The Trauma Symptom Inventory, Second Edition (TS1-2), developed by John Briere, was published in 2011 by the Psychological Assessment Resources, Inc. The revision of the Trauma Symptom Inventory (TSI) initially developed by Briere in 1995 was meant to update the inventory to reflect changes in the field. The TSI-2 provides a more comprehensive criteria of trauma related symptomatology that expands on the original TSI. Moreover, Briere was mindful to provide downward compatibility with the TSI. Both the TSI and TSI-2 also includes an alternate form-A (TSI-2-A) that does not include items that describe sexual symptomatology. However, at this juncture it appears that the TSI is no longer available to purchase. The manual for the TSI-2 offers data on the construction of the inventory, including norming sample, test-retest reliability and internal consistency; as well as evidence for its concurrent, criterion, discriminant, factorial, and construct validity (Briere, 2011).
The introductory TSI-2 test kit comes with the professional manual, a package of 10 reusable item booklets, 25 hand-scorable answer sheets, and 25 profile forms; it can be purchased for $229.00. The TSI-2-A introductory test kit comes with all the TSI-2 professional manual, but includes the alternative versions for the hand-scorable answer sheets and profile forms. It costs the same price and the TSI-2. Test components can also be purchased separately. The professional manual is available in print format and electronically for $69.00. The other test components are available in either version (TSI-2 and TSI-2-A) in the same package quantities that come standard with the introductory test kit. The reusable item booklets (10/pkg) cost $55.00, hand-scorable answer sheets (25/pkg) are $69.00, and profile forms (25/pkg) are $55.00. These items are also available in Spanish versions for the same quantities and price. Lastly, the TSI-2 is available in either downloadable software or CD-ROM for $394.00. Examiners may also use PARiConnect i-Admin ($4.50 per use; minimum order of 5) and Score Report ($4.00; minimum order of 5) for online administration, either in office or remote; as well as scores and profiles from Score Report based on data recorded via i-Admin (Psychological Assessment Resources 2017).
General Nature of the Test
Purpose and Conceptual Basis
The TSI-2 is a test that measures that identifies acute and chronic symptomology relating to trauma in adults, ages 18 years and older. Trauma can occur from exposure to many adverse and traumatic experiences, including childhood maltreatment, physical and sexual assault, intimate partner violence, combat exposure, torture, vehicular accidents, natural disasters and mass casualty events, trauma relating to a medical condition, and witnessing a traumatic event either directly or vicariously, and experiencing a traumatic loss (Briere, 2011). While the TSI-2 was created under the criteria relating to the DSM-IV, the most recent version of The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5; American Psychiatric Association, 2013), states that psychological distress relating to exposure from trauma exhibits a multitude of symptoms, ranging from anxiety and fear-based presentations, depressive symptoms (i.e., anhedonia and dysphoria), externalizing behaviors (i.e., anger and aggression) and dissociation. Clinically significant levels of symptoms meeting diagnostic criteria can be classified into disorders depending on the context and presentation, such as reactive attachment disorder, disinhibited social engagement disorder, and posttraumatic stress disorder (PTSD), acute stress disorder, and adjustment disorders.
Briere (2011) states that in North America, it was estimated that approximately 8% of the general population experiences a lifetime prevalence of posttraumatic stress disorder (PTSD); however, for certain populations, such as combat veterans, victims of sexual assault, torture survivors, refugees, and other individuals who experience an extreme or repeated stressor, increased rates can range from 30% to 60% prevalence. However, nonspecific PTSD related sequelae attributed to interpersonal victimization for both children and adults can include disturbances in mood, (i.e., anger, anxiety and depression), emotional dysregulation, somatic complaints, identity disturbance, disrupted and insecure attachment styles, chronic interpersonal difficulties, dissociation, externalizing or tension reduction behaviors (i.e., compulsive sexual behavior, aggression, bulimic eating, and self-mutilations), as well as suicidal ideation and acts (citations). Moreover, individuals are at-risk for complex traumas (e.g., complex PTSD) if exposure to mediating event was early, or repeated; as well as if they were exposed to a later traumatic event (citation). Briere (2011) also states that given the diverse range of clinical presentations, utilizing a test that only addresses PTSD related sequelae, or a singular symptom or disorder will likely be limited in its ability to depict an accurate clinical profile of a trauma survivor. Moreover, most published and standardized tests for trauma do not capture the full range of outcomes, and are most applicable to only assessing specific criteria relating to a disorder, such as PTSD. Given these implications, the need for a broadband measure was needed to capture a more comprehensive trauma symptom profile of an individual. The TSI was originally constructed for this purpose; therefore, developing the TSI-2 was meant to revise and expand on the existing TSI subscales to reflect greater symptom coverage (Briere, 2011).
