Psychometric Functioning of The MMPI-2-RF VRIN-r and TRIN-r Scales With Varying Degrees of Randomness, Acquiescence, and Counter-Acquiescence

You might also like

You are on page 1of 10

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/41967807

Psychometric Functioning of the MMPI-2-RF VRIN-r and TRIN-r Scales With


Varying Degrees of Randomness, Acquiescence, and Counter-Acquiescence

Article  in  Psychological Assessment · March 2010


DOI: 10.1037/a0017061 · Source: PubMed

CITATIONS READS
46 3,262

4 authors, including:

Yossef Ben-Porath Robert P Archer


Kent State University Eastern Virginia Medical School
291 PUBLICATIONS   8,026 CITATIONS    165 PUBLICATIONS   4,017 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

MMPI-2-RF View project

All content following this page was uploaded by Robert P Archer on 22 May 2014.

The user has requested enhancement of the downloaded file.


Psychological Assessment © 2010 American Psychological Association
2010, Vol. 22, No. 1, 87–95 1040-3590/10/$12.00 DOI: 10.1037/a0017061

Psychometric Functioning of the MMPI-2-RF VRIN-r and TRIN-r


Scales With Varying Degrees of Randomness, Acquiescence,
and Counter-Acquiescence

Richard W. Handel Yossef S. Ben-Porath


Eastern Virginia Medical School Kent State University

Auke Tellegen Robert P. Archer


University of Minnesota Eastern Virginia Medical School
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

In the present study, the authors evaluated the effects of increasing degrees of simulated non-content-
based (random or fixed) responding on scores on the newly developed Variable Response Inconsistency-
Revised (VRIN-r) and True Response Inconsistency-Revised (TRIN-r) scales of the Minnesota Mul-
tiphasic Personality Inventory-2 Restructured Form (MMPI-2-RF; Y. S. Ben-Porath & A. Tellegen,
2008) and compared the performance of these new scales with the existing VRIN and TRIN scales of the
MMPI-2 (J. N. Butcher et al., 2001). The results support the interpretation of VRIN-r and TRIN-r scores
as measures of random and fixed responding, respectively. Furthermore, the authors examined how
scores on the Restructured Clinical (RC) scales (A. Tellegen et al., 2003) are affected by increasing levels
of non-content-based responding and offer practical interpretive recommendations for test users. Finally,
the results of the present study indicate that RC validity coefficients are relatively robust in the face of
moderate degrees of non-content-based responding.

Keywords: MMPI-2-RF, VRIN-r, TRIN-r, random responding, fixed responding

Test takers must respond to the content of self-report measures sure content non-responsiveness, Scales L (Lie) and K (Correc-
of personality and psychopathology to produce meaningful and tion) were sensitive to fixed responding because of their unbal-
useful test scores. Failure to do so threatens the validity of inter- anced scoring keys (i.e., all L and all but one of the K items are
pretations based on scale scores. Content non-responsiveness can keyed false).
involve various degrees of random or fixed responding to test Tellegen (1988) cited Buechly and Ball’s (1952) Test-Retest
items. Fixed responding refers to a tendency to answer true–false Index (TR) as an early attempt to develop an inconsistency scale
test items with mostly indiscriminate true (i.e., acquiescent) or for the MMPI. The TR was composed of repeated items (16 total
indiscriminate false (i.e., counter-acquiescent) responses. pairs) that were included in the original MMPI for early machine-
The original Minnesota Multiphasic Personality Inventory scoring systems. However, as noted by Tellegen (1988), the TR is
(MMPI; Hathaway & McKinley, 1942) included several scales not sensitive to acquiescent or counter-acquiescent fixed respond-
sensitive to content non-responsiveness. The F (Infrequency) scale ing. Tellegen (1982) developed scales designed to assess random,
was designed to identify problematic protocols marked by exces- acquiescent, and counter-acquiescent responding for the Multidi-
sive infrequent responding, which can be a result of content mensional Personality Questionnaire (MPQ; Tellegen, 1982; Tel-
non-responsiveness. Although not designed specifically to mea- legen & Waller, 2008). These scales were named Variable Re-
sponse Inconsistency (VRIN) and True Response Inconsistency
(TRIN). These MPQ scales influenced the subsequent develop-
Richard W. Handel, Department of Psychiatry and Behavioral Sciences, ment of similar scales for the MMPI-2 (Butcher, Dahlstrom, Gra-
Eastern Virginia Medical School; Yossef S. Ben-Porath, Department of ham, Tellegen, & Kaemmer, 1989; Butcher et al., 2001).
Psychology, Kent State University; Auke Tellegen, Department of Psy- The MMPI-2’s VRIN scale consists of 67 item pairs with either
chology, University of Minnesota; Robert P. Archer, Department of Psy- similar or opposite content. Inconsistent responses to a pair of
chiatry and Behavioral Sciences, Eastern Virginia Medical School. items result in one raw score point being added to the raw score
Yossef S. Ben-Porath and Auke Tellegen are paid consultants to the total. The raw score total is transformed to a linear T score. Since
publisher of the MMPI instruments, the University of Minnesota Press. the introduction of the VRIN scale in the MMPI-2, numerous
They receive royalties on sales of the MMPI-2-RF. Preliminary analyses
analog studies have demonstrated that it is sensitive to random
were presented at the 42nd Annual Symposium on Recent Developments in
responding (Berry et al., 1992, 1991; Clark, Gironda, & Young,
the use of the MMPI-2/MMPI-A, Fort Lauderdale, Florida, April 2007.
Correspondence concerning this article should be addressed to Richard 2003; Cramer, 1995; Gallen & Berry, 1996; Paolo & Ryan, 1992;
W. Handel, Department of Psychiatry and Behavioral Sciences, Eastern Wetter, Baer, Berry, Smith, & Larsen, 1992).
Virginia Medical School, Hofheimer Hall, 825 Fairfax Avenue, Norfolk, The MMPI-2 TRIN scale consists of 20 item pairs that are
VA 23507. E-mail: handelrw@evms.edu opposite in content. Of the 40 possible true–true and false–false

