ADD5106- Week 2 Discussion 1: Norm-Referenced Tests
Tierainie 
ValidationoftheAdultSubstance.pdf
Multistudy Report
Validation of the Adult Substance Abuse Subtle Screening Inventory-4 (SASSI-4) Linda E. Lazowski1 and Brent B. Geary2
1The SASSI Institute, Springville, IN, USA 2Independent Clinical Practice, Phoenix, AZ, USA
Abstract: The study objective was to develop a revision of the adult Substance Abuse Subtle Screening Inventory-3 to include new items to identify nonmedical use of prescription medications, as well as additional subtle and symptom-related identifiers of substance use disorders (SUDs) and to evaluate its psychometric properties and screening accuracy against a criterion of DSM-5 diagnoses for SUD. Clinical professionals throughout the nine US Census Bureau regions and two Canadian provinces who used the SASSI Online screening tool submitted 1,284 completed administrations of the provisional SASSI-4 along with their independent DSM-5 diagnoses of SUD. Validation sample findings demonstrated SASSI-4 sensitivity of 93% and specificity of 90%, AUC = .91. Items added to identify respondents who were abusing prescription medications showed 94% overall screening accuracy. Logistic regression showed no significant effects of client demographic characteristics or type of screening setting on the accuracy of SASSI-4 screening outcomes. In Study 2, 120 adults in recovery from SUD completed the SASSI-4 under instructions to fake good. Sensitivity of 79% was demonstrated for the full scoring protocol and was 47% when only face valid scales were utilized. Clinical utility is discussed.
Keywords: alcohol and drug screening, prescription drug abuse, substance use disorders, screening accuracy
Substance use disorders (SUDs) have received extensive attention as a significant public health concern (Bouchery, Harwood, Sacks, Simon, & Brewer, 2011; US Department of Justice, 2011). Costs of SUD to the US economy are profound, estimated at $193 billion annually related to lost work productivity, healthcare, early mortality, and crime, for abuse of illicit substances, with alcohol-related costs exceeding $220 billion (National Institute on Drug Abuse, 2015). Recent investigations estimate that more than 24.6 million Americans over age 12 use illegal drugs, 16.5 million drink heavily, and 60.1 million are past-month binge drinkers (Substance Abuse and Mental Health Services Administration, 2014). Additionally, the misuse of prescription medication has steadily increased and is a major contributing factor in drug overdose deaths, rise in emergency room visits, and neonatal opioid withdrawal syndrome, among other serious consequences (US Department of Health & Human Services, 2013). Screening instruments that facilitate detection of SUDs are desirable to address their significant human, economic, and societal costs.
Validated SUD screening instruments are integral to the work of substance use assessment professionals and treatment providers. Increasingly, screening for SUD has
been advanced among best practices in primary healthcare settings (Babor et al., 2007), treatment of chronic pain (Compton, Darakjian, & Miotto, 1998), and care of military service members (Santiago, 2014). Such tools also play a significant role in evaluations done in criminal justice probation and reintegration programs (Belenko, 2006), interventions for domestic violence (Easton, Swan, & Sinha, 2000), treatment of patients with spinal cord and traumatic brain injuries (Andelic et al., 2010; Hawkins & Heinemann, 1998), and identification of depression severity risk factors (Williams et al., 2014). Early identification and treatment for SUD can improve outcomes in many domains including family cohesion, crime reduction, and vocational rehabilitation (e.g., Heinemann, Moore, Lazowski, Huber, & Semik, 2014).
The Substance Abuse Subtle Screening Inventory (SASSI; Miller, 1985) was designed to be an easily administered and objective tool that would assist practitioners in identifying persons with a high probability of a substance use disorder so that additional evaluation and treatment could be initiated when appropriate. Key to the design of the inventory was the inclusion of subtle items. Such items make no obvious reference to substance use but are effective in identifying individuals likely to have an SUD,
European Journal of Psychological Assessment (2019), 35(1), 86–97 �2016 Hogrefe Publishing DOI: 10.1027/1015-5759/a000359
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
despite inability or unwillingness to acknowledge substance use behaviors (Laux, Piazza, Salyers, & Roseman, 2012; Miller, 1985). Modifications to the scale composition of the inventory during two revisions, SASSI-2 (Miller, 1994) and SASSI-3 (Lazowski, Miller, Boye, & Miller, 1998; Miller & Lazowski, 1999), enhanced its screening accuracy and clinical utility. In 2013, the American Psychiatric Association (APA) published revised criteria for diagnosing substance use disorders (5th ed.; DSM-5; APA, 2013). The present study was designed to formulate a revision of the SASSI and to validate screening accuracy estimates for the instrument in a heterogeneous sample of adults diagnosed with and without SUD according to the most cur- rent and widely accepted diagnostic standards, the DSM-5. An additional aim was to improve the utility of the instrument by adding a brief new scale to identify prescrip- tion medication misuse and to evaluate its screening efficacy. Subtle items that were unscored on the SASSI-3, as well as new items designed to represent DSM-5 symptoms of SUD, also were evaluated for screening accuracy and improved clinical utility. In a second study, an independent sample of respondents completed the instrument under honest and fake good instructions to examine sensitivity when respondents attempted to conceal evidence of SUD.
Methods
Clinical assessment professionals who were qualified users of the SASSI Institute SUD web-based screening application were recruited to administer a research version of the adult SASSI-4 to their clients. They were also asked to submit an independent DSM-5 SUD diagnostic evaluation for each client who completed the screening instrument based on the proposed diagnostic criteria previously published online (http://www.dsm5.org) by the APA in 2012. Scoring and screening reports for each client were provided by the Institute. Study enrollment lasted from September 2012 through December 2013. Eligibility criteria were at least 18 years of age and English speaking. Assessment profes- sionals from 38 states in all nine US Census Bureau regions, as well as two Canadian provinces, participated.
Using the SASSI Online web application, after counselors provided a client’s demographic and available history data (e.g., arrest history, blood alcohol content [BAC] level, prior alcohol or drug treatment instances), a link to the questionnaire was provided for client responses, which were submitted directly to the secure web server upon completion. Following submission of the counselor diagnos- tic evaluation for the client, the screening responses were scored, and results were made available to the counselor in an online dashboard. Counselors were able to opt out
of the study at any time or exclude client data from the investigation. Figure 1 indicates data collection steps for client data, including rates of exclusions, diagnoses, SUD screening, and randomization of the sample into develop- ment and validation groups.