Application and Population
The TSI-2 was developed to be a broadband assessment tool that can be used to detect trauma related sequelae in adults, ages 18 years and older. Its application is appropriated for a variety of clinical settings, including hospitals, both inpatient and outpatient clinics, and schools. Being a comprehensive test that captures symptomatology relating to both early and later exposure to trauma, it is appropriate for use with insecure attachment, maladaptive self-reference, somatization, and externalizing behaviors (e.g., acting out); since it can capture multiple presentations of exposure to traumatic experience, it is appropriate to use for more conditions aside form PTSD. Additionally, the test was standardized and validated on the normal population using both men and women, with normative data available for different variations of both gender and age; therefore, the test is appropriate for use with a variety of combinations of age and gender for adults, ages 18 years and up (Briere, 2011).
General Type Test
The TSI-2 is a diagnostic tool that measures both potentially acute and chronic symptomatology relating to exposure to trauma in individuals, ages 18 years and older. The TSI-2 regular form consists of 136 items; whereas, the alternate form, TSI-2-A is a 126-item inventory that does not contain any of the factors relating to sexual symptoms; both forms require a fifth-grade proficiency level. The TSI-2 is constructed to evaluate the extent that a respondent will endorse 12 different types of trauma related symptomatology; six of these clinical scales also have scoreable subscales. In turn, these clinical scales are grouped together to make up four broad categories, or factors of distress: Self-Disturbance (SELF), Posttraumatic Stress (TRAUMA), Externalization (EXT), and Somatization (SOMA). (Briere, 2011).
The TSI-2 can be administered in both individual and group settings; it takes approximately 20-30 minutes for respondents to complete. Examinees should be given a calm and quiet area to test in. While it was not normed to be broken up and taken in multiple sessions, if the respondent becomes distressed or is distracted during test taking, it’s permissible to allow small breaks. However, notes should be included since whenever the test is not administered according to standardization. Due to the nature of the content, it’s advisable that when being used in a group setting, enough physical space should be given to the examinees so that their answers can remain private. After the administration of scoring, the respondent sores associated with clinical and subscales can be categorized as normative, problematic, or clinically elevated. Lastly, this TSI-2 can be utilized in tracking any changes in degree of symptomatology for the same individual using the TSI-2 change score summary table, included on the back of the profile form. This could be particularly helpful in tracking and detecting any reductions in symptoms if the respondent undergoes clinical intervention for their presenting problems (Briere, 2011).
The complete test kit provides the administrator with a professional manual, hand-scoreable answer sheets and profile forms. The TSI-2 is a 136-item format and the alternate form, TSI-2-A, has 126 items. Items on both forms ask respondents to rate their degree of self-reported symptoms during the previous six months using a Likert scale; item questions do not reference any specific trauma. Scale scores range from 0 (never) to 3 (often), with higher scores indicating a greater degree of distress and symptomatology. An examinee’s raw scores are then converted to T-scores and corresponding percentiles to depict an appropriate clinical picture of their psychological functioning. Moreover, the TSI-2 contains eight critical items (TSI-A includes seven) that denote clinical significance that may warrant further investigation if endorsed by the respondent with scores that are higher than zero (i.e., 1, 2, 3); these may signify severe psychological distress that could result in harm to self or others (Briere, 2011).
The TSI-2 has 12 clinical scales, six of which have their own subscales. As per the test manual, these clinical scales are further organized to produce a four-factor model of distress, as identified by using confirmatory factor analysis (CFA; be later discussed): Self-Disturbance (SELF), Posttraumatic Stress (TRAUMA), Externalization (EXT), and Somatization (SOMA). The SELF scale measures difficulties with one’s own self-awareness, which is usually negative; TRAUMA captures PTSD related stress, anxiety, and dissociation; EXT evaluates one’s inclination to engage in maladaptive behaviors; and SOMA denotes somatic complaints, preoccupation, and distress (Briere, 2011).
Each factor scale is comprised of one or more clinical scales. The factor scale SELF, includes the clinical scales Depression (D) and Insecure Attachment (IA), which measure typical symptoms of depression (e.g., affect, cognition, and somatic), and difficulties with interpersonal relationships. Factor scale TRAUMA incorporates clinical scales Anxious Arousal (AA), Intrusive Experience (IE), Defensive Avoidance (DA), and Dissociation (DIS), all of which measure common symptoms and criteria of a PTSD diagnosis. Factor scale EXT, includes clinical scales Anger (ANG) and Sexual Disturbance (not included on the TSI-2-A), Suicidality (SUI), and Tension Reduction Behavior (TRB). These scales measure different types of externalizing behavior that can be potentially harmful to the individual, such as aggression, risky sexual behavior, self-harming behaviors, and suicidal ideation and acts. Finally, factor scale SOMA, includes the clinical scale, Somatic Preoccupation (SOM), which measures preoccupation with somatic concerns and distress (Briere, 2011).