87
88 HANDEL, BEN-PORATH, TELLEGEN, AND ARCHER

response pairs, 14 true–true and nine false–false response pairs are 2. The content of each c-composite, as judged by Tellegen
scored as inconsistencies. The MMPI-2 TRIN scale is scored by and Ben-Porath (2008), had to be inconsistent when
summing the total number of inconsistent true responses, subtract- scored as a positive.
ing the total number of inconsistent false responses, and adding a
constant of 9. Adding 9 to the total raw score precludes negative 3. Each c-composite was required to have a low ratio of
raw score values. Raw score values of less than 9 indicate a observed frequencies to expected frequencies to ensure
counter-acquiescent response style, whereas values greater than 9 that any response pair scored as an inconsistency was
suggest acquiescence. Furthermore, TRIN scale T scores below 50 statistically improbable.
are reflected such that all T scores above 50 are labeled either T or
4. The c-composite was also required to have a low degree
F to indicate a deviation in the acquiescent or counter-acquiescent
of content saturation as determined by a negligible or
direction. Although multiple research studies have investigated the
weak correlation with an additive composite consisting of
psychometric properties of the VRIN scale, only one published
the same two inventory items but now scored to measure
study has focused on TRIN. Handel, Arnau, Archer, and Dandy
the shared content component of the two items. For
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(2006) found that the MMPI-2 (Butcher et al., 2001) TRIN scale
example, suppose the hypothetical item pair (“I am tall”;
This document is copyrighted by the American Psychological Association or one of its allied publishers.

