network size as well as lower levels of closeness, trust, contact, or overall support—are more likely to include incarcerated peers into their core support network.

Homophily

Homophily (Kossinets and Watts, 2009; McPherson et al., 2001)— the tendency for individuals to form connections with those who share similar characteristics—is a key theory in explaining tie formation. Studies on homophily and social support have found that individuals are more likely to develop and maintain supportive relationships with others who share similar experiences (Suitor, Keeton, and Pillemer, 1995; Thoits, 2011). This is because experiential homophily increases empathetic understanding, which is crucial to the support process. In particular, others who have been through the same experience are better prepared to help stressed individuals understand and manage their feelings.

In the prison context, shared experiences such as opioid use disorder (OUD) and a history of incarceration create a unique common ground that can foster stronger bonds among individuals. Individuals who have struggled with OUD or been involved with the criminal legal system may find that their network members lack the lived experience necessary to understand and empathize with their specific challenges. Consequently, incarcerated peers, who share these experiences, may be seen as more reliable sources of emotional and practical support. This similarity may not only increase trust and mutual understanding but also facilitate more effective communication and support exchange, making the inclusion of incarcerated peers in one’s core support network a useful coping mechanism during incarceration. Therefore, we expect that incarcerated individuals with fewer network members who share experiences of OUD and incarceration will be more inclined to include incarcerated peers in their core support network.

Methods

Data

Data used in this research are part of an ongoing longitudinal observational cohort study, titled the Geographic variation in Addiction Treatment Experiences (GATE) study (Oser et al., 2023). The overall goal of the GATE study is to examine multilevel factors, including social network characteristics, influencing prison-based initiation of medications for opioid use disorder (OUD), predictors of medications for OUD utilization post-release, and adverse outcomes among rural and urban people with OUD (Oser et al., 2023). Informed consent was obtained before data collection and varied by data collection modality (i.e., written consent for in-person data collection, verbal consent for participants who completed the survey over the phone or using Zoom). Participants received $30 for the completion of each part of the two-part baseline survey. Network data was collected using Network Canvas (Birkett et al., 2021) and all other survey measures were collected using Research Electronic Data Capture (Harris et al., 2009). The GATE study was approved by the university’s Institutional Review Board (IRB #508380) and categorized as a prisoner protocol. Thus, a prisoner representative was included on the IRB review committee and additional regulatory considerations were addressed (45 CRF 46, Subpart C) as the protocol was also approved by the Office for Human Research Protections through the U.S. Department of Health and Human Services. Participants were protected by a Certificate of Confidentiality and no data was shared with the Department of Corrections.

Participants were recruited from 14 substance use disorder (SUD) treatment programs across 11 prisons in one southern state. The prison-based SUD treatment program follows a modified therapeutic community approach based on peer support to encourage responsibility and accountability during the treatment process and after community re-entry. Each month, a prison liaison provided a report containing the names of individuals who participated in the prison-based SUD treatment program and were within 90 days of release (e.g., either serving out their sentence or parole). Study eligibility was assessed by trained research staff and included: 1) participating in the prison-based SUD treatment program (completion was not required), 2) having a history of moderate or severe opioid use disorder using DSM-5 criteria, 3) being within 90 days of release from prison, and 4) having a planned release to a county within the state during the study timeframe. Screening included in-person, mail, and phone processes (due to COVID-safety protocols), and all individuals who completed the screening received $5. Staff screened 1668 individuals, and 898 met the eligibility criteria.

Baseline data collection included completing two surveys either in-person, via phone, or using Zoom due to COVID-19 restrictions, with all participants completing the second survey within three months of their release from prison. Of the 898 eligible individuals, 492 individuals consented to participate and completed both parts of the two-part baseline survey between June 2021 and March 2024. The baseline survey was divided into two parts due to the length of the survey, and covid protocols within the prison at the start of the study. Of the 492 participants, 437 (88.8 %) were released from prison and thus eligible to complete the second baseline survey. A total of 395 (90.4 %) participants completed the two-part baseline survey, which is the data used for this analysis. Three participants (0.8 %) were excluded due to missing network data, and three participants (0.8 %) were excluded due to missing demographic data. This resulted in a final analytic sample of 389 participants.