Moreover, six of the clinical scales include subscales that further define and categorize symptoms relating to their construct. Clinical scale IA, includes subscales Relational Avoidance (IA-RA) and Rejection Sensitivity (IA-RA), which measure avoidance of relational intimacy and dependence, and preoccupation with rejection or abandonment. Clinical scale ISR has two subscales, Reduced Self-Awareness (ISR-RSA) and Other-Directedness (ISR-OD). These subscales measure factors, such as one’s disrupted thoughts and beliefs towards one’s own self, and associations with childhood maltreatment, and other traumas. Clinical scale AA, includes the subscales, Anxiety (AA-A) and Hyperarousal (AA-H), which measure specific symptoms of anxiety and posttraumatic hyperarousal, such as hypervigilance. Clinical scale, SXD has two subscales, Sexual Concerns (SXD-SC) and Dysfunctional Sexual Behavior (SXD-SC), that evaluates negative thoughts about sex and problematic behavior; these are not included on the TSI-2-A. Clinical scale SUI has the subscales, Ideation (SUI-I) and Behavior (SU-B) that denotes what kind of suicidality the examinee has engaged in (e.g., thoughts and behavior). Lastly, clinical scale, SOMA, has two subscales, Pain (SOM-P), and General (SOM-G) which measures the type of somatic complaints that are experienced (e.g., specific pains and generalized complaints (Briere, 2011).
Lastly, the TSI-2 has two validity scales, the Response Level (RL) and the Atypical Response (ATR) scales. The RL scale detects inconsistencies with reporting when respondent answers in a random manner. The scale contains items that are unlikely to receive a zero rating in a normative sample. Additionally, this could be attributed to an individual responding in a random manner or underreporting concerns with symptoms (e.g., “faking good”); conversely, the ATR scale indicates the degree that an individual may have overreported concerns with symptoms (e.g., “faking bad); this may be related to malingering, or could be seen as “a cry for help” (Briere, 2011).
Administration and Scoring
For convenience, the TSI-2 can be administered to either an individual or to a group of examinees. Respondents are given a pencil, flat writing surface, test item booklet and hand-scorable answer sheet to record their answers; examinees should not write in the item test booklets. The testing environment should be well-lit and free from distractions. If administering the test to a group, the examiner should orally provide instructions for the test. Additionally, there should be enough space between examinees so that their answers remain confidential. Respondents may take a brief break in the event they become distressed or distracted. The examiner should be prepared to answer any questions that may arise during testing Items on the TSI-2 refer to a variety of phenomena which ask the examinee to record the degree of occurrence with the item presented using a 3-point Likert scale: 0 (never), 1 (rarely), 2 (sometimes), and 3 (often); greater scores indicate more difficulties with trauma related sequalae. Once a respondent has completed the TSI-2, the examiner should scan the answer sheet for any missing items or multiple responses. The respondent should be prompted to fill out unanswered items and clarify items with multiple questions whenever possible; too many unanswered or unscorable items deems the protocol invalid. Typical time to complete the test is between 20 to 30 minutes (Briere, 2011).
Scoring for the TSI-2 is a relatively straightforward process. Perforated and carbonless answer sheets allow transfer of demographical information and scores to the scoring sheet underneath. As per the professional manual, the following steps are recommended for scoring:
- Using the scoring sheet with the transferred scores, sum up circled responses for each row across, except for the RL scale and the Pain (SOM-P) and General (SOM-G) subscales.
- For the RL scale, sum all circled responses indicating 0 in that row.
- For the SOM-P subscale, sum responses that are not shaded
- For the SOM-G subscale, sum all responses that are shaded, plus the remaining shaded responses for SOM-P.
- From there, the examiner will raw score calculations using the table located at the bottom of the answer sheet; subscale scores are summated to determine the clinical scale scores; then, the clinical scores are summated for overall all factor scores. Final raw scores are recorded on the summary sheet.
- Corresponding T-scores and percentiles are found in conversion charts categorized by gender and age for males age 18-54 years, males age 55-90 years, females 18-54 years, and females 55-90 years. Using the appropriate chart (e.g., gender/age by T-score/percentiles), raw scores are plotted for each scale (i.e., validity, factor, clinical and subscale) with their corresponding T-scores and percentiles.
Computer scoring is available via the TSI-2 software or PARiConnect i-Admin (onscreen test administration) (Briere, 2011).
The TSI-2 is a multidimensional tool designed to measure the broad spectrum of behaviors and psychological functioning attributed to trauma related sequalae. Higher raw scores and higher T-scores indicate a greater a degree of symptomatology experienced by the respondent. For the purposes of classification and clinical significance, T-scores are divided into score ranges. Based on the normal distribution of the population, a T-score of 50 (T = 50) is the population mean. A T-score of 60 (T = 60) indicates one standard deviation (SD = 10) above the mean; a T-score of 65 (T = 65) indicates 1.5 standard deviations above the mean. Therefore, classifications for T-sores create three ranges: normal (T < 60), problematic (T = 60–64), and clinically elevated (T > 65). For example, T-scores classified as problematic may not be clinically elevated, but signifies an above average endorsement of symptoms that are likely to have clinical implications. In other words, an individual endorsing problematic symptoms suggests some impairment of the individual’s functioning. Likewise, clinically elevated scores indicate severity of symptoms that warrants a significant clinical concern, and is likely impacting the individual’s psychosocial functioning. Moreover, the TSI-2 contains eight critical items, that if endorsed by the respondent may signify severe psychological disturbance and should be heeded by the clinician/examiner, as these could represent a danger to oneself or to others. It should be noted that critical items vary slightly between the TSI-2 and TSI-2-A. If any critical items are endorsed above a zero (i.e., 1-3), then the clinician should be prepared to intervene if necessary. Further probing should be conducted to assess the reasons behind the endorsement and potential repercussions, particularly items associated with the Suicidality scale. Moreover, other elevated responses relating to other items belonging to the scales, Suicidality, Tension Reduction Behavior, or Sexual Dysfunction, should be of note to the clinician and may warrant further exploration (Briere, 2011).