was sensitive to random insertion of either true or false responses


“I am short”) had been combined to create TT and FF
into existing MMPI-2 protocols.
c-composites for possible use as supposedly content-free
The recently released MMPI-2-Restructured Form (MMPI-2-
(counter)acquiescence indicators. Criterion 4 would have
RF; Ben-Porath & Tellegen, 2008) is a 338-item revised version of
called for computing correlations of both these
the MMPI-2 designed to yield an efficient and comprehensive set
c-composites with a measure made up of the same two
of scales that assess clinically relevant and salient constructs
inventory items, but now keyed in opposite directions
represented in the MMPI-2 item pool. The MMPI-2-RF includes
and combined additively. This two-item scale would
two inconsistency measures, VRIN-Revised (VRIN-r) and TRIN-
therefore assess the shared content component of the two
Revised (TRIN-r), that were developed using methods similar to
items, namely, the respondent’s self-reported height.
those used to construct the MMPI-2’s VRIN and TRIN scales.
Low correlations of the c-composites with this scale
However, as indicated by Tellegen and Ben-Porath (2008), some
would have been required as empirical evidence that the
changes were introduced in the development process to accommo-
c-composites were indeed relatively free of content.
date the substantially smaller MMPI-2-RF item pool, yet conduct
a rigorous item selection procedure, and increase the independence 5. The final criterion was designed to prevent the response
of VRIN-r and TRIN-r scores. to a single inventory item from exerting excessive influ-
Tellegen and Ben-Porath (2008) have described the develop- ence on the total VRIN-r or total TRIN-r scale score. It
ment of the VRIN-r and TRIN-r scales in considerable detail. required that no inventory item belong to more than one
Briefly, each raw score unit consists of a pair of inventory items of each of the four kinds of c-composites (TT, TF, FT, or
that are scored configurally to form a compound statement. That is, FF). For example, a T response to a given inventory Item
of the four possible pairs of responses, true-true, true-false, false- I could not be included in two different TF composites,
true, and false-false (or TT, TF, FT, and FF), one is keyed as a but could belong to both a TF composite and a TT
positive score and coded “1,” whereas the three other possible composite.
response pairs are coded “0.” Tellegen and Ben-Porath (2008)
referred to this type of compound statement as a configural com- The resulting VRIN-r scale of the MMPI-2-RF consists of 53
posite, or c-composite. To identify suitable c-composites for c-composites, and the TRIN-r scale consists of 26 c-composites.
VRIN-r and TRIN-r, Tellegen and Ben-Porath (2008) applied the Tellegen and Ben-Porath (2008) noted that these revised inconsis-
following five criteria: tency scales sample a larger percentage of the reduced MMPI-
2-RF item pool compared with the corresponding scales in the
1. The two component inventory items in each c-composite MMPI-2.
had to be reliably correlated. Furthermore, the compo- The total VRIN-r score equals the sum of its c-composite scores;
nent items of each TRIN-r c-composite had to be nega- in other words, it simply equals the number of scored inconsisten-
tively correlated so that the TRIN-r scale would assess cies. Scoring TRIN-r is somewhat more complex, the algorithm
only TT (acquiescent) and FF (counter-acquiescent) in- being similar to the one used for the MMPI-2 TRIN scale. That is,
consistencies. However, only c-composites made up of the total TRIN-r raw score is obtained by summing the individual’s
positively correlated inventory items were considered for TT c-composite scores and subtracting from this value the sum of
the VRIN-r scale because VRIN-r was intended to assess the FF c-composite scores. Consequently, higher TRIN-r raw
exclusively TF and FT inconsistencies, and not (counter) scores indicate higher levels of acquiescence, whereas lower
acquiescence. In other words, item pairs were selected scores indicate higher levels of counter-acquiescence. A constant
such that TRIN-r consisted only of “same response” of 11 is added to preclude negative raw score values.
c-composites and VRIN-r only of “different response” Both VRIN-r and TRIN-r raw scores are converted to linear T
c-composites. These selection rules served to enhance the scores, but in the case of TRIN-r, all T scores below 50 are
independence and distinctiveness of TRIN-r (as an index “reflected.” For example, an initial TRIN-r T score of 35 is
of fixed responding) and VRIN-r (as an index of variable reflected to a T score of 65F. As in the case of the MMPI-2 TRIN
responding). scale, TRIN-r scale T scores below 50 are not possible.
VRIN-r AND TRIN-r SCALES 89

The MMPI-2-RF technical manual (Tellegen & Ben-Porath, these two measures were designed to be content free; furthermore,
2008) includes a brief description of preliminary VRIN-r and largely cooperative individuals in the normative sample would not
TRIN-r analyses presented by Handel, Ben-Porath, Tellegen, and be expected to produce reliable and large variations in invalid
Archer (2007). We update these earlier analyses in the present responding (Tellegen & Ben-Porath, 2008). Tellegen and Ben-
study and present them in their entirety. Specifically, we used a Porath (2008) also indicated that VRIN-r and TRIN-r variances are
computer simulation to introduce increasing degrees of random low and that their standard errors are small enough to warrant their
and fixed responding into existing protocols to evaluate VRIN-r use in invalidating protocols at the recommended cutoff scores. In
and TRIN-r scores in comparison with their MMPI-2 counterparts. a test–retest subsample (n ⫽ 193) of the MMPI-2-RF normative
In addition, we investigated the effects of introducing increasing sample, Tellegen and Ben-Porath (2008) reported test–retest cor-
levels of non-content-based responding on the correlations of relations of .52 for VRIN-r and of .40 for TRIN-r and standard
selected MMPI-2 RC scales with the Brief Psychiatric Rating errors of measurement (SEMs) of 7 T-score points for VRIN-r and
Scale (BPRS; Overall & Gorham, 1988). of 8 for TRIN-r.
For the clinical sample used in the present study, Handel and
Archer (2008) reported RC scale alpha coefficients ranging from
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Method
This document is copyrighted by the American Psychological Association or one of its allied publishers.

