Sample
The sample for the current study included 282 freshman students at two universities in the United States collected across two points in time (2008, 2011). This sample resulted from taking the freshman students from a larger sample (described below; N = 1,980) and first selecting the students who identified as currently engaging in nonsuicidal self-injury (NSSI). Currently engaging in NSSI was defined as self-reporting engaging in NSSI behaviors within the past 90 days and included a total of 99 participants in the sample. Once those who self-reported currently engaging in NSSI were identified, they were matched with students who had never self-injured and students who had previously self-injured (past NSSI) on race, sex, and university (the latter was matched given the two different locations). When more than one match existed for a currently engaged in NSSI participant, participants with complete data were randomly selected. Of the original 99 currently engaged in NSSI, five students could not be matched due to a lack of matching students in the other two self-injury categories, resulting in a total of 94 current engaged in NSSI students, 94 past engaged in NSSI students, and 94 never engaged in NSSI students (total N = 282) matched on race, sex, and university.
This sample of 282 originated from a larger sample of 1,980 college freshman. The matched participants (N = 282) did not significantly differ from the unmatched participants (n = 1,698) regarding age and sex. A significant difference did exist in regards to race (X = 21.01, p < 0.01). Specifically there were less Asians and African Americans, and significantly more Hispanic and Whites observed in the NSSI matched sample than expected. This is expected as researchers have found lower prevalence of self-harm behaviors among Asian and African American individuals [18]. Of the 282 freshmen student sample used in this study, the majority of participants were female (72%) with 28% identifying as male, with a mean age of 18.50 (SD = 2.32). The majority of participants were White (48.9%), followed by Hispanic (24.5%), Black/African American (10.6%), Multiracial (11.7%), and Asian (4.3%). Slightly over half of the sample came from University B (n = 150, 53.2%) with the remainder coming from University A (46.8%). Equal numbers from each sex, university, and racial category were present in current engagement in NSSI, past engagement in NSSI, and never engaged in NSSI groups as the groups were matched on these demographics.
Procedures
Incoming freshman at the two universities (A and B) were targeted. University A was a midsized university located in the southeast United States and University B was a mid-sized Hispanic Serving Institution located in the southwest United State. Procedures from the two time points, and at both universities, were the same. Information for both the 2008 and 2011 samples will be provided here to better understand each individual sample; however, final sample demographics for the 282 participants were given in the sample section above for this study.
At both universities, freshman participants were randomly selected from the larger freshman student body. Specifically in 2008, a random selection of 2,400 incoming freshman consisted at University A and 8,000 at University B. Out of those freshmen, 1,396 students responded (13.5% response rate). Similarly, in 2011, a random selection of 2,525 freshmen from University A and 4,953 freshmen from University B was sampled. A total of 584 students responded (8% response rate: 300 University A; 284 University B). Samples were compared by data collection point, and by university, and no significant differences were found between groups on NSSI behavior variables or independent variables. Final respondents were similar to their university freshman student body on race, age, and sex. Therefore, the two universities and two time point samples were collapsed into one larger sample (N = 1,980), with the final matched sample being used for this study (N = 282).
Both the 2008 and 2011 freshman samples were sent an e-mail through their university email account inviting them to participate in this study. The e-mail contained a link to an online survey, which was the primary method of data collection. If they did not respond to the first e-mail they were sent a follow-up 1–2 weeks later for a total of three e-mails. Both samples had incentives for participation: In 2008 students were offered the possibility of winning one of three $50 raffles; in 2011 students were offered the possibility of winning an Apple iPod Touch.
Instruments
Participants were asked to complete a demographic form that included sex, age, year in school, and race. They were also asked to complete measures of NSSI, ethnic identity, depression, anxiety, coping behaviors, and perceptions of interpersonal support. Each of these measures are described below.
Nonsuicidal self-injury
NSSI was measured through the use of an adapted version of the Deliberate Self Harm Inventory (ADSHI; original DSHI was developed by Gratz [1]). The ADSHI assessed NSSI engagement (yes/no), number and type of method, and frequency of engagement (count frequency within past 90 days). The ADSHI contains 12 items that assess for lifetime and current (90 days) engagement in particular NSSI behaviors (e.g., cut, burn, skin pick). If participants indicated they currently utilized a specific method of NSSI, they were asked to report their frequency of engagement with that method in the past 90 days. The ADSHI has been found to have adequate estimates of reliability (Cronbach α = 0.70 on both lifetime and current engagement [18, 31].
Ethnic identity
The Multi-Ethnic Identity Measure (MEIM) [32] was used to assess identification with participants’ self-identified ethnic group. The MEIM consists of 12 items rated on a 4-point Likert-type scale. The MEIM consists of two subscales [20]: Affirmation, Belonging, Commitment (MEIM-A) which measures the participant’s attitudes and feelings surrounding their identification with their ethnic group as well as the degree to which the participant identifies with their ethnic group; and Ethnic Identity Achievement (MEIM-EI), which measures the level of the participants understanding of and awareness about their ethnicity. Internal consistency of the measures for the current study was adequate (Cronbach α = 0.90 entire scale; MEIM-A α = 0.86; MEIM-EI α = 0.76).
Depression
The Center for Epidemiological Studies for Depression Scale (CES-D), short version [33] was used to measure participants’ level of depression. This assessment consists of a 10-item scale measuring the participant’s amount of depressive symptoms. Research has shown that the CES-D has good predictive accuracy for depression and adequate reliability (0.64), with Cronbach’s α of 0.61 in the current study. High scores on this assessment indicate higher levels of depression.
Anxiety
The 5-item Anxiety subscale of the PGI General Well Being Scale [34] was used to assess for anxiety. Respondents answered each item on a 4-point Likert-type scale, rating the frequency of occurrence of each item. High scores on this subscale indicate higher levels of wellness and less anxiety. Cronbach’s alpha for the current study was .78.
Coping
The Brief COPE [35], which consists of 28-item to assess 14 different coping styles, was used to assess maladaptive and adaptive coping. Participants rate their use of various coping skills from (0) “I usually don’t do this at all” to (3) “I usually do this a lot”. This assessment has scale reliabilities of 0.71 [35]. For the purpose of this study the various coping styles were organized into two subscales. The first subscale is Avoidant/Maladaptive Coping (Cronbach’s alpha = 0.73), which consists of denial, self-distraction, venting, substance abuse, behavioral disengagement, and self blame. The second subscale is Active/Adaptive Coping (Cronbach’s alpha = 0.79), which consists of active, planning, instrumental support, positive reframe, humor, acceptance, religion, and emotional support.
Interpersonal support
The Multidimensional Scale of Perceived Social Support (MSSPSS) [36] utilizes a Likert scale (“very strongly” to “disagree”) to assess for an individual’s perceived social support from family, friends, and significant others. For this study, the full scale of interpersonal support was used, in addition to the three individual subscales. In previous studies reliability for this assessment has been found to range on the scales from .81 to 0.98. Cronbach alpha in the current study were 0.91.
Data analysis
Descriptive statistics were conducted to examine the NSSI behaviors engaged in by the current and past NSSI groups. One-way ANOVAs were used to examine if the current, past, and never engaged in NSSI groups significantly differed on coping, interpersonal support, depression, anxiety, and ethnic identity after they were matched on sex, race, and university (environmental context). Due to the matched nature of this sample, if one of the participants was missing data on a particular scale (e.g., depression), all matched individuals were removed from that analysis.