The TSI-2 is also constructed with two validity scales that can alert the clinician to potential issues regarding whether results on a protocol should be inferred. The Response Level (RL) scale is intended to measure inconsistent answers, such as marking zeros for items that are unlikely to receive zeros from other respondents, or the tendency for a respondent to deny concerns with pathology (e.g., “faking good). Those who are oppositional towards test-taking, or otherwise may be avoidant or defensive are likely to excessively mark items as zero. Based off the standardized sample, those who endorsed RL items with T-scores greater than 70, were among the top 4%. For the purposes of interpretation, it’s recommended that protocols receiving a T-score greater than 75 be considered invalid.
The Atypical Response (ATR) scale indicates whether a respondent has overreported concerns with pathology (e.g., “faking bad”). Very high scores on this scale may represent a generalized overendorsement of trauma-related symptoms, symptomatology specific to PTSD, random response patterns that overendorse items that are not typically endorsed, or extreme levels of distress. While high scores could denote a possibility of factious disorder or malingering in clinical contexts, clinicians should consider that the individual’s experience may be presented as more extreme to facilitate efforts in obtaining clinical assistance (e.g., “cry for help”) (Briere, 2011). It should be noted that victims of intense trauma or child abuse may yield higher scores on scales measuring “faking bad” or infrequency type scales, due extreme or atypical symptomatology sometimes connected to posttraumatic disturbance (CITATIONS).
For interpretation of the ATR scale, Briere (2011) recommends that raw scores greater than 15 should invalidate a protocol if used in clinical and forensic settings. If used on general populations (e.g., nonclinical populations) for research or screening, then a raw score of 8 should be used (CITATION). Lastly, it may be helpful to evaluate elevated ATR responses with respondents to discuss how the item in question relates specifically to the context of their situation. For example, a highly traumatized, retuning combat veteran may accurately report their experience by endorsing TSI-2 item 14 (“Feeling so irritable after a trauma that you got into physical fights with strangers”); whereas, many other respondents would only engage in verbal altercations. In situations like these, it’s permissible for the clinician to adjust the protocol by substituting the item score with the mean (round to the next whole number) of the remaining ATR scale items and calculate a new raw score for the scale. However, this should only be done for only one item belonging to the ATR, and only after consideration about whether a viable explanation exists for the elevated item in question. Of note, this method of “average score minus one item response” may still be quite high depending if multiple ATR responses were endorsed. If using the Scoring Program, it’s recommended that the decision is made to substitute the mean score made prior to entering data because this option is not supported by the software; therefore, the revised ATR should be entered at the time of initial entry of data (Briere, 2011).
As per the professional manual, formal training in either clinical or counseling psychology, or a closely-related field is not needed for an individual to administer the TSI-2, if the examiner has proficiently familiarized themselves with the scoring guidelines in the professional manual to ensure that procedures conducted result in a normative administration of the test. A licensed or certified agency that requires proper training and experience in the ethical and competent use of psychological tests is also qualified. The website, Psychological Assessments Resources, where the TSI-2 can be purchased, provides a more detailed description of examiner qualifications, gives the test a B-rating. A B-rating indicates that qualifications entail a degree from a 4-year accredited educational institution, with a degree in psychology, counseling, speech-pathology or other closely related field; coursework in the statistics, test interpretation, measurement theory, or other closely related courses are also required (Psychological Assessment Resources, 2017). Conversely, to interpret the results of the TSI-2 as per the manual, one must have a received professional training in psychology, counseling, or other closely related field. Likewise, the aforementioned criteria regarding licensed or certified agencies is also permissible to interpret test results (Briere, 2011).