Measurement

Details on survey questions and response categories are provided in Table S1 in the supplemental materials.

Incarcerated peers in core support network

In the GATE baseline survey, participants were asked to complete an egocentric network battery. To collect egocentric network data, name generators were used to elicit names of individuals who were activated in the past 6 months without capping the total number of names that could be listed. Core support network is measured by the name generator: “Who have you been able to count on for support during last 6 months in prison?” drawing from the Arizonia Social Support Interview Schedule (Barrera, 1980), which has been used in other egocentric network studies (Bunting et al., 2025; Tillson et al., 2024). After names were provided, a series of questions called name interpreters were asked about each person in the network. These included characteristics of network members and relationships (e.g., gender, relationship strength).

To identify incarcerated peers in the core support network, we use the question on the incarceration history of network members. The response categories are 1 = Never been incarcerated after a conviction, 2 = Incarcerated after a conviction, but wasn’t in prison with me for the last 6 months of my incarceration, or 3 = In prison with me during the last 6 months of my incarceration. We identify a tie as an incarcerated peer tie if 3 is selected.

Network characteristics of participants

We created network measures to examine whether the network characteristics of an incarcerated individual are associated with including incarcerated peers in the core support network. They are divided into two groups: Measures of homophily and unmet social needs.

Two measures of homophily are included. Proportion ties ever incarcerated is calculated as the number of ties with an incarceration history divided by the total number of ties in the core support network. Proportion ties ever use illicit opioids is calculated as the number of ties that ever used illicit opioids divided by the total number of ties in the core support network.

We also include measures of unmet social needs. Core support network size is measured by the number of ties named in the core support network name generator. Proportion kin is calculated as the number of kin ties divided by the total number of ties in the core support network. Mean closeness is calculated based on the question: “How close are you to each person?” The response categories are 1 = not close at all, 2 = not very close, 3 = moderately close, 4 =very close, or 5 = extremely close. Mean contact frequency is calculated based on the question: “How often did you communicate with this person through visitation, phone, letters, or email during the last 6 months you were in prison?” The response categories are 1 = never, 2 = less than once monthly, 3 = monthly, 4 = weekly, or 5 = daily. Mean trust is calculated based on the question: “ How much do you trust each person?” The response categories are 1 = don’t trust at all, 2 = barely trust, 3 = sometimes trust, 4 = mostly trust, or 5 = trust with life. Proportion provide emotional support is calculated based on the question: “Does this person provide emotional support for you?” 0 = no or 1 = yes. Proportion do leisure activities is calculated based on the question: “Does this person participate in fun and leisure activities with you?” 0 = no or 1 = yes. Proportion provide advice is calculated based on the question: “Does this person gives you practical suggestions or advice?” 0 = no or 1 = yes.

Demographic characteristics of participants

We include two measures of contact opportunities. Time in prison is measured by the number of months in prison for their current sentence. Completion of SUD Treatment is measured as 0 = no, 1 = currently in SUD treatment, or 2 = completed SUD treatment.

We also include demographic controls to examine whether the demographic characteristics of participants are associated with including incarcerated peers in the core support network. Age is measured as years of age. Gender is measured as 0 = women, 1 = men, or 2 = other. Race is measured as 0 = non-White or 1 = White. Education is measured as 1 = less than high school, 2 = GED/High school, or 3 = more than high school. Marital status is measured as 0 = not married or 1 = currently married.

Analysis plan

To examine whether incarcerated peers are part of a participant’s core support network, we conducted ego-level analyses. These analyses included generating histograms (Fig. 1) to display the distribution of both the number and proportion of incarcerated peers included in the core support network.

Fig. 1.

Distribution of the number and proportion of incarcerated peers in the core support network.