.83 to .94 for men and from .82 to .94 for women. These RC alpha
coefficients were based on a sample size of 544 because Handel
Participants
and Archer (2008) used MMPI-2 (rather than MMPI-2-RF) valid-
Two previously published data sets were used in the present ity scales to remove invalid protocols. For VRIN-r, alpha coeffi-
study. The first data set was the MMPI-2 nongendered normative cients of .34 for men and of .44 for women were obtained. For
sample (Ben-Porath & Forbey, 2003), which consisted of 2,276 TRIN-r, alpha coefficients were .49 for men and .45 for women.
participants (1,138 men and 1,138 women) and is a subset of the The BPRS. The 18-item version of the BPRS (Overall &
MMPI-2 (Butcher et al., 2001) normative sample. The sample Gorham, 1988) was used in the present study as an external
comprised 1,861 Whites, 264 Blacks, 71 Native Americans, 67 criterion measure. This version of the BPRS is an extension of
Hispanics, and 13 Asians. Additional demographic information for Overall and Gorham’s (1962) original 16-item version. Although
this sample is provided in Ben-Porath and Forbey (2003). the 18-item version of the BPRS has been in the public domain
The second data set consisted of 704 mental health inpatients since 1965, Overall and Gorham (1988) provided an updated
(363 men and 341 women) originally published by Archer, Griffin, reference for the 18-item version. The BPRS is a clinician rating
and Aiduk (1995). This data set includes individuals who were scale that was designed to provide a global assessment of symp-
administered the MMPI-2 and other measures in one of three toms across patients with a broad range of psychiatric conditions
inpatient psychiatric settings. The mean age of participants in this within the context of a clinical interview (Faustman & Overall,
sample was 33.7 years (SD ⫽ 9.4). The majority of participants 1999). Clinicians rate patients on each of the 18 BPRS dimensions
were White (n ⫽ 461), followed by Black (n ⫽ 144), and those of using a 7-point scale ranging from 1 (not present) to 7 (extremely
other racial/ethnic groups or who did not report race/ethnicity (n ⫽ severe) (Faustman & Overall, 1999). The validity of BPRS score
99). The primary Diagnostic and Statistical Manual of Mental interpretations has largely been evaluated within the context of
Disorders, 3rd edition, revised (DSM–III–R; American Psychiatric controlled studies of medication response (Rhoades & Overall,
Association, 1987) Axis I diagnoses were as follows: schizophre- 1988, as cited in Faustman & Overall, 1999). Faustman and
nia or other psychotic disorders (n ⫽ 91), major depressive disor- Overall (1999) reported that the likely number of published studies
der (n ⫽ 117), bipolar disorder (n ⫽ 60), adjustment disorders in this area is in the thousands. Hedlund and Vieweg (1980)
(n ⫽ 70), substance or alcohol abuse or dependence (n ⫽ 122), provided a comprehensive review of studies that evaluated the
other diagnoses (n ⫽ 111), and missing diagnoses (n ⫽ 133). psychometric properties of BPRS scores. Median interrater reli-
ability for BPRS item scores based on Pearson’s product–moment
correlations ranged from .67 (Blunted Affect) to .88 (Hallucinatory
Measures
Behavior). Hedlund and Vieweg (1980) identified over 150 pub-
The MMPI-2. The MMPI-2-RF item set is a subset of the lished drug treatment studies that provided validity data. These
MMPI-2 item pool. Furthermore, Tellegen and Ben-Porath (2008) studies provided evidence for the validity of BPRS score interpre-
have shown that the properties of MMPI-2-RF scales scored from tations based on relations to other variables. Hedlund and Vieweg
the 567-item MMPI-2 protocols and from the 338-item MMPI- (1980) concluded that BPRS change scores have consistently been
2-RF itself are essentially interchangeable. Consequently, it is shown to reflect treatment changes that are corroborated with
possible to evaluate scores on both pairs of scales, the MMPI-2 extratest data.
VRIN and TRIN scales and the MMPI-2-RF VRIN-r and TRIN-r
scales, based on the current MMPI-2 study samples. Procedure
Tellegen and Ben-Porath (2008) reported alpha coefficients for
VRIN-r of .39 for men and of .20 for women in the MMPI-2-RF Details of the data collection procedures for the MMPI-2 nor-
normative sample. TRIN-r alpha coefficients were .37 for men and mative sample and the clinical sample are provided in Butcher et
.23 for women (Tellegen & Ben-Porath, 2008). Furthermore, Tel- al. (2001) and in Handel and Archer (2008), respectively. For
legen and Ben-Porath (2008) reported RC scale alpha coefficients simulated random responding analyses using the MMPI-2-RF nor-
ranging from .63 to .87 for men and from .63 to .89 for women in mative sample, an n of 2,109 was arrived at by applying validity
the MMPI-2-RF normative sample. The relatively low alpha co- criteria using scales (other than the VRIN and VRIN-r) from both
efficients for VRIN-r and TRIN-r are not surprising given that the MMPI-2 and MMPI-2-RF. All MMPI-2 validity scale T scores
90 HANDEL, BEN-PORATH, TELLEGEN, AND ARCHER