Participants
Clinical Sample Counselors in 163 practices and organizations throughout the US and Canada administered the screening instrument. Descriptive statistics for cases screened by assessment setting type and client demographic characteristics are shown in Table 1. Diagnosed incidence of SUD in these six assessment setting types ranged from 60% to 86%. The frequency of criterion positive cases submitted by the individual practices and organizations ranged from 0 to 100%.
Retest Sample Retest reliability of SASSI-4 scale scores was examined with an independent sample of 40 participants who were not receiving treatment for substance use. Project staff recruited volunteers aged 18 and older in community organizations (e.g., nonprofits, jobs programs, philanthropic organizations) to complete the screening questionnaire anonymously on two occasions, 1–8 weeks apart (M = 22.8 days, SD = 12.3, range 8–60 days) and offered participants a $10 honorarium. The sample included 12 (30%) men and 28 (70%) women. Their mean age was 61.9 years (SD = 14.5, range 24–85 years). Ethnicities of sample participants included 38 (95%) Caucasians, 1 (2.5%) African American, and 1 (2.5%) person of Hispanic origin. Forty percent of the sample (n = 16) had a high school diploma or equivalent, 27% (n = 11) had a vocational or 2-year college degree, and 33% (n = 13) had 3 or more years of postsecondary education.
Study 2 Testing resistance to faking good. To replicate earlier work on the development of the SASSI in which respondents were instructed to hide signs of unfavorable characteristics and substance misuse (Miller, 1985) and to test the utility of new subtle items as indicators of SUD, an independent sample of 120 respondents was recruited to complete the SASSI-4. Participants were recruited from community organizations that provide support programs (e.g., housing or job search assistance) for adults in recovery from substance use disorders. Adult volunteers were invited to complete the screening questionnaire anonymously on two occasions, approximately 2 weeks apart (M = 11.8 days, SD = 8.5), under two types of instructions. In the first administration, participants were asked to complete the
L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4 87
�2016 Hogrefe Publishing European Journal of Psychological Assessment (2019), 35(1), 86–97
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
questionnaire honestly, based on their actual attitudes and experiences with alcohol, illicit drugs, and nonmedical use of prescription medications. For the second administration, participants were directed to respond as someone would if they were attempting to conceal unfavorable characteristics and substance misuse while still appearing to be believable and forthcoming in their responses. A $10 honorarium was offered. The sample’s mean age was 42.2 years (SD = 13.2, range 19–71 years). Seventy-nine percent (n = 95) of the sample participants were men. Respondents’ race/ethnicity included White (77%; n = 92), Black (18%; n = 22), and Native American, biracial, or other races (5%; n = 6). Highest education included 15% (n = 18) with less than a
high school diploma, 31% (n = 37) with a high school diploma or equivalent, and 54% (n = 65) with one or more years of vocational or other postsecondary education.
Measures
SUD Diagnoses In DSM-IV, SUDs had been classified into categories of substance dependence and substance abuse. In DSM-5, the former abuse and dependence symptom criteria are combined into a single diagnostic set. The 11 specific symptoms remain the same as in DSM-IV, with the exception that the former symptom “substance-related
Figure 1. SASSI-4 clinical sample flow diagram.
88 L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4
European Journal of Psychological Assessment (2019), 35(1), 86–97 �2016 Hogrefe Publishing
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
legal problems” was removed and the new symptom “craving or strong desire or urge to use a substance” was added. In DSM-5, specifiers for SUD severity are delineated by the number of symptoms evidenced: 2–3, mild; 4–5, moderate; 6 or more, severe.
Counselors indicated the presence or absence of the 11 DSM-5 SUD symptom criteria and specified for which
drug class the symptom was evidenced, within the time period (past 12 months, lifetime) for which they conducted each diagnostic evaluation. In addition to those with current SUDs, cases with SUD in remission were considered criterion positive. Counselors also indicated diagnoses of any non-substance related psychological disorders. Cases without sufficient diagnostic information were excluded from analyses.
Logistic regression was used to evaluate whether excluded cases differed in client demographic characteris- tics or type of assessment setting from clinical cases that did include sufficient diagnostic information. Age, gender, ethnicity, education level, employment status, marital status, and type of screening setting were used as predictor variables of case inclusion status. The omnibus test of the model coefficients showed no significant effects of the demographic or setting variables on case inclusion status, w2(29, n = 1,320) = 28.1, p = .45.
The incidence of diagnosed SUD in this screening sample was 75.7%. Sixteen percent of these cases met criteria for mild SUD, 12% for moderate SUD, and 48% for severe SUD. Three hundred eighty cases (29.6%) included diagnoses of non-substance related psychiatric disorders, both with and without coexisting SUD. Non-substance related diagnoses in the clinical sample included: depression (17.1%), anxiety (14.3%), bipolar disorder (5.8%), attention-deficit hyperactivity disorder (5.0%), post- traumatic stress disorder (4.0%), and other mental health diagnoses (3.5%).
Research Version of the SASSI-4 The SASSI-3 consists of 67 true-false items that identify SUD through obvious and subtle content. The instrument also includes 26 face valid alcohol and other drug frequency items. Face valid and subtle items are organized into seven scales that are utilized in a series of decision rules to produce a dichotomous SUD screening classification.