We also conducted ego-level analyses to determine which participants rely on support from incarcerated peers. Specifically, we conducted logistic models to predict whether participants have at least one incarcerated peer or not. First, we included all ego demographic controls (i.e., age, gender, race, education, and marital status) and contact opportunities measures (i.e., time in prison and completion of the SUD treatment program) in the model.

Second, we added core support network size in the model. Ten participants reported having zero members in their core support network. Because a network size of zero perfectly predicts having zero incarcerated peers, these cases result in perfect separation. To avoid estimation issues associated with perfect separation, all network analyses were limited to participants with at least one network member.

Finally, we added all other network measures individually (i.e., proportion ties ever use illicit opioids, proportion ties ever incarcerated, proportion kin ties, mean closeness, mean contact frequency, proportion provide emotional support, proportion do leisure activities, proportion give advice, and mean trust) while adjusting for demographic variables and core support network size to examine the effect of other network characteristics. This is a common practice in network analysis to avoid multicollinearity issues, as many network measures tend to be correlated. In these models predicting whether a participant includes incarcerated peers in their network, all network measures must exclude ties to incarcerated peers. This ensures that the independent variables are not influenced by the dependent variable, thereby avoiding endogeneity issues and maintaining the validity of the analysis. For example, incarcerated peer ties (the dependent variable) are more likely to be non-kin, same-gender, and characterized by lower levels of trust and closeness—attributes that could bias the network measures (independent variables) if included. Therefore, all network measures in regression models are adjusted to exclude incarcerated peer ties.

To examine how ties with incarcerated peers differ from other ties in the core support network, we conducted tie-level analyses. Specifically, we used multilevel linear regression models to compare the difference between incarcerated peer ties and other ties across different tie characteristics and functions (i.e., closeness, contact frequency, providing emotional support, engaging in leisure activities, giving advice, trust, and ever use illicit opioids). Each tie measure was analyzed as an outcome in separate models, with an indicator for incarcerated peer ties as the key independent variable, adjusting for demographic variables including age, gender, race, education, marital status, time in prison, and completion of the SUD treatment program. We also adjusted for whether the participant had at least one incarcerated peer tie to account for differences in non-incarcerated peer ties between those who did and did not include incarcerated peers. Multilevel regression is necessary to account for the nesting of ties within each ego. We provided Fig. 2 to visualize the differences. While multilevel logistic regression is technically better suited to model binary variables (i.e., providing emotional support, engaging in leisure activities, giving advice, trust, and ever use illicit opioids), our sensitivity analyses indicate that the results from multilevel logistic regressions are consistent with those from multilevel linear regressions (See Figure S1 in the Supplemental Materials).

Fig. 2.

Differences between ties to incarcerated peers and other ties. Note: 95 % confidence interval. All tie measures are modeled individually while adjusting for age, gender, race, education, marital status, time in prison, completion of the SUD treatment program, and whether the participant had at least one incarcerated peer tie. See Table S3 in the supplemental materials for full tables.

Therefore, we chose to present the results of multilevel linear regressions for better visualization, as all coefficients are presented on the same scale.

Since less than 5 % of the data is missing, listwise deletion is recommended for handling missing values (Allison, 2010). All analyses were conducted in Stata 18. Replication code is available online at [redacted for review].

Results

Table 1 presents descriptive statistics for the sample. The average age of participants is 39 (SD = 9). The majority are men (75 %), White (82 %), not married (91 %), and have a high school education (60 %). On average, they have served 53 months in prison, with a mean core support network size of 3.52.

Table 1.

Descriptive statistics (N = 389).

	Mean/Prop.	SD	Range
Age	39.20	8.99	21–66
Gender
Women	.24
Men	.75
Other	.01
White	.82
Education
<HS	.15
GED/HS	.60
>HS	.25
Married	.09
Months in prison	52.53	51.20	1–535
Completion of SUD Treatment Program
No	.08
Completed	.63
Currently in the SUD treatment program	.29
Core support network size	3.52	2.27	0–14
Have a tie with incarcerated peers	.20
Proportion ever incarcerated	.26	.31	0–1
Proportion ever use illicit opioids	.28	.34	0–1
Proportion kin	.79	.28	0–1
Mean closeness	4.59	.59	2–5
Mean contact	4.17	.73	1–5
Proportion provide emotional support	.89	.23	0–1
Proportion do leisure activities	.67	.40	0–1
Proportion give advice	.92	.17	0–1
Mean trust	4.60	.63	1–5

Note: SUD = Substance use disorder; Prop. = Proportion; SD = Standard deviation.