More than five years in the making, the TSI-2 is a significant revision from the first edition of the TSI. Briere’s goal was to address feedback from other professionals in the field (e.g., clinicians, researchers) regarding the test’s potential strengths and weaknesses. Revising the TSI was meant to reflect new developments elucidated in trauma-related research by updating the existing items of the scale and adding newer scales that would capture a wider range of symptomology not measured by the first edition, but are significant factors relating to posttraumatic complexities (Briere, 2011). Furthermore, in forensic applications, some concerns were raised about the original ATR scale’s utility in identifying PTSD malingering (e.g., Efendov, Sellbom, & Bagby, 2008; Elhai et al., 2005); however, Edens, Otto, and Dwyer (1998) did not find this to be the case. While the original ATR was not created specifically for this application (i.e., detection of malingering in PTSD symptoms), the ATR scale for the TSI-2 was redeveloped to specifically identify misrepresentation of PTSD symptoms; as well as a more comprehensive measure of overreporting for trauma symptoms. Finally, the inclusion of a subscale for the Anxious Arousal scale for the TSI-2 was planned to specifically measure posttraumatic hyperarousal because the original scale failed to fully capture autonomic activation associated with anxiety (Briere, 2011).
During the development of the TSI-2, while it was hoped that the TSI-2 would be an improved tool in measuring trauma related sequelae compared to the original TSI, it was also important to provide good downward compatibility with the original TSI. In evaluating both existing and new domains of trauma, the goal was that equivalent scales for both the original TSI and the TSI-2 continue to tap into the same constructs (e.g., same symptom domain); thus, providing high correlations between the scales from the TSI-2 to the TSI (Briere, 2011). Briere (2011) states that the process of adjusting the content and psychometrics of the TSI-2 took considerable attention and ongoing evaluation to assure that that newly developed content remained true to the original TSI measure.
During development, items written for new scales and subscales, newly written replacements for existing TSI items, and items carried over from the original test consisted of a total of 233 possible items for the TSI-2. Additionally, items pertaining to the original TSI Response Level and Atypical Response scales were entirely rewritten. As previously mentioned, the ATR needed to be more sensitive in detecting malingering in PTSD symptoms or overreporting. Moreover, a Sleep Disturbance scale was created, but was ultimately dropped due to complicating the utility of the TSI-2 relative to its expected usefulness. From there, 83 items were discarded due to being poorly written, redundant compared to other preferred items, or did not accurately represent the item content domain. This culminated in 150 items that were administered to a standardized sample to assess for internal consistency in their respective scales. Finally, items deemed unnecessary or redundant were omitted, resulting in a final measure of 136 items. While psychometric reliability was considered paramount, each item was analyzed for its usefulness in contributing to the capturing the construct for its associated scale. Even though it slightly reduced the internal consistency of a scale, some items were retained because it evaluated an important characteristic of a given symptom construct (Briere, 2011).
Using a standardized sample, a structural hypothesis was tested using confirmatory factorial analysis (CFA) to evaluate the contribution of the newly developed scales and subscales. First, it was predicted that the Trauma factor from the original TSI would be replicated in the TSI-2, but would also contain the new subscales created for the Anxious Arousal. Secondly, it was hypothesized that the TSI Self factor might split into two new factors in the TSI-2: the new Insecure Attachment subscale and Depression scale might load onto a new self or identity factor, and the new Suicidality and Disturbance subscales, along with the anger scale, might load onto an externalization factor. Third, it was thought that the Dysphoria factor belonging to the TSI would be eliminated, because the Anxious Arousal subscales would be incorporated into the TSI-2 TRAUMA factor; as well as, the Depression and Anger scales would be included in the TSI-2 EXT and Self factors. Lastly, it was theorized that the new TSI-2 Somatic Preoccupations scale would form its own factor or load with the TRAUMA factor (Briere, 2011).
According to Godbout et al. (2010a, 2010b), versions of the structural hypothesis were examined using associated fit indexes, Lagrange multiplier, and Wald’s test. CFA indicated the model of best fit was that of the hypothesized SELF, TRAUMA, EXT, and SOMA factor scales (Chi-square ratio = 3.56; Comparative Fit Index (CFI) = .94). It’s also stated that no other significant unexplained variance existed, as per the Root Mean Square Error of Approximation (RMSEA = 0.06). Additionally, all path coefficients belonging to the factors were found to be statistically significant (p < 0.001). Lastly, to validate the generalizability of the four-factor model from the standardize sample to those of belonging to the general population, Godbout et al. (2010a, 2010b), conducted CFAs of the TSI-2 in both a university and online subsample belonging to a larger university sample. Results indicated that the same four-factor model was also a good fit for each subsample (Briere, 2011).
Psychometric evidence of reliability for the TSI-2 was conducted through several analyses to determine its degree of temporal stability, consistency, and accuracy. Cronbach’s alpha (coefficient alpha) a measure of internal consistency, describes the extent that items on the test all measure the underlying construct; if all items are tapping into the same construct, then they should correlate highly with each other. To evaluate this, the TSI-2 was administered once to a standardized sample. The alpha coefficients indicate high levels of internal consistency among the clinical scales and subscales that ranged from 74 to .94. Three subscales were found to be at the lower end of the that estimate (αs = .74–76), with the remaining clinical and subscales beginning at .80 for the next lowest coefficient alpha. The validity scales, RL (α = .81) and ATR (α = .72) showed slightly lower alphas, but were still determined to be very good. The validity scales function as a measure of bias or error variance, therefore lower alpha coefficients were expected. Finally, TSI-2 four- factor model also indicated very good internal consistency in the standardized sample: SELF (α = .93), TRAUMA (α = .93), EXT (α = .86), and SOMA (α = .77) (Briere, 2011).
