Psychometric evaluation of the Turkish version of the Substance Use Risk Profile Scale in adolescents

Abstract

The Substance Use Risk Profile Scale (SURPS) was developed as a self-report measure to screen four high-risk personality dimensions (hopelessness, anxiety sensitivity, impulsivity, and sensation seeking) that predict specific patterns of vulnerability to substance use and other problematic behavior. The scale has previously been shown to have adequate psychometric properties in various other languages. The aim of the present study was to evaluate the psychometric properties of the Turkish version of the SURPS. A cross-sectional total sampling design was adopted in which 875 Turkish participants took part (mean age 15.58 years, SD = 1.03). Using exploratory factor and correlational analyses, the internal consistency, test–retest reliability, construct validity, and concurrent validity were evaluated. The SURPS demonstrated good psychometric and construct validity. The results provide clear evidence that the Turkish version of the SURPS is valid and reliable. The Turkish version of the SURPS may become an important tool for daily clinical practice and clinical research.

Introduction

The increasing prevalence of adolescent substance abuse over the past 25 years has made it one of the major global health problems faced today (Peiper, Ridenour, Hochwalt, & Coyne-Beasley, 2016). Adolescence is a period of dynamic biological, psychological, and social growth. Substance use during this period can disrupt growth significantly, impairing an otherwise natural course of cognitive, emotional, and social development. The deleterious effects of substance abuse on the individual, families, and the larger community are well documented and have proved costly. Compared to their nonabusing counterparts, substance-abusing adolescents underperform academically, are less employable, and lack the necessary family and social support during a critical period of social development. Moreover, there are strong associations among substance abuse, comorbid mental illness, and problematic behavior (e.g., violence and criminal and risky sexual behavior) (Borus, Parhami, & Levy, 2016). The significant social and economic costs of substance abuse to national economies has therefore led to large health initiatives to relieve the burden and eliminate clinical onset, making substance abuse prevention paramount (LeNoue & Riggs, 2016).

Substance abuse prevention begins with assessment. One promising approach to risk assessment has been to identify personality vulnerability. Individuals would use substances to obtain a positive outcome or to avoid a negative outcome. According to Cooper’s motivational model, these two motives (enhancement or coping) are explained by positive (e.g., positive mood change and social approval) and negative (e.g., avoiding negative emotions and social censure) reinforcement processes in conditioning. Motives for substance use appear to be the final common pathway to substance abuse through which more distal risk variables, such as personality factors, apply their influences (Cooper, 1994). Motivation and reinforcement-specific personality profiles (Comeau, Stewart, & Loba, 2001; Woicik, Stewart, Pihl, & Conrod, 2009) have been shown to be particularly predictive of subsequent development of substance abuse (Castellanos-Ryan & Conrod, 2011; Conrod & Woicik, 2002). Woicik et al. (2009) found that four personality traits (hopelessness, anxiety sensitivity, impulsivity, and sensation seeking) explained drug abuse vulnerability. The relationship between these personality traits and substance abuse is mediated by different motivational patterns and reinforcement processes (Castellanos-Ryan, O’Leary-Barrett, Sully, & Conrod, 2013; Conrod, Pihl, Stewart, & Dongier, 2000; Conrod, Stewart, et al., 2000). In their model, individuals who score high on either of the internalizing traits (anxiety sensitivity or hopelessness) are susceptible to negative reinforcement from alcohol and drug use. However, the reinforcement patterns are different; while individuals high in anxiety sensitivity use alcohol and drugs to cope and reduce anxiety and panic symptoms, those high in hopelessness are differentially reinforced by alcohol and drugs, using them instead to relieve negative affective states (e.g., negative thinking, depressive thoughts) (Comeau et al., 2001; Krank et al., 2011). Individuals scoring high on the externalizing traits of the model (sensation seeking and impulsivity) are thought to be more susceptible to positive reinforcement related to intense novelty desire (seeking novel and thrilling experiences), risk taking, and poor behavioral inhibition to enhance positive affective states (Castellanos-Ryan & Conrod, 2011; Sher, Bartholow, & Wood, 2000; Woicik et al., 2009). These findings are relevant in the context of the Research Diagnostic Criteria framework, which integrates many levels of information (from genomics and circuits to behavior and self-reports) for understanding negative and positive valence systems and developing new approaches for prevention of mental disorders associated with substance use exposure (Kozak & Cuthbert, 2016).