The Face Valid Alcohol (FVA) and Face Valid Other Drug (FVOD) scales measure how often (0 = never to 3 = repeat- edly) respondents have engaged in and experienced effects from the use of alcohol and other drugs within a specified time frame (e.g., lifetime, past 12 months). All other SASSI-3 scales utilize a true-false response format. The Symptoms (SYM) scale contains face valid items that assess substance use history and consequences. The Obvious Attributes (OAT) scale is compiled empirically of items shown to discriminate between SUD criterion groups under standard instructions to answer honestly. The Subtle Attributes (SAT) scale consists of items found to discrimi- nate between individuals with and without SUDs when respondents answered honestly and when they attempted to conceal signs of substance misuse. The Defensiveness (DEF) scale discriminates responses given under honest
Table 1. SASSI-4 clinical sample participant characteristics
Characteristic n = 1,245 %
Assessment setting
Substance use treatment programs 564 46
Criminal justice: drug courts, probation and parole, community corrections
209 18
Private practice 190 16
Behavioral health facilities 107 9
DOT and DUI screening and education 75 6
Social service programs 59 5
Clinical SUD diagnosis
Criterion positive 945 76
Criterion negative 300 24
Age (range 18–79)
M 34.4
SD 12.1
Gender
Male 781 63
Female 464 37
Education
Less than high school diploma 199 17
High school diploma 479 41
One or more years postsecondary education
490 42
Race/Ethnicity
White 900 75
Black 133 11
Hispanic 83 7
Native American or Alaska Native 45 4
Asian, Native Hawaiian, or Pacific Islander 11 1
Biracial or other 26 2
Employment status
Full time 445 38
Not employed 381 32
Part time 146 12
Retired due to disability 85 7
Student 81 7
Homemaker 25 2
Retired due to age 13 1
Marital status
Single 596 49
Married or cohabiting with partner 330 28
Divorced or separated 270 22
Widowed 16 1
L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4 89
�2016 Hogrefe Publishing European Journal of Psychological Assessment (2019), 35(1), 86–97
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
versus fake good instructions. The measure indicates the extent to which respondents are willing to acknowledge minor, socially acceptable limitations or attempt to deny such flaws. The SAM (Supplemental Addictions Measure) scale consists of items that discriminate between SUD crite- rion groups and is used in the SASSI-3 decision rules. These seven scales are utilized in the dichotomous screening out- come. The inventory also contains two supplementary clin- ical scales that are not used to screen for SUD but provide information that can be useful in evaluation and treatment planning. Finally, the Random Answering Pattern (RAP) scale is used to identify profile invalidity that might be due to deliberate noncompliance, insufficient reading comprehension, inattention, or due to other processes.
Fifteen new items were added to the research version of the SASSI-4 to assess their accuracy in identifying persons with SUDs based on subtle identifiers, nonmedical use of prescription drugs, and symptom criteria for a DSM-5 diagnosis of SUD (e.g., craving for a substance).
The Lexiler Framework for Reading was used to assess readability of the questionnaire (Stenner, Burdick, Sanford, & Burdick, 2007). The Lexiler Measure for the provisional instrument was 740L, which corresponds to the reading text complexity band for 4th and 5th grade students (Nelson, Perfett, Liben, & Liben, 2012). Average time to complete the inventory was approximately 15 min (M = 14.8 min, SD = 5.9).
SASSI responses were evaluated for profile invalidity. Thirty-nine cases indicated elevated Random Answering Pattern scores (2 or higher) and were excluded from analyses. A total of 1,245 complete cases, including a valid and complete SASSI-4, a diagnostic evaluation for SUD, and client demographic information, served as the primary dataset (see Figure 1).
Analysis Strategy Overview
Five psychometric indices of the screening inventory scores and their performance were evaluated: retest reliability; internal consistency reliability; screening accuracy; impacts of assessment setting and client demographic characteris- tics on accuracy; and robustness of screening performance when clients attempt to fake good.
Data Analyses
Reliability Retest reliability was evaluated with Pearson correlations. Because these correlation coefficients measure the repro- ducibility of the rank order of participant scores on retest
but do not allow one to assess magnitude of change in raw scale scores, paired sample t-tests also were conducted on participant responses to assess raw scale score stability.
Internal Consistency The SASSI is composed of multiple subscales that were compiled empirically on the basis of whether items discriminated between known criterion groups under various instructional sets (Miller, 1985, 1994). SASSI items were not designed or selected based on a criterion of constant item variances. Noting how infrequently the assumptions of essential tau-equivalence are met, and yet are necessary for the appropriate use of coefficient alpha as an estimate of internal consistency reliability, recent papers in the psychometric literature have suggested the use of coefficient omega as a preferred measure of internal consistency (Dunn, Baguley & Brunsden, 2014; Gignac, 2014; McDonald, 1999; Revelle & Zinbarg, 2009). Because the congeneric measurement model underlying the use of omega does not require homogeneity of item variances, the use of coefficient omega as an estimate of internal consistency is appropriate for the current data. Internal consistency reliability was estimated for the SASSI-4 by calculating omega coefficients (McDonald, 1999) for the item set composed of all items included on the screener as indicators of SUD likelihood (Table 2: “SASSI-4 overall”) and for each scale individually. DEF and RAP scale items, as well as items in the two supplementary clinical scales, FAM and COR, were excluded from the overall analysis because they were not included on the instrument as measures of SUD likelihood. Omega coefficients were computed using R open source software statistical functions (R Development Core Team, 2014).
In contrast to coefficient omega, omega hierarchical estimates how reliable test scores are as indicators of the target construct of interest (here SUD) and as such is a measure of general factor saturation (McDonald, 1999; Zinbarg, Revelle, Yovel & Li, 2005). Omega hierarchical for the SASSI-4 was calculated using the omega function implemented in the psych package (Revelle, 2014) of R statistical software (R Development Core Team, 2014).
Criterion Validity The clinical sample was divided randomly into a develop- ment sample and a reserve, cross-validation sample. The development sample was used to establish a scoring protocol that maximized correspondence with clinicians’ diagnoses of SUD while balancing the false positive and false negative error rates. Screening accuracy of the new scoring protocol was validated on the reserve set of clinical cases. We computed various measures of accuracy,
90 L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4
European Journal of Psychological Assessment (2019), 35(1), 86–97 �2016 Hogrefe Publishing
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
including: sensitivity (true positives/criterion positives), specificity (true negatives/criterion negatives), positive pre- dictive value (true positives/test positives), and negative predictive value (true negatives/test negatives). Percent correctly classified, and Areas Under the Curve (AUCs) of Receiver Operating Characteristic (ROC) curves were uti- lized as summary indices of agreement between screening outcomes and DSM-5 SUD diagnoses. Likelihood ratio chi- square statistics tested whether the observed agreement differed from chance.
To examine whether the accuracy of SASSI-4 screening outcomes was impacted differentially by respondents’ demographic characteristics or the type of assessment setting in which they were screened, we conducted a logistic regression analysis on screening accuracy in the validation sample using age, gender, ethnicity, education level, employ- ment status, marital status, and screening setting as predic- tor variables. Gender was the only predictor that had been utilizedwhen formulating the scoring rules with the develop- ment sample and was included as a predictor in the logistic regression to assess whether any remaining gender-asso- ciated score variance impacted screening accuracy.
