Psychometric properties of the Trauma Symptom Inventory (TSI; Briere, 1995)

Type of Instrument

The Trauma Symptom Inventory (TSI) is a test containing 100 items claiming to measure ‘posttraumatic stress and other psychological sequelae of traumatic events’. It was devised to be used in the assessment of ‘acute and chronic traumatic symptomatology’, such as rape, physical assault, spouse abuse, major accidents, combat trauma, natural disasters and the enduring effects of childhood abuse and early childhood trauma (Briere, 1995).

The TSI has 3 validity scales and 10 clinical scales that assess a broad range of psychological symptoms including those related to Posttraumatic Stress Disorder (PTSD) and Acute Stress Disorder (ASD) plus intra and interpersonal difficulties associated with chronic psychological trauma. The test is self-administered and is intended for a fifth grade and above reading level (Briere 1995). Items are scored on a four point scale with 0 = Never through to 3 = Often, and are rated in terms of frequency of occurrence over the previous six months. Due to this time frame the TSI was ‘not intended to generate a DSM IV PTSD diagnosis’. The TSI takes approximately 20 minutes to complete and around 15 minutes to score (Briere and Elliott, 1997).

Reliability

Table 1 lists the scales along with their internal reliability correlation coefficient for the standardised sample (N = 836), and the number of items for each scale (Briere, 1995).

Table 1. Clinical scales & Mean alpha coefficients & number of test items for each scale.

Scales

alpha

Items

Validity Scales:

Response level (RL)

Atypical Response (ATR)

Inconsistent Response (INC)

 

.80

.75

.51

 

10*

10

20*

Clinical Scales:

Anxious Arousal (AA)

Depression (D)

Anger/Irritability (AI)

Intrusive Experiences (IE)

Defensive Avoidance (DA)

Dissociation (DIS)

Sexual Concerns (SC)

Dysfunctional Sexual Behaviour (DSB)

Impaired Self-Reference (ISR)

Tension Reduction Behaviour (TRB)

 

.86

.91

.90

.89

.90

.82

.87

.85

.88

.74

 

8

8

9

8

8

9

9

9

9

8

(Source: Briere 1995) *RL = 5 independent items & 5 combined *INC = 20 combined items

The Mean alpha correlation scores show a high reliability on all scales except for the INC validity scale.

Briere (1995), states that the mean intercorrelations of the 10 clinical scales for the ‘TSI are internally consistent with Mean alpha coefficients’ of .86 for the standardised (N=836), .87 for the clinical (N=370), .84 for the university (N=279), and .85 for the military samples (N=3659).

Runtz & Roche (1999), in their study of a group of 775 ‘previously victimized’ Canadian university women calculated internal consistency reliabilities for the clinical scales of the TSI. They found that the TSI internal consistency is strong (alpha = .64) ‘as all reliabilities were above alpha = .80, except for TRB’. They cite other researchers who found that TRB scales have lower reliability with student samples. In their sample more than 90% of those surveyed were aged under 25.

Validity

  • Criterion Validity: Discriminant function analysis, using the standardised clinical scales of the TSI to predict PTSD status were compared to the subscales on the Brief Symptom Inventory, (BSI), and the Impact of Events Scale (Norris & Raid, 1997 p.30). This comparison indicated, that all TSI scales were associated with PTSD (Briere, 1995 p.44). From a sample of 449 of the general population the TSI scales predicted 24 of the 26 true positive cases of PTSD (92%). The TSI predicted 91% of true negative cases of PTSD, identifying 385 of 423 PTSD negative cases. The TSI also predicted 89% of a clinical sample ‘independently diagnosed with Borderline Personality Disorder’ (Briere, 1995).

  • Runtz & Roche (1999), found in their study that both childhood sexual assault (CHA) as measured by an events checklist created by the researchers, and childhood physical maltreatment as measured by a modified version of the Physical Maltreatment scale (PHY), were linked to all 10 scales of the TSI. However, other studies have not firmly established a link between childhood physical abuse and the TSI scales.