The Substance Use Risk Profile Scale (SURPS) was developed to screen for the four high-risk personality traits (hopelessness, anxiety sensitivity, impulsivity, and sensation seeking) for substance abuse (Conrod & Woicik, 2002; Woicik et al., 2009). The SURPS is a self-report instrument of 23 items rated on a Likert scale (1 = strongly disagree to 4 = strongly agree). It has demonstrated satisfactory internal consistency, test–retest reliability, and construct, concurrent, and predictive validity while measuring these personality variables for the goal of predicting the development of substance misuse (Castellanos-Ryan et al., 2013; Woicik et al., 2009). Psychometric properties of the SURPS have been evaluated with adequate results in English (Castellanos-Ryan et al., 2013; Krank et al., 2011; Newton et al., 2016; Woicik et al., 2009), Dutch (Malmberg et al., 2010), Sinhala (Chandrika Ismail, De Alwis Seneviratne, Newcombe, & Wanigaratne, 2009), French-Quebequois (Castonguay-Jolin et al., 2013), Spanish-speaking Mexican (Robles-Garcia et al., 2014), Japanese (Omiya, Kobori, Tomoto, Igarashi, & Iyo, 2015), Korean (Saliba, Moran, & Yoo, 2014), Chinese (Siu, 2011), German (Jurk et al., 2015), Portuguese-speaking Brazilian (Canfield, Gilvarry, & Koller, 2015), clinical (Schlauch et al., 2015), and adolescent population–based samples (Castonguay-Jolin et al., 2013; Chandrika Ismail et al., 2009; Jurk et al., 2015; Krank et al., 2011; Malmberg et al., 2010; Newton et al., 2016; Siu, 2011; Woicik et al., 2009).

Cross-cultural differences are able to influence adolescents’ decisions about substance use. It is not clear what mediates this influence. The association between acculturation and substance use behavior suggests that traditional cultural values could be protective or risk factors for substance abuse (Johnson et al., 2006). For example, especially in Asian cultures, filial piety, which emphasizes obedience to parents, may protect adolescents from substance use. Similarly, in Latin American cultures, familism, a sense of obligation to, and connectedness with, one’s immediate and extended family, may be associated with a lower risk of substance use. Conversely, in Hispanic cultures, machismo, which prescribes differentiated gender roles (unlike women, the qualities associated with men are believed to be masculinity, male dominance, sexual prowess, physical strength, and honor), may be related to a higher risk of substance use among boys (Unger et al., 2002). In addition to cultural values, there are also studies that revealed cross-cultural differences on personality traits (Eysenck & Haapasalo, 1989). English and French adolescents scored higher than German adolescents in the Hopelessness, Impulsivity, and Sensation Seeking subscales of the SURPS (Jurk et al., 2015). Koreans consistently scored higher than Australians in all subscales. Both nations nevertheless demonstrated a similar relationship between the personality traits and substance use (Saliba et al., 2014). Although the SURPS is sensitive to national differences in terms of subscale scores, the four personality dimensions were found to be usually correlated with substance use in different nations (Omiya et al., 2015). These personality traits have not yet been evaluated in Turkish adolescents.

Main effects for gender were found on anxiety sensitivity (females scored significantly higher) and sensation seeking (males reported significantly higher) in the other studies of gender associations with personality and in the original study that informed construction of the SURPS (Canfield et al., 2015; Siu, 2011; Woicik et al., 2009). Males are thought to be more prone to personality traits related to externalizing, whereas predominant personality traits in females are associated with internalizing (Muris, Meesters, & Blijlevens, 2007). Gender effects on the personality traits related to substance use have not yet been replicated in Turkish adolescents.

Taken together, the previous studies have shown that the SURPS is a reliable instrument in assessing high-risk personality traits and developing early interventions for prevention of substance use disorders. There is currently no Turkish version of SURPS. Therefore, the aim of the present study was to evaluate the psychometric properties of a Turkish version of SURPS and to establish normative scores in this population. We hypothesize that the Turkish version of the SURPS will be a suitable and valid measure of personality in Turkish adolescents.