To assist practitioners in identifying individuals likely to have an SUD related to the misuse of prescription medica- tions, six prescription drug abuse items were included on the SASSI-4. The items have face valid content such as “Used prescription drugs that were not prescribed for you” and “Took a higher dose or different medications than your doctor prescribed in order to get the relief you need.” Internal consistency and retest reliability estimates were calculated for this new scale. Accuracy of this screening measure was estimated using a diagnosis of SUD for opioid or sedative drugs as the criterion for presence of the disorder. Cases without an SUD served as the criterion negative sample.
Results
Modifications to SASSI-3 Scales
Discriminant analyses of development sample responses indicated a 90.5% correct classification rate based on the discriminant function equation. Items that discriminated between SUD criterion groups were retained in the SASSI-4 item pool. Three items that were scored in the SASSI-3 decision rules and four subtle, previously unscored SASSI-3 research items on the published instrument were not significant criterion group discriminators and were excluded from consideration in the new scoring rules. The six remaining previously unscored subtle research items and the 15 newly added items were significant discriminators of SUD criterion groups and were added to the FVA, FVOD, SYM, or SAT scales based on item content. To evaluate the effectiveness of the subtle items, we conducted an additional analysis that utilized only the SAT items as discriminators of criterion negative and fake good respondents. The discriminant function equation indicated the SAT items alone yielded a correct classifica- tion rate of 82.1%, indicating effectiveness of the item set in identifying persons with SUD, even when respondents attempted to conceal signs of substance use disorder.
Reliability
Test-Retest Reliability As shown in Table 2, Pearson correlation estimates of SASSI-4 score reliability in this sample ranged from .78 to .99 and indicate high consistency in scores retested over the 8 to 60-day interval. DEF scores were more variable, as would be expected based on the likelihood that defensive
Table 2. Test-retest reliability and internal consistency reliability estimates for the SASSI-4
Test-retest Internal consistency
Reliability coefficientsa Reliability coefficientsb
Scale Pearson r 95% CI Omega 95% CI
SASSI-4 overallc .99 [.98, .99] .97 [.96, .97]
Face valid alcohol .99 [.98, .99] .93 [.93, .94]
Face valid other drug .99 [.98, .99] .96 [.96, .97]
Symptoms .97 [.94, .98] .90 [.89, .90]
Obvious attributes .91 [.83, .95] .74 [.71, .76]
Subtle attributes .84 [.70, .91] .70 [.67, .72]
Defensiveness .78 [.61, .88] .72 [.71, .75]
Supplemental addiction measure .91 [.83, .95] .83 [.81, .84]
Family vs. controlsd .89 [.80, .94] .70 [.67, .72]
Correctionald .95 [.90, .97] .81 [.80, .83]
Notes. CI = confidence interval. aN = 40. bN = 1,245. cOnly items that are utilized as indicators of SUD likelihood were included in this analysis. dScale is not used to classify respondents regarding likelihood of SUD.
L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4 91
�2016 Hogrefe Publishing European Journal of Psychological Assessment (2019), 35(1), 86–97
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
responding is impacted by psychological state and situational factors (Flett, Besser, & Hewitt, 2005; Knee, Porter, & Rodriguez, 2014).
Raw Scale Score Stability Overall magnitude of change was very small on all scales. The largest mean change occurred in OAT scores and indicated less than a 1-point increase (Δ .38) in the mean OAT scale score at Time 2. None of the differences were significant for any of the SASSI-4 scale scores (all ps > .05), indicating high temporal stability in raw scale scores in the absence of intervention.
Internal Consistency Omega coefficients are shown in Table 2 and indicate overall internal consistency of .97 with scale reliability estimates ranging from .70 to .97.
Findings also indicated an omega hierarchical coefficient of .78. Dividing omega hierarchical by omega total (here .78/.97) estimates 80.4% of the reliable variance in the overall measure is attributable to SUD likelihood, the construct of interest, while the remaining reliable variance is due to unique variance associated with the subscales (see also, Reise, Bonifay, & Haviland, 2013).
Criterion Validity
Correspondence of SASSI-4 Screening Outcomes With DSM-5 SUD Diagnoses Beginningwith the SASS-3decision rules and thenewly com- piled SASSI-4 scales, an iterative process was used to adjust
scale cutoffs in the development sample of clinical cases (N = 620). At each iteration we assessed the impact on screening sensitivity and specificity in this sample with the aim of optimizing overall accuracy while balancing both types of screening errors. Via this process, we formulated a scoring protocol that demonstrated overall development sample classification accuracy of 92.6%: sensitivity 94.1%, specificity 87.9%, positive predictive value (PPV) 91.2%, and negative predictive value (NPV) 82.4%; likelihood ratio (1, N = 620) = 383.8, p < .001. ROC analysis indicated an AUC of .91 (SE = .02), p < .001, 95% CI [.88, .94]. Screening accuracy indices resulting from the applica- tion of the new scoring rules to the reserve validation sample (N = 625) are shown in Table 3, as are accuracy findings for lifetime and past year SUD, and by SUDseverity.
Cases diagnosed with mild SUD reveal that 15 of the 21 (71%) cases with negative screening outcomes were diagnosed based on evidence of only twoDSM-5 symptoms; the remaining six caseswerediagnosedwithmild SUDbased on evidence of three symptom criteria. The most frequent pattern observed in test misses with two SUD symptoms showed the individual had been arrested for DUI and the clinician indicated symptom criteria of “used more than intended” and “used in hazardous situations.”