Another measure of internal consistency for the test consists of evaluating the mathematical average of the item-total correlation for each item belonging to a scale. An item-total correlation discerns whether an individual item is consistent with the rest of the items of the scale by correlating its score with the overall test score (minus the item score). The average item-total correlations using the standardized sample were found to be in the moderate or desirable ranges. Average item-total rs for the validity, clinical scales and subscales ranged from .61 to .84, with average scale correlations for the four-factor model also in the moderate to desirable range: SELF (r = .80), TRAUMA (r = .88), EXT (r = .72), and SOMA (r = .91) (Briere, 2011).
Finally, temporal stability refers to the concept that a test will produce the same results if given to the initial sample at a later point in time. This is used to evaluate if the TSI-2 would produce the same measurements for respondents if they were to take the test again a various timepoints. Using a subset of the standardized sample (n = 31), the TSI-2 was administered at two different times. The target interval was determined to be one week; testing times ranged from 1–14 days (M = 5.19 days, SD = 2.81 days). According to the results, all but one of the scales was determined to have excellent test-re-test stability (p < .01). The result for the low correlation for Suicidality-Behavior subscale (r = .15, ns) was better explained by the fact that in nonclinical populations, there is a low prevalence of suicidal behavior. The four-factor model for the TSI-2 was also found to have high test-retest stability: SELF (r = .96), TRAUMA (r = .93), EXT (r = .91), and with SOMA having a slightly lower correlation (r = .84) (Briere, 2011).
Validity is a multi-faceted concept that involves the evaluation of several sources of evidence; a test is only considered valid when it accurately measures the construct in question. First, the manual cites psychometric evidence for the convergent/discriminant validity of the TSI-2 through a series of analyses with several samples. One way this was assessed was how well the factors, clinical, and subscales of the revised TSI-2 correlated with their counterparts associated with the original TSI. To begin with, a sample of university students (University of California-Riverside) (N = 477) were administered both the original TSI and the TSI-2. It was hypothesized that scales tapping the same construct would correlate highly with each other (e.g., depression scales for both should show a strong correlation). All correlations were found to be statistically significant (p = .01) and indicated high correlations between original scales of the TSI and their revised counterparts on the TSI-2; for example, Defensive Avoidance (r = .98), Intrusive Experiences (r = .96), Anger (Anger/Irritability on TSI) (r = .94), and Impaired Self-Reference (r = .94). Moreover, newly developed subscales for the TSI-2 strongly correlated with the original parent scales; as well as the TSI-2 factors correlating highly with original TSI factors; likewise, high intercorrelations were also found in the standardized sample (N = 678) and statistically significant (p < .01). Moreover, concurrent measures, such as the Major Depression Inventory (MDI), the PTSD Checklist-Civilian Version (PCL-C) (Weathers, Huska, & Keane, 1991), Impact of Event Scale-Revised (IES-R) (Marmar, Weiss, & Metzler, 1997), and the Borderline (BOR) scale items from the Personality Assessment Inventory (PAI) (Morey, 1991) were administered in a combined clinical validity (CCV) sample (N = 125) to assess the TSI-2 scales’ correlational strength among the associated constructs between the tests. The TSI-2. Evidence of strong correlations were found between the TSI-2 and those of similar constructs belonging to other scales, and were statistically significant (p < .01). Moreover, in the standardized sample (N = 678) and a different university sample (University of Victoria; N = 1,051), the PTSD subscales belonging to the TSI-2 (i.e., Anxious Arousal scale, the Intrusive Experiences and Defensiveness subscales) showed correlational evidence with equivalent scales belonging to the PCL-C (i.e., Reexperiencing, Avoidance, and Hyperarousal scales), respectively. Moreover, using the University of Victoria sample, two of the newly developed subscales, Somatic Preoccupations, and Insecure Attachment were evaluated against the Health Symptoms Checklist (HSC) (Runtz, 2002), and the subscales (i.e., Avoidance and Anxious) belonging to the Experiences in Close Relationships-Revised scale (Fraley, Waller, & Brennan, 2000), respectively. Results for the Somatic Preoccupations scale of the TSI-2 indicated correlational evidence of concurrent and discriminant validity with the total score of the HSC. The subscales, Relational Avoidance and Rejection Sensitivity, belonging to the Insecure Attachment scales of the TSI-2, also indicated strong correlations with the subscales of the ECR-R. The manual posits that the body of aforementioned analyses support the convergent and discriminant validity for the TSI-2 factors and scales (Briere, 2011).
As per the manual, criterion validity for the TSI-2 was assessed for its ability to accurately distinguish individuals with purported symptomatology associated with trauma. The criterion validity was assessed by comparing scores produced by the TSI with various groups; as well as analyzing the interrelationships between the scales and external variables of criterion. Since the TSI-2 was developed to be an instrument that measures trauma related sequalae, the scales were assessed in terms of their association with exposure to trauma and variation among several groups. Populations used to establish criterion related validity were combat veterans, sexual abuse victims, intimate partner victims (i.e., domestic violence), and clients who had a diagnosis of borderline personality disorder (BPD).