Are incarcerated peers part of a participant’s core support network?

As shown in Fig. 1, approximately 20 % of participants include at least one incarcerated peer in their core support network. Among these, about half include only one incarcerated peer; the maximum number of incarcerated peers in a single core support network is five. However, the distribution shifts when examining the proportion of incarcerated peers in these networks. Among the 20 % of participants who include at least one incarcerated peer, the proportion of incarcerated peers in their core support networks varies widely, ranging from 0.07 to 1, indicating a less concentrated and more dispersed pattern. This divergence between the number and proportion of incarcerated peers arises from the wide variation in core support network sizes. This suggests that participants include incarcerated peers in their networks regardless of the overall size of their core support networks, challenging the notion that incarcerated persons are only included in small or nonexistent networks.

Who are the participants including incarcerated peers in their core support network?

Table 2 presents demographic characteristics associated with including at least one incarcerated peer in the core support network. Gender is the only statistically significant demographic control, with men being substantially less likely than women to name incarcerated peers (OR = 0.211, p < 0.001). Of the two measures of contact opportunities, greater time spent in prison is associated with higher odds of including incarcerated peers in the core support network (OR = 1.006, p < 0.05), whereas completion of the SUD treatment program is not significantly associated.

Table 2.

Demographic variables associated with naming at least one incarcerated peer.

	Odds Ratio
Age	1.017
	(0.016)
Gender (Ref. = Women) Men	0.211***
	(0.064)
Other	1.061
	(1.548)
White	1.612
	(0.659)
Education (Ref. = GED/HS)
Less than high school	1.235
	(0.466)
More than high school	0.873
	(0.291)
Married	0.759
	(0.365)
Contact opportunities
Months in prison	1.006*
	(0.003)
Completion of the SUD treatment program (Ref. = Completed)
No	2.071
	(0.991)
Currently in the SUD treatment program	1.211
	(0.378)
N	389
BIC	413.689

Note:

^*p < 0.05

^**p < 0.01

^***p < 0.001.

SUD = Substance use disorder; Ref. = Reference group.

Table 3 highlights the network correlates associated with the inclusion of at least one incarcerated peer in a participant’s core support network. As expected, experiential homophily plays a significant role. A lower proportion of ties with prior incarceration experience is associated with higher odds of including incarcerated peers (OR = 0.112, p < 0.001). However, the proportion of ties who have ever used illicit opioids shows no significant association. Unmet social needs are also significantly associated with the outcome. Specifically, a lower proportion of kin ties (OR = 0.235, p < 0.01) and less trust with the network members (OR = 0.551, p < 0.01) are related to higher odds of including incarcerated peers. However, other dimensions of unmet social needs, such as core support network size, mean closeness, mean contact frequency, proportion providing emotional support, proportion doing leisure activities, and proportion giving advice, are not statistically significant.

Table 3.

Network characteristics associated with naming at least one incarcerated peer.

	Odds ratio
Homophily
Prop. ties ever use illicit opioids	0.666
	(0.246)
Prop. ties ever incarcerated	0.112***
	(0.069)
Unmet social needs
Core support network size	1.126
	(0.075)
Prop. kin ties	0.235**
	(0.112)
Mean closeness	0.706
	(0.172)
Mean contact frequency	1.182
	(0.254)
Prop. provide emotional support	2.004
	(1.455)
Prop. do leisure activities	0.743
	(0.251)
Prop. give advice	1.051
	(0.795)
Mean trust	0.551**
	(0.118)

Note:

^*p < 0.05

^**p < 0.01

^***p < 0.001.