Logistic Regression Using Demographic and Screening Setting Variables as Predictors of SASSI-4 Screening Accuracy Screening accuracy by assessment setting types ranged from 87% in government and community social service
Table 3. Validation sample estimates of SASSI-4 screening accuracy by DSM-5 diagnostic features
Criterion status Prev AUC [95% CI] Sens Spec PPV NPV %CC
Any SUD (n = 625) 76 .91 [.88, .95] 93 90 97 80 92
Lifetime SUD (n = 480) 79 .92 [.89, .96] 92 92 98 76 92
Past year SUD (n = 145) 66 .90 [.84, .97] 95 86 93 90 92
SUD severity (n = 625)
Mild (n = 97) 16 .84 [.79, .90] 78 90 84 87 86
Moderate (n = 77) 12 .91 [.87, .96] 92 90 83 96 91
Severe (n = 300) 48 .94 [.91, .97] 98 90 95 95 95
No SUD (n = 151) 24
Other mental health
Diagnoses (n = 380)
Co-occurring SUD (n = 320) 84 .96 [.92, .99] 98 93 99 89 97
Non-substance
Related disorder (n = 60) 16
Rx screen (n = 246)
Opioid or sedative
SUD (n = 95) 39 .93 [.89, .97] 87 98 97 93 94
No SUD (n = 151) 61
Notes. All figures other than AUC are percentages. Prev = Prevalence of SUD within the diagnostic category; AUC = area under the curve; CI = confidence interval; Sens = Sensitivity; Spec = Specificity; PPV = Positive Predictive Value; NPV = Negative Predictive Value, %CC = percent correctly classified (i.e., accuracy); SUD = substance use disorder, Rx Screen = SASSI-4 prescription drug abuse scale.
92 L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4
European Journal of Psychological Assessment (2019), 35(1), 86–97 �2016 Hogrefe Publishing
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
programs to 97% in criminal justice settings. The omnibus test of the model coefficients showed no significant effects of the demographic or screening setting variables on accu- racy of the screening classification, w2(29, N = 554) = 28.1, p = .51. The null model, which included the accuracy of the SASSI-4 screening outcome as the constant, correctly predicted the classification of 510 cases (92.1%) and the full model that included the six client demographic variables and type of assessment setting showed no improvement to accuracy prediction; 510 cases were correctly classified in this model.
Accuracy in Detecting Likely SUD in Persons Diagnosed With Non-Substance Related Disorders Clinicians’ assessments of individuals in the SASSI-4 validation sample indicated that 380 individuals had been diagnosed with non-substance related psychiatric disorders, either in addition to SUD (co-occurring disorders, n = 320) or with a non-substance related psychiatric disorder only (n = 60), whereby they were considered criterion negative with respect to SUD. SASSI-4 screening accuracy for these cases is also presented in Table 3 under the heading “Other Mental HealthDiagnoses.” These findings provide evidence of construct validity and indicate that the SASSI-4 is effec- tive in distinguishing identifiers associated with likely SUD from those associated with other behavioral health needs.
Identification of Likely Prescription Medication Abuse Reliability analyses for the SASSI-4 Rx scale indicated a Pearson retest reliability coefficient of .95, 95% CI [.90, .97], and an internal consistency reliability coefficient omega of .88, 95% CI [.87, .89]. With cases in the development sample, a cutoff score of 3 or more on the pre- scription drug scale produced sensitivity of 83.3% and specificity of 96.0%, likelihood ratio (1, N = 251) = 186.5, p < .001; AUC = .90 (SE = .02), p < .001, 95% CI [.85, .94]. When this rule was applied to validation sample cases diagnosed with or without opioid or sedative SUDs, screening accuracy findings for the SASSI-4 Rx scale, shown in Table 3 under the heading “Rx Screen,” indicated strong correspondence between the prescription drug scale screening classification and clinicians’ diagnoses.
To further specify parameters for the effective function- ing of the SASSI-4 prescription drug scale, we assessed its screening sensitivity in identifying cases diagnosed with any type of SUD, using the same cutoff score of 3 used for opioid and sedative related SUDs. Thirty-five percent (165/474) of all SUDs in the validation sample were identified with this rule. Even when the cutoff score was lowered to 2, screening sensitivity for any type of SUD was 45% (214/474 cases), indicating that SASSI-4 prescrip- tion drug scores are not effective as a stand-alone screening measure for all SUDs.
Study 2 Results
SASSI-4 Resistance to Faking Good Prior alcohol and drug treatment was reported by 89% of this sample of adults recovering from substance use disorders. Screening outcomes for responses in the honest condition indicated all 120 individuals tested positive on the SASSI-4; five of these individuals (4%) had an elevated score on the SASSI RAP scale, signaling that responses might not be valid, and were excluded in subsequent analyses. At Time 2 when asked to fake good, 23 additional respondents (20%) had elevated RAP scores and were excluded. The rate of invalid profiles in the fake good condition was significantly elevated above the 3% rate observed in the clinical sample and the 4% rate of these same participants in the honest condition, providing evidence that RAP scores can reflect deliberate response styles. Faking participants also omitted responses on one or more SASSI-4 scales (n = 18), which were sufficient to prevent calculation of a conclusive screening outcome for only one of these cases. Mean scores for cases with com- plete pairwise responses are shown by instructional set in Table 4.
As shown in Table 4, under instructions to fake good, participants’mean scores on the face valid scales decreased significantly – between one and two standard deviations. In addition, participants’ Defensiveness scores increased significantly, consistent with early research on the instru- ment (Miller, 1985) and the intended design to provide practitioners a way of identifying possible response mini- mization. Scores on the SAT scale decreased significantly, although on average, less than one standard deviation.
Screening outcomes based on the full SASSI-4 scoring protocol, which utilizes both subtle and face valid scale scores, indicated sensitivity of 79.1% (72/91 cases), 95% CI [.70, .86] in identifying likely SUD in individuals who attempted to hide signs of substance misuse. Further, 15 of the 19 respondents (79%) who were able to dissimulate a negative screening outcome had elevated DEF scores that met author-recommended guidelines for further evaluation regarding possible response minimization. In contrast to the full SASSI-4 scoring protocol, when only the face valid scale scores, FVA, FVOD, and SYM, were used to produce screening outcomes, sensitivity was 47.3%, (43/91 cases), 95% CI [.37, .57], indicating vulnerability of face valid SUD screening measures to attempts to minimize signs of SUD.
Discussion
The study objective was to formulate a revision of the adult SASSI screening tool that would demonstrate high
L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4 93
�2016 Hogrefe Publishing European Journal of Psychological Assessment (2019), 35(1), 86–97
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
correspondence with the current diagnostic criteria for substance use disorders. An additional aim was to balance sensitivity and specificity so the screener can be used in a variety of settings. The study sample included SUD screen- ings from a diverse sample of practices throughout the nine US Census Bureau regions and two Canadian provinces. Analyses indicated 92% overall correspondence with clini- cians’ DSM-5 diagnoses of SUD, with no significant varia- tion from this accuracy level across a range of assessment setting types and respondent demographic characteristics.