Methods

Procedure

Permission to validate the SURPS in Turkey was obtained from P. Woicik, who developed the scale. The SURPS was translated into Turkish by three child and adolescent psychiatrists proficient in English [SDU, OSU, and EC]. The scale was then blindly back-translated to English by a bilingual translator in the Department of Linguistics at Stony Brook University [AY]. Finally, Woicik and her colleagues compared the translated and back-translated versions with the original English version. There were a few discrepancies. Minor corrections were made by consensus.

Measures

The assessment consisted of the Turkish version of SURPS and four additional self-administered instruments: the Beck Hopelessness Scale, the State Trait Anxiety Inventory, the Barratt Impulsivity Scale, and the Arnett Inventory of Sensation Seeking. These measures were used to determine concurrent validity of the SURPS subscales.

1. The SURPS is a 23-item (4-point Likert scale: 1 = strong disagree to 4 = strongly agree) instrument designed to assess the personality dimensions such as anxiety sensitivity, hopelessness, impulsivity, and sensation seeking (items of the SURPS are presented in Table 1). The SURPS has demonstrated good psychometric properties in nonclinical samples, including convergent, discriminant, and predictive validity (Krank et al., 2011; Woicik et al., 2009).

2. The Beck Hopelessness Scale (BHS), a 20-item instrument rated on a “yes” or “no” response (Beck, Weissman, Lester, & Trexler, 1974), has shown adequate internal consistency in the Turkish population (Cronbach’s alpha = .93) (Durak & Palabıyıkoğlu, 1994; Seber, Dilbaz, Kaptanoğlu, & Tekin, 1993).

3. The State Trait Anxiety Inventory (STAI) State subscale consists of 20 items describing how a person feels at a particular moment rated on a Likert scale (1–4 points) (Spielberger, Gorsuch, & Lushene, 1970). The STAI Trait subscale consists of 20 items describing how a person generally feels rated on a Likert scale (1–4 points) (Spielberger et al., 1970). The State and Trait subscales of STAI have been validated with alpha values of .83–.87 and .94–.96, respectively, in the Turkish population (Oner & Le Compte, 1985).

4. The Barratt Impulsivity Scale (BIS), which comprises 30 items scored on a 4-point Likert instrument ranging from 1 (rarely/never) to 4 (almost always/always), includes three subscales labeled Attentional, Non-planning, and Motor (Patton, Stanford, & Barratt, 1995). The validation of the BIS has been replicated in Turkey (Güleç et al., 2008). In the current study, Cronbach’s alpha for internal consistency was .78.

5. The Arnett Inventory of Sensation Seeking (AISS), composed of 20 items rated on a Likert scale (1–4 points) (Arnett, 1994), has adequate internal consistency in the Turkish population (with alpha values of .84 and .75) (Sümer, 2003).

Table 1.

Factor Loadings of the SURPS Items.

Item		Factor
		Lack of self-contentment (formerly hopelessness)	Impulsivity	Sensation seeking	Anxiety sensitivity
1	I am content	.751
4	I am happy	.845
7	I have faith that my future holds great promise	.662
13	I feel proud of my accomplishments	.447
17	I feel that I’m a failure	.338			.383
20	I feel pleasant	.749
23	I am very enthusiastic about my future	.599
2	I often don’t think things through before I speak		.567
5	I often involve myself in situations that I later regret being involved in		.546
11	I usually act without stopping to think		.768
15	Generally, I am an impulsive person		.740
22	I feel I have to be manipulative to get what I want		.343
3	I would like to skydive			.674
6	I enjoy new and exciting experiences even if they are unconventional			.661
9	I like doing things that frighten me a little			.660
12	I would like to learn how to drive a motorcycle			.614
16	I am interested in experience for its own sake, even if it is illegal		.348	.558
19	I would enjoy hiking long distances in wild and uninhabited territory			.551
8	It’s frightening to feel dizzy or faint				.668
10	It frightens me when I feel my heartbeat change				.632
14	I get scared when I’m too nervous				.495
18	I get scared when I experience unusual body sensations				.547
21	It scares me when I’m unable to focus on a task				.617
	Eigenvalue	3.77	2.96	2.24	1.41
	Variance (%)
	45.18	16.42	12.87	9.75	6.13
	Cronbach’s alpha
	.728	.758	.668	.707	.658
	Mean	13.96	10.57	16.28	11.00
	SD	4.34	3.46	4.36	3.50

Note. SURPS = Substance Use Risk Profile Scale.