     

  • Construct Validity: Discriminant function analysis was used to examine the relationship, in the normative sample, between TSI T scores and four types of traumatic experience – adult interpersonal violence, adult disaster, childhood interpersonal violence, and childhood disaster, and it was found that ‘all four trauma types were significantly associated with elevated TSI scores’ (Briere, 1995 p.38). Analysis also justified ‘conceptualising the scales in terms of three higher order constructs’, traumatic stress – IE, DA, DIS, and ISR, dysphoria – AI, D, and AA, and Self – ISR, SC, DSB, TRB, AI. However, these factors were ‘highly interrelated’, and ISR and AI in the Self construct scored low correlations which indicates that the Self construct would be more related to ‘sexual trauma and dysfunction’ (Briere, 1995; Norris and Raid, 1997).

  •  

  • Convergent & Discriminant Validity: The ATR and RL validity scales on the TSI correlated with other validity scales on the PAI and MMPI-2. The ATR correlated at .52 with the PAI Negative Impression Management (NIM) scale of the Personality Assessment Inventory and .50 with the MMPI-2 F scale. The TSI RL scale positively correlated with the PAI Positive Impression Management scale and at .46 on the MMPI-2 K scale. The TSI INC scale was uncorrelated with the PAI ICN scale (Briere, 1995).

  •  

    The TSI clinical scales were compared to the scales on the Brief Symptom Inventory (BSI). ‘Reasonable convergent validity was observed between those scales expected to correlate positively’ ie: Anxiety vs AA at .75, Depression scales at .82, Hostility vs AI .77. The TSI clinical scales were also compared with the IES and SCL scales with the IES Avoidance vs DA at .69, and SCL Avoidance vs DA at .68. IES Intrusion vs IE correlated at .67, and SCL .73. SCL Arousal scale vs AA correlated at .75. This again would suggest reasonable convergent validity but low reliability on discriminant validity (Briere, 1995).

    Norms

    The standardisation sample Norms and T scores were derived from the general population in a mail out of 836 American males and females 18 years and over. The mean age of subjects 47.3 years (SD = 16.6%) range = 18-88). Of the sample 57.1% were married, 16.6% separated, 16.5% single, 50.8% males, 77.5% Caucasian, 10.3% African American, 6.1% Hispanic, 2.9% Asian, and 2.3% Native American. Normative data for the TSI scales were derived from the raw score data of the above standardisation sample. There are separate norms for a sample of 3,659 male and female navy recruits (Briere, 1995). Analyses of variance revealed differences in age, sex and race. Based upon age and sex groupings normative data were derived with Linear T scores having a mean of 50 and a standard deviation of 10. Separate norms have been calculated for age 18 – 55 and 55 and over for both male and females.

    The Runtz & Roche (1999) study confirmed Briere’s (1995) observation that student samples report ‘greater difficulties on many of the TSI scales’ than a survey of the general population. The overall means for the 10 TSI scales were higher than the standardisation sample of females 18 to 54, compared to the student sample. They had an average of .42 SD (ranging from .15 to .81).

    References

    Briere, J. (1995). Trauma Symptom Inventory (TSI) Professional Manual, Psychological Assessment Resources, Inc.

    Briere, J., & Elliott D.M. (1997) Psychological Assessment of Interpersonal Victimisation Effects in Adults and Children. Psychotherapy, 34, 353 – 364.

    Norris, F.H., & Raid, J.K. (1997). Standardised self-report measures of civilian trauma and posttraumatic stress disorder. In J.P. Wilson and T.M Keane (Eds.) Assessing psychological trauma and PTSD. The Guilford Press: New York.

    Runtz, M.G., & Roche, D.N. (1999) Validation of the Trauma Symptom Inventory in a Canadian sample of university women. Journal of the American Professional Society on the Abuse of Children, 4, 69 – 80.

    The actual questionnaire cannot be found here as it is copyrighted.

    Above written by: Mr. Philip Byrne

    Reviewed, edited and approved by: Dr. Grant J. Devilly