The data provide evidence of SASSI-4 screening accuracy in detecting lifetime and past year presence of SUDs, which enhances the utility of the inventory for use in programs with varying screening objectives. Findings also provide evidence of screening sensitivity across a range of SUD symptom severity, improving use of the instrument in programs that serve individuals with varying levels of the disorder.
Less sensitivity was observed in a sample of respondents with mild SUDs, particularly in cases where diagnoses evi- denced only two of the 2–3 criteria necessary for this level of SUD severity. Examination of client data for those who screened negative when clinicians had indicated mild SUD severity revealed 57% had a DWI violation. Other research has found that individuals who received a DSM-IV diagnosis of alcohol abuse based solely on the criterion of driving while intoxicated differed from non- substance abusers on approximately half of the external diagnostic validating criteria. In contrast, persons who met substance abuse criteria by other symptoms differed from non-substance abusers on all external validating criteria (Hasin, Paykin, Endicott, & Grant, 1999). For example, the drinking-driver abuser group did not differ from those without abuse diagnoses on drinking during the week, depressed mood, or self or others’ perceptions that they needed treatment. Whether the hazardous and illicit substance-related behaviors in which they engage
meet the standard of a psychiatric substance use disorder merits further attention. In the current study, it is possible that the legal consequences associated with clients’ alcohol or drug use affected their diagnoses, even though legal consequences have been removed as a diagnostic symptom of SUD in DSM-5. This phenomenon might be especially likely when a legal consequence is the precipitating event for the assessment or when it is salient in a profile with otherwise low evidence of SUD. Further validation research is needed to gather additional estimates of SASSI-4 screening sensitivity for mild SUD severity. In addition, research on whether legal consequences resulting from individuals’ substance use play a role in clinicians’ DSM-5 SUD diagnoses might also be informative about the ways in which current diagnostic standards are utilized clinically.
Excellent SASSI-4 sensitivity in identifying SUD in individuals experiencing co-occurring psychiatric disorders, and high specificity for individuals without SUD who instead were experiencing cognitive, mood, somatic, or related impairments associated with other psychiatric disorders was also demonstrated. Given that screening for SUD occurs in settings where individuals are likely to present with a variety of behavioral health needs, it is important that the screener effectively discriminate between symptoms associated with SUDs versus those indicative of problematic functioning in other areas.
New items added to assist practitioners in identifying individuals likely to be abusing prescription medications had an overall accuracy rate of 94% in a criterion sample diagnosed with opioid or sedative related SUDs. This extends the utility of the SASSI-4 to practitioners seeking to address the escalating prevalence of misuse of prescrip- tion medications and can be useful in settings where clients may be at high risk for medication abuse (e.g., disability and pain management populations). Since this scale is a face valid measure, individuals who are unwilling to acknowledge prescription drug misuse can avoid detection
Table 4. Paired samples t-tests: SASSI-4 mean scale scores as a function of instructional set
Instructional set
Honest Fake good
SASSI-4 scale M SD M SD t df
Face valid alcohol 25.5 10.1 10.6 10.8 10.26* 81
Face valid other drugs 36.0 17.1 12.3 14.8 10.71* 74
Symptoms 12.8 3.8 6.0 5.6 9.59* 83
Obvious attributes 7.3 2.2 4.2 3.0 8.99* 83
Subtle attributes 7.1 2.7 5.0 2.7 6.39* 83
Defensiveness 3.9 1.8 6.8 2.5 �9.16* 84
Supplemental addiction measure 10.9 1.8 5.6 4.2 10.68* 86
Family vs. controls 5.6 2.1 8.7 2.9 �8.39* 88
Correctional 10.0 2.5 4.3 4.0 11.18* 84
Note. *p < .001.
94 L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4
European Journal of Psychological Assessment (2019), 35(1), 86–97 �2016 Hogrefe Publishing
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
on this scale. In such instances, respondents’ scores on the SASSI-4 DEF scale can be a useful tool in discerning response minimization.
It is also important to note that in the current study clinicians specified the class of drugs for each diagnosed client symptom but were not asked to indicate the specific drugs clients used. Since prescription opioid pain medica- tions (e.g., Vicodin, OxyContin, and Hydrocodone) and sedatives used to treat anxiety and sleep disorders (e.g., Valium, Xanax, and Ambien) are among the most widely abused prescription medications (US Department of Health & Human Services, 2014; Volkow, 2014) we reasoned that a diagnosis of opioid or sedative related SUD had utility as a criterion variable for validating the SASSI-4 prescription drug abuse scale. Ninety-seven percent of clients whose reported nonmedical use of prescription medications met the Rx scale cutoff were independently diagnosed with an opioid or sedative related SUD (PPV), and 93% of clients who screened negative on the Rx scale were diagnosed as not having an SUD (NPV). It is possible, given the use of this more general criterion variable, that false negative cases in this analysis included some who were diagnosed with an opioid use disorder based on their heroin use, a nonprescription opiate, and thus should not be considered legitimate false negatives. Indeed, findings indicated that of the 12 criterion positive “misses,” seven screened positive on the overall SASSI-4 screening outcome for any SUD. Thus, this evaluation of the Rx scale screening accuracy may overestimate its false negative rate. Future validation studies of the SASSI-4 Rx scale that include diagnostic specification of the prescription and other drugs evidenced in criterion positive cases are needed to provide more precise estimates of screening error rates for this scale.
SASSI-4 Resistance to Faking Good
The SASSI was designed to identify individuals in need of diagnostic evaluation for SUD, including individuals who may be unable or unwilling to acknowledge their substance misuse. Findings demonstrated SASSI-4 sensitivity of 79% even when respondents deliberately attempted to conceal their substance use. Moreover, only four participants out of 120 were successful in faking a negative screening out- come without producing elevations on the RAP or DEF scales. Elevated scores on these scales alert practitioners that responses are atypical and warrant further evaluation to determine whether invalid profiles are motivated and whether cases indicating elevated defensiveness are attributable to unwillingness to acknowledge SUD, lack of insight into the causes of any substance-related
consequences evidenced, or to situational or dispositional factors unrelated to substance misuse. In each of these cases, elevated RAP and/or DEF scores provide valuable information to practitioners and have utility in directing the course of additional assessment beyond that which is available through face valid screening instruments.