Participants

Approval for the design and data collection procedures was obtained from the ethics review committee at the Yildirim Beyazit University, Ankara. The study sites included two public high schools in the Altindag district of Ankara from low and high socioeconomic income census areas, respectively. Additional approvals were elicited from the school authorities.

To minimize selection bias, a total sample of students from the two schools (n = 1,074) was approached, and 905 (84%) students returned the distributed scales with informed consent, including the additional self-report study measures with similar response rates between grades. Participant questionnaires that were incomplete were removed from the analyses (n = 30), yielding the final study sample of 875 participants (81.5%; n = 349 from a low socioeconomic neighborhood school, and n = 526 from a high socioeconomic neighborhood school). All the survey data were collected on a class basis during the guidance hours. The student ages were 14 to 19 years (mean 15.58, SD = 1.03; 502 males, 369 females, and 4 adolescents who left the gender question blank in self-reports). The grade distribution for the combined schools was 350 in ninth grade; 271 in 10th grade; 171 in 11th grade; 83 in 12th grade. Seventy-five students were randomly selected and further requested to complete the SURPS a second time after one month to examine test–retest reliability of the SURPS subscales.

Statistical analyses

First, a principal-component analysis with varimax rotation was performed to assess the construct validity of the Turkish version of the SURPS. Kaiser–Meyer–Olkin’s measure of sampling adequacy and Bartlett’s test of sphericity were calculated to verify the appropriateness of using the factor analysis. The exploratory factor analysis was completed to explore the factor structure of the SURPS. Interpretation of the factor loadings for all analyses was based on the guidelines set forth by Comrey and Lee. They suggested that loadings greater than .71 should be considered excellent, .63 very good, .55 good, .45 fair, and .32 poor (Comrey & Lee, 1992). Items whose factor loadings were greater than .32 were retained and allocated to factors. The factor extraction was performed with Eigenvalues set at greater than 1. The scree plot graph was utilized to determine the number of factors by counting the number of points where the slope of the curve clearly leveled off. Second, to evaluate the internal consistency, Cronbach’s alpha coefficients of the Turkish version of SURPS and all four subscales, Cronbach’s alpha if item deleted, and item total correlation coefficients of all 23 items were calculated. Third, Pearson correlation coefficients for the test–retest reliabilities of all four subscales were also calculated. Fourth, the concurrent validity of the SURPS subscales was obtained by determining their association with BHS (for the Hopelessness subscale), STAI (for the Anxiety Sensitivity subscale), BIS (for the Impulsivity subscale), and AISS (for the Sensation Seeking subscale) with Pearson correlation coefficients. Statistical analyses were carried out using SPSS 17.0 software. A significance level of p ≤ .05 was set for all analyses.

Results

Factor structure of the Turkish version of SURPS

Five factors (instead of the expected four) accounted for 49.67% of the variance in item scores. Eigenvalues (>1) were 3.77, 2.96, 2.24, 1.41, and 1.03. All item loadings were above .32 and none was excluded. Observation of the inflexion point after fourth factor in the scree plot graph suggested a four-factor model (Figure 1). Four of the factors resembled the original defined structure. Among the 23 items, only three (items 7, 13, and 23) loaded on a separate fifth factor. All other items replicated the expected factor structure as reported in the literature. Therefore, the four-factor model was “forced” in the second factor analysis. In addition to items 7, 13, and 23 (with factor loading of .662, .447, and .599 on the hopelessness factor, respectively), all items loaded on the original defined factors at greater than .32. Four factors accounted for 45.18% of the variance, and eigenvalues (>1) were 3.77, 2.96, 2.24, and 1.41 representing hopelessness, impulsivity, sensation seeking, and anxiety sensitivity, respectively (presented in Table 1).