For those in the standardized sample (N = 678), participants with a history of trauma endorsed more items on the TSI-2 than those with either no history or exposure to at least one event, for the exception of the sole non-discriminating variable, the Dysfunctional Sexual Behavioral subscale belonging to Sexual Disturbance scale. Construct validity of the TSI-2 was supported by the broad correlations between the trauma related symptom scales and trauma exposure. The complexities of trauma were highlighted in the small to moderate correlations associated with the relationship between past trauma and the TSI-2 factors, scales and subscales, exemplifying that many variables are likely to serve as mediators and moderators of trauma outcomes (Briere, 2011; Briere & Jordan, 2009).
The criterion validity was also assessed by performing independent t-tests for TSI-2 scores between the combat veteran group, sexual abuse and domestic violence victims, and individuals with a diagnosis of BPD, to control groups obtained from the standardized sample. Significant differences were found between the associated control groups and clinical groups for the results on the Response Level validity scale, except for each clinical group scored higher on the Atypical Response scale. In combat veterans, expected patterns of high endorsements for items relating to Anxious Arousal-Hyperarousal, Anxious Arousal-Anxiety, and Intrusive Experiences were synonymous with symptoms that are central to combat-related PTSD. Additionally, those with a borderline personality diagnosis fit the expected patterns of endorsing items pertaining to Insecure Attachment- Rejection Sensitivity, Depression, and Anger scale/ subscales; as well as the Externalizing factor being quite high. Notably, scores for the Suicidality scale were well into the clinically elevated rang, with the Suicidality-Behavior subscale having the highest score (M = 73.3, SD = 18.60; M = 75.5, SD = 21.08); these results are aligned with theoretical, empirical, and diagnostic criteria associated with BPD. Moreover, sexual abuse victims generally scored in the hypothesized direction compared to its associated control group, endorsing items relating to the Intrusive Experiences, Tension Reduction Behavior, and Suicidality scales. In general, sexual abuse victims score higher on all factors except for SOMA than controls. Lastly, Guyton and colleges (2010) administered to the TSI-2 to a female prison population (N = 125) and found a significant difference between controls relating to both validity scales (i.e., Response Level, Atypical Response). Furthermore, higher scores were also found on the TRAUMA factor, Defensive Avoidance, Intrusive Experiences, Dissociation, Depression and Tension Reduction Behavior scales; as well as the Anxious Arousal-Hyperarousal, Sexual Disturbance-Sexual Concerns, Sexual Disturbance-Dysfunctional Sexual Behavior subscales. In sum, the TSI-2 is reported to have good evidence in supporting criterion validity by the number and degree of statistically significant differences between the various clinical samples used compared to their respective controls (Briere, 2011).
The professional manual also states evidence of predictive validity. In assessing whether the TSI-2 scales tapping into PTSD constructs can predict actual cases of diagnosed PTSD (according to criteria based on the DSM-IV). A discriminant analysis was conducted using T-scores for the TSI-2 related PTSD scales (i.e., Anxious Arousal-Hyperarousal, Intrusive Experiences and Defensive Avoidance) to evaluate if the TSI-2 could predict PTSD based on the PTSD Checklist (PCL). According to the PCL, scores indicative of PTSD are equal to or greater than 50 (Weathers, Litz, Herman, Huska, & Keane, 1993). As per the manual, the discriminant analysis elucidated that an optimally weighted combination of posttraumatic stress related TSI-2 scales significantly predicted PTSD. Furthermore, a receiver-operator analysis (ROC) of the TRAUMA factor of the TSI-2 to assess its predictive strengths. The area under the curve (AUC) indicates how well the TSI-2 can identify a given condition (e.g., PTSD). The manual states that the analysis presented indicated that the TRAUMA scales were a strong approach to predicting criteria of PTSD, as per the PCL. Moreover, predictive capabilities of the ATR validity scale were analyzed in both the standardized sample and a university student sample (Gray, Elhai, & Briere, 2010); University information is undisclosed, but denotes affiliation from two regionally affiliated Midwest and Western institutions (N =124). Results from the standardized sample indicated that then TSI-2 could predict overreporting of PTSD symptoms, but also an overall general measure of trauma related overreporting; particularly relating to the four factor scales. Finally, using the university sample was divided into a group trained to feign PTSD symptoms, and a group who legitimately met criteria according to the PCL. discriminant analysis indicated that the ATR could discern between malingered PTSD from genuine responding. Together these analyses indicated evidence for predictive validity (Briere, 2011).