Study Limitations The SASSI is a screening instrument, designed to be used as one source of information in clinicians’ decision-making regarding individuals in need of diagnostic evaluation for the presence of an SUD and the potential need for treatment; it does not provide a diagnosis. Respondents in early or sustained remission were among the criterion positive cases in this study. Therefore, persons who screen positive on the SASSI-4 might already be in remission.
Data used to validate the screening instrument were submitted by practitioners engaged in ongoing pro- grams of substance use screening. In one respect, this is a study advantage in that we measured clients’ responses in actual settings where decisions regarding further evaluation for treatment needs would ensue from clients’ self- reports. Additional validation in practices that serve lower rates of SUD can extend the generalizability of the current findings.
Conclusions
We formulated a revision of the SASSI SUD screening instrument for alcohol, illicit drugs, and nonmedical use of prescription medications that demonstrates high sensitiv- ity and specificity, and consistency in results across hetero- geneous samples of adults in substance use treatment, criminal justice, behavioral health, and social service programs. Results also demonstrate accuracy of screenings for lifetime and past year presence of SUD. The SASSI-4 can have considerable clinical utility in forensic, vocational, and psychological evaluations. The inventory can assist practitioners in identifying individuals in need of further assessment for SUD, and treatment recommendations can be enhanced by information regarding clients’ defensive- ness or acknowledgment of their substance use, and specific substance use related consequences they have experienced. Accurate detection of an SUD can facilitate differential diagnosis, client supervision and rehabilitation processes, and decisions regarding case management service delivery, particularly where substance misuse can adversely affect potential benefits of other medical and behavioral health treatment interventions.
L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4 95
�2016 Hogrefe Publishing European Journal of Psychological Assessment (2019), 35(1), 86–97
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
Acknowledgments
The authors gratefully acknowledge the contributions of: Jennifer Cullen Meyer for consultation on analyses; Tim Baker, Carl Briggs, Allen Heinemann, Jonathan Kaplan, Kristin Kimmell, and Nelson Tiburcio for their many helpful comments on earlier drafts of this manuscript; Tom Cox, Anne Hazeltine, David Helton, Melissa Renn, and Lewis Seward for their assistance with onsite data collection; Adrian Hosey and Lauren Nelson for assistance with online data collection; and Scarlett Baker for study coordination and manuscript preparation. Finally, we express our sincere gratitude to the assessment profession- als and community service organizations that were instru- mental in respondent recruitment and diagnostic evaluations for this study.
References
American Psychiatric Association. (2013). Diagnostic and statisti- cal manual of mental disorders (5th ed.). Washington, DC: Author.
Andelic, N., Jerstad, T., Sigurdardottir, S., Schanke, A., Sandvik, L., & Roe, C. (2010). Effects of acute substance use and pre-injury substance abuse on traumatic brain injury severity in adults admitted to a trauma centre. Journal of Trauma Management & Outcomes, 4, 1–12. doi: 10.1186/1752-2897-4-6
Babor, T. F., McRee, B. G., Kassebaum, P. A., Grimaldi, P. L., Ahmed, K., & Bray, J. (2007). Screening, brief intervention, and referral to treatment (SBIRT): Toward a public health approach to the management of substance abuse. Substance Abuse, 28, 7–30. doi: 10.1300/J465v28n03_03
Belenko, S. (2006). Assessing released inmates for substance- abuse-related service needs. Crime & Delinquency, 52, 94–113. doi: 10.1177/0011128705281755
Bouchery, E. E., Harwood, H. H., Sacks, J. J., Simon, C. J., & Brewer, R. D. (2011). Economic costs of excessive alcohol consumption in the US, 2006. American Journal of Preventive Medicine, 41, 516–524. doi: 10.1016/j.amepre.2011.06.045
Compton, P., Darakjian, J., & Miotto, K. (1998). Screening for addiction in patients with chronic pain and “problematic” substance use: Evaluation of a pilot assessment tool. Journal of Pain and Symptom Management, 16, 355–363. doi: 10.1016/ S0885-3924(98)00110-9
Dunn, T. J., Baguley, T., & Brunsden, V. (2014). From alpha to omega: A practical solution to the pervasive problem of internal consistency estimation. British Journal of Psychology, 105, 399–412. doi: 10.1111/bjop.12046
Easton, C., Swan, S., & Sinha, R. (2000). Motivation to change substance use among offenders of domestic violence. Journal of Substance Abuse Treatment, 19, 1–5. doi: 10.1016/S0740- 5472(99)00098-7
Flett, G. L., Besser, A., & Hewitt, P. L. (2005). Perfectionism, ego defense styles, and depression: A comparison of self-reports versus informant ratings. J of Personality, 73, 1355–1396. doi: 10.1111/j.1467-6494.2005.00352.x
Gignac, G. E. (2014). On the inappropriateness of using items to calculate total scale score reliability via coefficient alpha for multidimensional scales. European Journal of
Psychological Assessment, 30, 130–139. doi: 10.1027/1015- 5759/a000181
Hasin, D., Paykin, A., Endicott, J., & Grant, B. (1999). The validity of DSM-IV alcohol abuse: Drunk drivers versus all others. Journal of Studies on Alcohol, 60, 746–755.
Hawkins, D., & Heinemann, A. W. (1998). Substance abuse and medical complications following spinal cord injury. Rehabilitation Psychology, 43, 219–231. doi: 10.1037/0090- 5550.43.3.219
Heinemann, A. W., Moore, D., Lazowski, L. E., Huber, M., & Semik, P. (2014). Benefits of substance use disorder screening on employment outcomes in state-federal vocational rehabil- itation programs. Rehabilitation Counseling Bulletin, 57, 144–158. doi: 10.1177/0034355213503908
Knee, C. R., Porter, B. A., & Rodriguez, L. M. (2014). Self- determination and regulation of conflict in romantic relation- ships. In N. E. Weinstein (Ed.), Human motivation and interper- sonal relationships: Theory, research and applications (pp. 139–158) Amsterdam, Netherlands: Springer. doi: 10.1007/ 978-94-017-8542-6_7
Laux, J. M., Piazza, N. J., Salyers, K., & Roseman, C. P. (2012). The Substance Abuse Subtle Screening Inventory-3 and stages of change: A screening validity study. Journal of Addictions & Offender Counseling, 33, 82–92.