Figure 1.

The Scree Plot Graph of the Five-Factor Structure.

A common feature of items 7 (“I have faith that my future holds great promise”) and 23 (“I am very enthusiastic about my future”) is that they are positive statements about the future, so they may have diverged from the other statements that comprise the Hopelessness subscale. When assessing the contextual meaning of items 7 and 23, they were included in the Hopelessness subscale (as in the original SURPS) because future orientation increases during adolescence, and thus, viewing the future in a negative light, such as giving a low endorsement to these items, may be more relevant to adolescents’ hopelessness. The content for item 13 (“I feel proud of my accomplishments”), which loaded at .447 on the hopelessness factor, was interpreted as lack of self-contentment rather than hopelessness, similar to item 17 (“I feel that I’m a failure”), whose factor loading was .338 on the same factor. When the meanings of hopelessness items (items 1, 4, 7, 13, 17, 20, and 23) and factor loadings for this factor were examined, although those for items 13 and 17 were interpreted as poor, replacing the subscale name of “Hopelessness” with “Lack of Self-Contentment” made it more inclusive by supporting the original four-factor structure. Therefore, the name “Lack of Self-Contentment” was used instead of the “Hopelessness” subscale for the Turkish version of the SURPS.

Only items 16 and item 17 cross-loaded on to another factor as well as the expected factor of the original SURPS (Table 1). The factor loadings for item 16 (“I am interested in experience for its own sake, even if it is illegal”) were .348 and .558 on impulsivity and sensation seeking, respectively. As in the original SURPS, item 16 was considered a sensation-seeking factor rather than an impulsivity factor when graded on loading according to Comrey and Lee (1992). Although the factor loadings on lack of self-contentment (formerly hopelessness) and anxiety sensitivity (.338 and .383, respectively) for item 17 were interpreted as poor, we preferred to consider it in the intended factor (lack of self-contentment [formerly hopelessness]) rather than anxiety sensitivity in terms of meaning.

Mean subscale scores of the Turkish version of the SURPS fell within the normative scores found in English versions: Lack of Self-Contentment (formerly Hopelessness) subscale, 13.96, SD = 4.34 points; Anxiety Sensitivity subscale, 11.00, SD = 3.50 points; Impulsivity subscale, 10.57, SD = 3.46 points; and Sensation Seeking subscale, 16.28, SD = 4.36 points. Males scored significantly higher on the sensation seeking [t(805) = 2.736, p < .05] and the impulsivity [t(827) = 2.038, p < .05] subscales, and females scored significantly higher on the anxiety sensitivity subscale [t(815) = −3.960, p < .001].

Internal consistency

The Turkish version of SURPS showed adequate to good internal consistency (Cronbach’s alpha coefficient = .728). Alpha coefficients were .758, .658, .668, and .707 for the Lack of Self-Contentment (formerly Hopelessness), Anxiety Sensitivity, Impulsivity, and Sensation Seeking subscales, respectively (Table 1). These values were similar to those reported in the previous validation studies and the original study (Chandrika Ismail et al., 2009; Jurk et al., 2015; Krank et al., 2011; Robles-Garcia et al., 2014; Siu, 2011; Woicik et al., 2009).

Test–retest reliability

For the test–retest reliability, we randomly selected 75 adolescents, who completed the Turkish version of the SURPS a second time one month after the initial evaluation. Pearson correlation coefficients for the test–retest reliabilities of the Lack of Self-Contentment (formerly Hopelessness), Sensation Seeking, Impulsivity, and Anxiety Sensitivity subscale scores were .779, .703, .608, and .646, respectively (p < .01) (Table 2). Similar to the other versions and the original SURPS (Omiya et al., 2015; Siu, 2011; Woicik et al., 2009), the Turkish version of SURPS showed quite strong test–retest reliability estimates.

Table 2.

Test–Retest Reliabilities of the SURPS Subscales.