were nongendered. The following validity criteria were used: items were randomly selected to evaluate TRIN, whereas 20% of
MMPI-2 CNS (Cannot Say; raw score) ⬍ 30, MMPI-2-RF CNS the 338 MMPI-2-RF items were selected to evaluate TRIN-r.
(raw score) ⬍ 18, F ⬍ 90, F-r (Infrequent Responses) ⬍ 90, Fp Because the items selected for modification were always randomly
(Infrequency-Psychopathology) ⬍ 80, Fp-r (Infrequent Psychopa- selected at the individual protocol level, it is highly unlikely that
thology Responses) ⬍ 80, FB (Back F) ⬍ 90, L ⬍ 80, L-r any two profiles would have the same exact set of randomly
(Uncommon Virtures) ⬍ 80, TRIN ⬍ 80, and TRIN-r ⬍ 80. For selected items. All mean T scores reported in this study were
simulated fixed responding analyses using the MMPI-2-RF nor- calculated using unrounded, untruncated T scores, as recom-
mative sample, profiles were again removed using the aforemen- mended by Tellegen and Ben-Porath (2008). Finally, mean TRIN
tioned validity criteria with two exceptions. Here, profiles with and TRIN-r T scores were calculated using T scores that were not
VRIN or with VRIN-r T scores ⱖ 80 were removed, whereas no reflected. In other words, T scores below 50 were permitted for the
profiles were removed on the basis of TRIN or TRIN-r scores, purposes of calculating a mean score, and the resulting mean T
resulting in an n of 2,130. score was then reflected when below 50 (e.g., a mean T score of 48
For the clinical sample, the following validity criteria were ob- was reflected to 52F).
served: Fp-r ⬍ 100 (T score), MMPI-2-RF CNS ⬍ 18, and TRIN-r ⬍
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