All network analyses are limited to participants with at least one network member to avoid perfect separation issue. All network measures exclude incarcerated peer ties to avoid endogeneity. Network measures are included in the model individually to avoid multicollinearity issue. All models adjust for age, gender, race, education, marital status, time in prison, completion of the SUD treatment program, and core support network size. See Table S2 in the supplemental materials for full tables.

How do ties with incarcerated peers differ from other ties in the core support network?

Based on analysis of 1352 ties, Fig. 2 illustrates the differences between ties with incarcerated peers and other ties across various network functions and characteristics. We find that participants are on average less close to and have less trust in incarcerated peer ties than other ties. In contrast, participants have more contact with and engage in more leisure activities with incarcerated peer ties than other ties. Ties to incarcerated peers are more likely to have a history of illicit opioid use compared to other ties. Surprisingly, there is no statistically significant difference in emotional support and advice support.

Discussion

This study used egocentric network data from 389 individuals who are incarcerated with a history of OUD, encompassing 1352 network members, to analyze the following research questions: (1) Do incarcerated individuals include at least one incarcerated peer in their core support network? (2) Who includes incarcerated peers in their core support network, and what demographic, network, and prison-related factors predict this inclusion? (3) How do ties with incarcerated peers differ from other ties in the core support network?

Do incarcerated individuals include at least one incarcerated peer in their core support network?

We find that approximately 20 % of participants include at least one incarcerated peer tie in their core support network, underscoring the potential importance of social ties formed within the prison environment. Notably, the core support network consists of a small group of core supporters, with an average network size of 3.52. This is comparable to the core discussion network size of the civilian population. For instance, U.S. older adults in the 2015/2016 wave of the National Social Life, Health, and Aging Project had a mean core discussion network size of 3.79 (Cornwell et al., 2021), while U.S. adults in the 2006 wave of the General Social Survey reported a mean core discussion network size of 2.08 (McPherson et al., 2006). Given the relatively small size of these core support networks, the inclusion of even one incarcerated peer significantly impacts the composition of the core support network. More importantly, our findings demonstrate that relationships between incarcerated persons can be examined using a core network framework, extending beyond weak ties (e.g., “get along the most”) approach previously used (Schaefer et al., 2017, 2021; Schaefer and Kreager, 2020), and filling a critical gap related to incarceration and core networks.

In the field of core discussion networks, there is an ongoing debate regarding the role of weak ties within these networks. Traditionally, scholars have argued that core discussion networks are composed of the closest and strongest ties (Burt, 1984; Granovetter, 1973; Marsden, 1987; McPherson et al., 2006). However, this long-standing view has been challenged by recent research emphasizing the importance of weak ties in discussing significant and personal issues (Small, 2013, 2017; Small et al., 2024). This study contributes to this debate by demonstrating that, even in the high-risk environment of prison, relationships between incarcerated persons can be strengthened and integrated as a core part of one’s support system.

Who are the participants including incarcerated peers in their core support networks?

Our findings offer support for contact opportunities theory (Blau, 1960; Feld, 1981; Fischer, 2011). In particular, longer prison stays are associated with an increased likelihood that participants include at least one incarcerated peer in their core support networks, suggesting that more contact opportunities foster deeper interpersonal connections. In contrast, completion of the SUD treatment program is not significantly related to the likelihood of including incarcerated peers. This result is somewhat unexpected given that these types of therapeutic community programs are designed to provide an immersive, pro-social environment in which residents provide social support and role modeling for one another (De Leon, 2000; Schaefer et al., 2021; Schaefer and Kreager, 2020); however, therapeutic communities also use accountability frameworks where participants are required to report their peers for rule infractions (De Leon, 2000; Schaefer et al., 2021; Schaefer and Kreager, 2020).