	SURPS
	Lack of self-contentment (formerly hopelessness)	Impulsivity	Sensation seeking	Anxiety sensitivity
Test mean (±SD)	13.94 ± 4.73	11.41 ± 3.35	17.08 ± 4.25	11.04 ± 3.86
Retest mean (±SD)	14.91 ± 4.99	11.38 ± 3.14	17.10 + 4.57	11.42 ± 3.41
Pearson correlation coefficient	.779**	.608**	.703**	.646**

Note. SURPS = Substance Use Risk Profile Scale.

^**p < .01;

^*p< .05.

Concurrent validity

Consistent with the results from the original study by Woicik et al. (2009), we found significantly low-moderate correlations in the range of .252–.589 between the SURPS subscales and other scales measuring the similar constructs (Table 3).

Table 3.

Correlation between the SURPS Subscales and Other Instruments.

Pearson correlation coefficient	SURPS
	Lack of self-contentment (formerly hopelessness)	Impulsivity	Sensation seeking	Anxiety sensitivity
BHS score	0.589**
BIS total score		0.530**
AISS score			−0.525**
STAI State subscale score				0.252**
STAI Trait subscale score				0.386**

Note. SURPS = Substance Use Risk Profile Scale; BHS = Beck Hopelessness Scale; STAI = State Trait Anxiety Inventory; BIS = Barratt Impulsivity Scale; AISS = Arnett Inventory of Sensation Seeking.

^**p<.01;

^*p<.05.

Discussion

The psychometric evaluation of the Turkish version of SURPS among 14- to 18-year-old adolescents attending two public high schools in Ankara supports the reliability and validity of this version and the cross-cultural use of the SURPS.

According to the present data, the factor structure of the Turkish version of the SURPS corresponded to the original (Woicik et al., 2009), including four dimensions: Lack of Self-Contentment (formerly Hopelessness), Anxiety Sensitivity, Impulsivity, and Sensation Seeking. The name “Lack of Self-Contentment” was suggested to replace “Hopelessness” for the subscale to make it include all of the intended items and maintain the original four-factor structure. This difference was raised from the contextual meaning triggered by Turkish language. In Turkish, expressions such as “I am content,” “I am happy,” “I feel proud of my accomplishments,” “I feel that I’m a failure,” and “I feel pleasant” are incapable of assessing negative expectancies that indicate hopelessness. In addition to the expressions regarding attitudes toward the future such as “I have faith that my future holds great promise” and “I am very enthusiastic about my future,” the aforementioned expressions appear to be closely related to the level of self-satisfaction.

Internal consistencies for the four subscales were all adequate to good, indicatingg that all of the items within each of the subscales acted as indicators of the same construct without redundancy. Items 22, 16, and 17 were carefully evaluated. The factor loading of item 22 (“I feel I have to be manipulative to get what I want”) on the Impulsivity subscale was interpreted as poor according to Comrey and Lee’s grading (Comrey & Lee, 1992); this was also noted with the Japanese and Spanish versions (Omiya et al., 2015; Robles-Garcia et al., 2014). Even so, item 22 was adopted as in the other validation studies and the original study (Canfield et al., 2015; Woicik et al., 2009). In terms of content in Turkish, manipulation of a situation or person for gain could be perceived as the result of more complex planning before action rather than acting without considering the result. Including item 22, all of the impulsivity items appear to measure deficits in behavioral response inhibition when an individual chooses immediate rewards in the environment despite longer term negative consequences (Woicik et al., 2009).

Only item 16 and item 17 cross-loaded on to another factor as well as the expected factor of the original SURPS. In the original study, Woicik et al. found item 16 to correspond to the intended factor (sensation seeking) by using both exploratory and confirmatory factor analyses (Woicik et al., 2009). However, Krank et al. and Castellanos-Ryan et al. removed this item in the other validation studies because it was not included in the proposed model as a sensation-seeking item (Castellanos-Ryan et al., 2013; Krank et al., 2011). As in the original SURPS, item 16 was included as a sensation-seeking factor in the exploratory factor analysis because the communality of item 16 was interpreted as good in comparison with the impulsivity factor according to Comrey and Lee’s grading (Comrey & Lee, 1992). Although they are known to be significantly related to each other, sensation seeking, or the desire for intense and novel experiences, can be distinguished from impulsivity, described as the inability to control behavior in the face of cues for reward and/or punishment (Conrod, Pihl, et al., 2000). While impulsivity reflects deficits in response inhibition, sensation seeking shows the tendency to seek out stimuli that enhance experience (Woicik et al., 2009). Nevertheless, the meaning of item 16 (“I am interested in experience for its own sake, even if it is illegal”) includes both a long-term punishment and a new experience.