Lazowski, L. E., Miller, F. G., Boye, M. W., & Miller, G. A. (1998). Efficacy of the Substance Abuse Subtle Screening Inventory-3 (SASSI-3) in identifying substance dependence disorders in clinical settings. Journal of Personality Assessment, 71, 114–128. doi: 10.1207/s15327752jpa7101_8
McDonald, R. P. (1999). Test theory: A unified treatment Mahwah, NJ: Erlbaum. doi: 10.1111/j.2044-8317.1981.tb00621.x
Miller, F. G., & Lazowski, L. E. (1999). The Substance Abuse Subtle Screening Inventory-3 (SASSI-3) manual. Springville, IN: The SASSI Institute.
Miller, G. A. (1985). The Substance Abuse Subtle Screening Inventory (SASSI) Manual. Spencer, IN: Spencer Evening World.
Miller, G. A. (1994). The Substance Abuse Subtle Screening Inventory (SASSI): Adult SASSI-2 manual supplement. Spencer, IN: Spencer Evening World.
National Institute on Drug Abuse. (2015). Trends & Statistics. Bethesda, MD: National Institute on Drug Abuse. Retrieved from http://www.drugabuse.gov/related-topics/ trends-statistics
Nelson, J., Perfett, C., Liben, D., & Liben, M. (2012). Measures of text difficulty: Testing their predictive value for grade levels and student performance [Supplemental information for Appendix A]. Retrieved from http://achievethecore.org/page/642/text- complexity-collection
R Development Core Team. (2014). R: A language and environment for statistical computing. Vienna, Austria: Foundation for Statistical Computing. http://www.R-project.org/
Reise, S. P., Bonifay, W. E., & Haviland, M. G. (2013). Scoring and modeling psychological measures in the presence of multidi- mensionality. Journal of Personality Assessment, 95, 129–140. doi: 10.1080/00223891.2012.725437
Revelle, W. (2014). psych: Procedures for personality and psycho- logical research (R package version 1.4.8). Evanston, IL: North- western University. http://CRAN.R-project.org/package=psych
Revelle, W., & Zinbarg, R. E. (2009). Coefficients alpha, beta, omega, and the glb: Comments on Sijtsma. Psychometrika, 74, 145–154. doi: 10.1007/s11336-008-9102-z
Santiago, P. (2014). Substance use disorders. In S. J. Cozza, M. N. Goldenberg, & R. J. Ursano (Eds.), Care of military service members, veterans, and their families (vol. 2, pp. 119–139). Arlington, VA: American Psychiatric Publishing.
96 L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4
European Journal of Psychological Assessment (2019), 35(1), 86–97 �2016 Hogrefe Publishing
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
Stenner, A. J., Burdick, H., Sanford, E. E., & Burdick, D. S. (2007). The Lexile Framework for Reading technical report Durham, NC: MetaMetrics. Retrieved from https://www.lexile.com/ research/9/
Substance Abuse and Mental Health Services Administration. (2014). Results from the 2013 National Survey on Drug Use and Health: Summary of national findings (NSDUH Series H-48, HHS Publication No. (SMA) 14-4863). Retrieved from http://www.samhsa.gov/data/sites/default/files/ NSDUHresultsPDFWHTML2013/Web/NSDUHresults2013.pdf
US Department of Health and Human Services, Behavioral Health Coordinating Committee, Prescription Drug Abuse Subcommittee. (2013). Addressing prescription drug abuse in the United States: Current activities and future opportunities. Retrieved from http://www.cdc.gov/homeandrecreational safety/ overdose/hhs_rx_abuse.html
US Department of Health and Human Services, National Institute on Drug Abuse. (2014). Research Report Series: Prescription Drug Abuse (Third Revision). (NIH Publication No. 15-4881). Retrieved from https://d14rmgtrwzf5a.cloudfront.net/sites/ default/files/prescriptiondrugrrs_11_14.pdf
US Department of Justice, National Drug Intelligence Center. (2011). The economic impact of illicit drug use on American society. (Product No. 2011-Q0317-002). Retrieved from http://www.justice.gov/archive/ndic/pubs44/44731/ 44731p.pdf
Volkow, N. D. (2014, May). America’s addiction to opioids: Heroin and prescription drug abuse. Testimony presented at the
Senate Caucus on International Narcotics Control, Washington, D.C. Retrieved from http://www.drugabuse.gov/about-nida/ legislative-activities/testimony-to-congress/2014/americas- addiction-to-opioids-heroin-prescription-drug-abuse
Williams, R. T., Wilson, C. S., Heinemann, A. W., Lazowski, L. E., Fann, J. R., & Bombardier, C. H., U Washington PRISMS Investigators. (2014). Identifying depression severity risk fac- tors in persons with traumatic spinal cord injury. Rehabilitation Psychology, 59, 50–56. doi: 10.1037/a0034904
Zinbarg, R. E., Revelle, W., Yovel, I., & Li, W. (2005). Cronbach’s alpha, Revelle’s beta, and Mcdonald’s omega H: Their relations with each other and two alternative conceptualizations of reliability. Psychometrika, 70, 123–133.
Received January 9, 2015 Revision received December 9, 2015 Accepted December 14, 2015 Published online October 7, 2016
Linda E. Lazowski The SASSI Institute 201 Camelot Ln Springville, IN 47462 USA Tel. +1 (800) 726-0526 Fax 1 (800) 546-7995 E-mail [email protected]
L. E. Lazowski & B. B. Geary, Validation of the Adult SASSI-4 97
�2016 Hogrefe Publishing European Journal of Psychological Assessment (2019), 35(1), 86–97
Th is
d oc
um en
t i s c
op yr
ig ht
ed b
y th
e A m
er ic
an P
sy ch
ol og
ic al
A ss
oc ia
tio n
or o
ne o
f i ts
a lli
ed p
ub lis
he rs
. Th
is a
rti cl
e is
in te
nd ed
so le
ly fo
r t he
p er
so na
l u se
o f t
he in
di vi
du al
u se
r a nd
is n
ot to
b e
di ss
em in
at ed
b ro
ad ly
.