Additionally, our findings support the unmet social needs theory, indicating that participants with limited core support networks are more likely to incorporate incarcerated peers into their core support systems (Sykes, 2007; Thibaut, 1959; Volker et al., 2007). Specifically, participants with a lower proportion of kin ties and less trust within their core support networks are significantly more inclined to include at least one incarcerated peer. However, other dimensions of unmet social needs—such as core support network size, average closeness, contact frequency, and the proportions of members providing emotional support, participating in leisure activities, or offering advice—are not statistically significant. These findings suggest that not all dimensions of social needs are equally influential in determining whether a participant includes incarcerated peers in their core support networks. The significant predictors (kin ties and trust) speak to the quality and stability of relationships rather than mere number or frequency of interactions. It may be that having a trusted and family-based support system is especially critical for individuals who are incarcerated; and it is possible that they are more likely to seek support from incarcerated peers when those social support structures are absent.

The theory of experiential homophily (Kossinets and Watts, 2009; McPherson et al., 2001; Suitor et al., 1995) highlights the importance of shared experiences in shaping incarcerated individuals’ core support networks. Our findings reveal that participants with fewer connections to individuals with prior incarceration experience are more likely to include at least one incarcerated peer in their core support networks. This highlights the unique role shared experiences of incarceration play in fostering support and mutual understanding. However, the proportion of ties with a history of illicit opioid use shows no significant association. This may suggest that, within the prison environment, the most salient stressors are those related to incarceration itself, while stressors associated with opioid use disorder may be less salient in shaping core support networks during imprisonment.

How do ties with incarcerated peers differ from other ties in the core support network?

As expected, ties with incarcerated peers are associated with higher levels of contact and shared leisure activities. However, they are characterized by lower levels of closeness and trust compared to other ties, who primarily consist of family members and close friends from the community. Nonetheless, there is no statistically significant difference in emotional or advice support between these two types of ties. These findings suggest that ties with incarcerated peers not only fulfill immediate needs for social contact and leisure—resources that are difficult to access from outside the prison—but also provide emotional support and advice on par with other ties. This highlights the multifaceted role of relationships between incarcerated individuals.

Limitations and future studies

Our findings demonstrate that gender significantly influences the core support networks of individuals who are incarcerated, with men being less likely than women to rely on incarcerated peers for support. This implies that a gender stratified approach is needed to fully understand gender differences in incarcerated individuals’ core support networks. Future research should examine these gender differences in greater depth to provide more nuanced insights into the role of gender in shaping core support networks within incarceration.

Our findings may not be generalizable to other populations of incarcerated individuals, given that the sample was recruited from prisons only, in one southern state, and only including individuals who met criteria for opioid use disorder and had currently or previously participated in a corrections-based SUD treatment program. Results may differ for individuals who do not have a history of SUD (or SUDs other than opioid use) or who are not treatment-seeking, those who are serving time in shorter-term carceral facilities (i.e., jails), or those from other regions of the United States with more racial/ethnic diversity.

Due to the cross-sectional nature of the analyses, it is not possible to determine the causal direction of tie formation. Establishing causality would require network data collected prior to incarceration. Additionally, future research should examine changes in network ties longitudinally—particularly after release from prison—to assess whether close relationships formed during incarceration are sustained in the community.

Conclusion

This study highlights the significant role that incarcerated peers play in providing support and helping individuals cope with the challenges of incarceration. Notably, 20 % of participants include at least one incarcerated peer in their core support networks, where these ties serve both practical and emotional support functions. Our findings demonstrate that relationships between incarcerated persons can be examined using a core network framework, extending beyond weak ties approach previously used (Schaefer et al., 2017, 2021; Schaefer and Kreager, 2020). Theoretically, this research offers a unique opportunity to test tie formation theories within the context of total institutions. Our findings provide evidence supporting the theories of contact opportunities (Feld, 1981), unmet social needs (Thibaut, 1959), and experiential homophily (McPherson et al., 2001) in shaping participant decisions to include incarcerated peers in their core support networks. By examining the dynamics of core support networks during incarceration, this study contributes to a deeper understanding of resilience, trust formation, and the critical role of “weak” ties in navigating high-stress, constrained environments.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.socnet.2025.12.004.