The content of item 17 (“I feel that I’m a failure”) overlapped on negative affectivity (including anxiety sensitivity and lack of self-contentment [formerly hopelessness]) in the present study. Similarly, item 17 modestly cross-loaded onto anxiety sensitivity in the original study that high school students who ranged in age from 14 to 18 years participated in; it was preferred to be kept in the intended factor (hopelessness) (Woicik et al., 2009). Item 17 was also accepted in the Turkish version according to the original SURPS, although the factor loadings on both lack of self-contentment (formerly hopelessness) and anxiety sensitivity were interpreted as poor according to Comrey and Lee’s grading (Comrey & Lee, 1992). Both hopelessness and anxiety sensitivity have already been shown to be associated with higher depression scores in the previous studies (Newton et al., 2016). The statement “I’m a failure” can be interpreted as a negative core belief about oneself in depression.

Reliability was also demonstrated by a test–retest analysis that revealed significant relationships between test and retest scores. Correlations between the SURPS subscales and the other scales indicated that the Lack of Self-Contentment subscale (formerly Hopelessness) was related to the Beck Hopelessness Scale assessing feelings about the future, loss of motivation, and expectations, and the impulsivity subscale of the Barratt Impulsivity Scale that assesses the personality/behavioral construct of impulsiveness. Surprisingly, the Arnett Inventory of Sensation Seeking, including 20 items such as “If I were to go to an amusement park, I would prefer to ride the rollercoaster or other fast rides” and “I would like to travel to places that are strange and far away” was inversely related to the Sensation Seeking subscale. Sensation seeking can be expressed in a variety of ways, some culturally and some socially acceptable or unacceptable, depending on the extent to which the individual’s socialization environment guides or suppresses the sensation-seeking tendency. The reliable completion of the Arnett inventory may have been influenced by cultural and social factors in Turkish adolescents due to the various aspects of the items such as “I can see how it would be interesting to marry someone from a foreign country” and “I would never like to gamble with money, even if I could afford it.”

The results of this study should be considered in the light of several limitations.

First, the predictive and convergent validity could not be assessed in the present study. The relationship between the personality traits and substance use have already been shown both concurrently and prospectively in adaptations of SURPS in different languages and cross-cultural samples (Canfield et al., 2015; Castellanos-Ryan et al., 2013; Castonguay-Jolin et al., 2013; Chandrika Ismail et al., 2009; Jurk et al., 2015; Krank et al., 2011; Malmberg et al., 2010; Newton et al., 2016; Omiya et al., 2015; Robles-Garcia et al., 2014; Saliba et al., 2014; Schlauch et al., 2015; Siu, 2011; Woicik et al., 2009). Although the present study provides robust evidence to support the sensitivity of SURPS in identifying adolescents at high risk by measuring the personality traits related to substance abuse, further studies assessing its predictive and convergent validity in Turkish samples are necessary to develop early interventions against the personality dimensions linked to specific drug abuse.

Second, the design of the present study involved a total school sample with a completion rate of 81.5%. The variability of the personality traits cannot be fully appraised. Longitudinal studies are needed to address variability in personality traits and subsequent development of substance abuse.

Third, the data were collected through self-reports, which can be problematic. Nevertheless, given assurance of confidentiality, assessing only personality dimensions instead of substance use directly seems to be a reasonable approach.

Fourth, the results comprise two school samples in a metropolitan area in the capital of Turkey and cannot necessarily be generalized to other regions and in particular more rural provinces in the country. Further studies should be planned to evaluate additional samples, including individuals who use substances.

Despite these limitations, the current study provides strong support for adequate reliability and validity of the Turkish version of the SURPS, which appears to be a promising measure for assessing adolescent personality dimensions related to substance abuse risk and for future use in prediction and prevention of substance use among high-risk youth in Turkey.