The TSI-2 was normed on a standardized sample reflective of demographical attributes that were synonymous with characteristics associated with the 2007 census for adults between the ages of 18 and 90 and years of age. Respondents were recruited through use of a sampling company contracted by the test-kit publisher, PAR. Exclusion criteria to participate in the study were currently incarcerated, residing in an impatient psychiatric or medical facility, under care for a psychotic disorder (e.g., schizophrenia), had hearing loss or uncorrected vision, could not understand and comprehend English or read at a third-grade level, and lastly, were unable to provide informed consent. The final sample reflected 678 participants whose characteristics closely paralleled the age, gender, ethnicity/race, and educational demographics of the US census. Raw scores of the standardized sample were converted to corresponding T-scores and percentile ranks. Using classical testing theory, standard error of measurements (SEMs) were calculated for each TSI-2 factor, scale and subscale for the normative sample (Briere, 2011).
For the validation studies, samples were derived from college students, University of Victoria sample (N = 1,051), University of California (N = 477), and two medium sized universities from the Midwest/Western regions (names undisclosed; N = 124); as well as a combined clinical sample (N = 125), and a sample of incarcerated women (N = 125) residing in a US west coast women’s penitentiary. Lastly a subset sample from the standardized sample was also utilized (n = 31) for the assessing temporal stability of the TSI-2 (Briere, 2011).
Critiques by Others
Research specifically evaluating the TSI-2 remains scant even though it was published six years ago. One possible theory is that psychometric evidence was elucidated that affirmed that downward compatibility between the TSI-2 and the original TSI. Respectively, given the similarities between the scales, it appears much of the research associated with the TSI-2 utilizes the test as part of a larger battery of instruments for the given study. Gray et al. (2010) conducted research to evaluate if the ATR scale was sensitive enough to discern PTSD malingering compared to those who genuinely struggled with PTSD related symptomatology. However, these results were included in the professional test manual as evidence for predictive validity for the ATR scale (and will not be discussed again here). While psychometric evidence remains scant, a recent study reexamined the structural properties for the TSI-2.
Godbout, N., Hodges, M., Briere, J., & Runtz, M. G. (2010a). TSI-2 Structural Analyses on three samples: The dimensions of trauma symptoms–A measurement study. Unpublished manuscript.
Godbout, N., Hodges, M., Briere, J., & Runtz, M. G. (2010b). Confirmatory Factor Analysis of the Trauma Symptom Inventory-2: A four factor model. Poster presented at the annual meeting of the International Society for Traumatic Stress Studies. Montreal, Quebec, Canada.
Guyton, M. R., Brown, C. S., Hinman, J., & Stotler-Turner, E. (2010). Women’s adjustment to prison study. Unpublished manuscript.
Briere, J. (2011). Trauma Symptom Inventory-2 (TSI-2): Professional manual: PAR.
Briere, J., & Jordan, C. E. (2009). Childhood maltreatment, intervening variables, and adult psychological difficulties in women: An overview. Trauma, Violence, & Abuse, 10(4), 375-388.
Edens, J. F., Otto, R. K., & Dwyer, T. J. (1998). Susceptibility of the Trauma Symptom Inventory to malingering. Journal of Personality Assessment, 71(3), 379-392.
Efendov, A. A., Sellbom, M., & Bagby, R. M. (2008). The utility and comparative incremental validity of the MMPI-2 and Trauma symptom Inventory validity scales in the detection of feigned PTSD. Psychological Assessment, 20(4), 317.
Elhai, J. D., Gray, M. J., Naifeh, J. A., Butcher, J. J., Davis, J. L., Falsetti, S. A., & Best, C. L. (2005). Utility of the Trauma Symptom Inventory’s Atypical Response Scale in detecting malingered post-traumatic stress disorder. Assessment, 12(2), 210-219.
Fraley, R. C., Waller, N. G., & Brennan, K. A. (2000). An item response theory analysis of self-report measures of adult attachment. Journal of Personality and Social Psychology, 78(2), 350-365. doi: http://dx.doi.org/10.1037/0022-35184.108.40.2060
Gray, M. J., Elhai, J. D., & Briere, J. (2010). Evaluation of the Atypical Response scale of the Trauma Symptom Inventory-2 in detecting simulated posttraumatic stress disorder. Journal of anxiety disorders, 24(5), 447-451.
Marmar, C. R., Weiss, D. S., & Metzler, T. J. (1997). The peritraumatic dissociative experiences questionnaire. Assessing psychological trauma and PTSD, 2, 144-168.
Morey, L. C. (1991). Personality assessment inventory (PAI): Wiley Online Library.
Runtz, M. G. (2002). Health concerns of university women with a history of child physical and sexual maltreatment. Child Maltreatment, 7(3), 241-253.
Weathers, F. W., Huska, J. A., & Keane, T. M. (1991). PCL-C for DSM-IV. Boston: National Center for PTSD-Behavioral Science Division.
Weathers, F. W., Litz, B. T., Herman, D. S., Huska, J. A., & Keane, T. M. (1993). The PTSD Checklist (PCL): Reliability, validity, and diagnostic utility. Paper presented at the annual convention of the international society for traumatic stress studies, San Antonio, TX.