80 (T score) for random response insertion. For true- and false-


response insertion, the protocol validity criteria were as follows: Results
Fp-r ⬍ 100 (T score), MMPI-2-RF CNS ⬍ 18, and VRIN-r ⬍ 80
(T score). For random response insertion, 525 valid MMPI-2-RF Table 1 shows the mean nongendered T scores for the VRIN and
profiles were available, but the sample size ranged from 442 to 446 VRIN-r scales using the various simulated random response con-
because of missing BPRS data. For true- and false-response insertion, ditions. Furthermore, the percentage of cases in each condition
the valid profiles numbered 545, whereas for the correlation coeffi- with T scores ⱖ 80 is provided. As seen in Table 1, mean scores
cients, ns ranged from 459 to 463. for both VRIN and VRIN-r increased monotonically as greater
For the present study, the procedure involved a computer sim- degrees of simulated random responding were introduced into
ulation of the effects of inserting increasing levels of simulated protocols. Mean scores for VRIN and VRIN-r were highly com-
random responding, acquiescence, and counter-acquiescence into parable in all conditions of simulated random responding. Further-
existing MMPI-2 protocols. To simulate increasing degrees of more, both scales identified a majority of these partially random
random responding, percentages (i.e., 10%, 20%, 30%, 40%, 50%, protocols after at least 50% of the responses were replaced with
60%, and 70%) of items were selected at the individual protocol random responses. The MMPI-2 VRIN scale produced a slightly
level and replaced with a randomly generated true or false re- greater number of elevated cases in most conditions.
sponse. The insertion of random responses for the VRIN-r analyses Similar analyses for TRIN and TRIN-r using simulated acqui-
was restricted to the subset of 338 MMPI-2-RF items in the escence are presented in Table 2. As seen in Table 2, mean T
MMPI-2 protocols. This ensured that exact percentages (e.g., scores again increased monotonically with increasing degrees of
exactly 20%) of items were selected for each case from the randomly inserted true responses. Again, mean T scores were
MMPI-2 and MMPI-2-RF item pools, allowing direct comparisons generally comparable for TRIN and TRIN-r under the various
between the VRIN and VRIN-r results. The same procedure was response insertion conditions. In contrast to the VRIN scales in the
followed to simulate increasing acquiescence and counter- assessment of simulated random responding, a much lower per-
acquiescence. For example, in the 20% acquiescent condition, 20% centage (i.e., 30%) of true-response insertion was required for the
of the items for each case were randomly selected and replaced majority of cases to produce T scores ⱖ 80 for both TRIN and
with true responses. In this example, 20% of the 567 MMPI-2 TRIN-r.

Table 1
MMPI-2 VRIN and MMPI-2-RF VRIN-r Mean T-Scores and Percentage of Cases With
T-Scores ⱖ 80 for Varying Degrees of Random Response Insertion—Normative Sample
(n ⫽ 2,109)

VRIN VRIN-r
Random insertion Percentage ⱖ Percentage ⱖ
percentage M SD T-score of 80 M SD T-score of 80

0% 49.6 9.8 0.4 49.5 9.5 0.5


10% 57.6 10.3 2.5 57.1 10.6 2.5
20% 65.0 10.6 9.3 64.3 10.8 8.1
30% 71.5 11.2 24.7 70.4 11.9 21.0
40% 77.4 11.5 43.8 76.0 12.7 36.8
50% 82.0 11.6 58.7 81.5 12.8 53.7
60% 86.6 12.0 73.3 86.0 13.7 66.5
70% 89.8 12.4 80.2 90.0 13.6 77.0

Note. MMPI-2 ⫽ Minnesota Multiphasic Personality Inventory-2; MMPI-2-RF ⫽ Minnesota Multiphasic


Personality Inventory-2-Restructured Form; VRIN ⫽ Variable Response Inconsistency; VRIN-r ⫽ Variable
Response Inconsistency-Revised.
VRIN-r AND TRIN-r SCALES 91

Table 2
MMPI-2 TRIN and MMPI-2-RF TRIN-r Mean T-Scores and Percentage of Cases With
T-Scores ⱖ 80 for Varying Degrees of True-Response Insertion—Normative Sample (n ⫽ 2,130)

TRIN TRIN-r
True-insertion Percentage ⱖ Percentage ⱖ
percentage M SD T-score of 80T M SD T-score of 80T

0% 50.2F 9.4 0.6 50.2F 9.3 0.8


10% 58.9T 11.4 6.5 59.5T 11.8 8.0
20% 67.7T 12.8 25.3 69.7T 13.2 29.4
30% 77.3T 14.1 52.4 79.9T 14.5 58.8
40% 87.6T 14.9 78.0 90.1T 15.0 82.3
50% 97.4T 14.4 93.1 101.1T 15.0 95.2
60% 108.3T 14.0 98.9 113.1T 14.7 99.3
70% 119.0T 13.2 99.8 125.5T 13.4 100.0
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Note. MMPI-2 ⫽ Minnesota Multiphasic Personality Inventory-2; TRIN ⫽ True Response Inconsistency;
MMPI-2-RF ⫽ Minnesota Multiphasic Personality Inventory-2-Restructured Form; TRIN-r ⫽ True Response
Inconsistency-Revised; T ⫽ True; F ⫽ False.

Table 3 provides the results for simulated counter-acquiescence. column, followed by a “T” for simulated acquiescent responding,
Here, we used a T score of ⱖ 79 to define an elevation for an “F” for simulated counter-acquiescent responding, or an “R” for
nongendered TRIN, whereas we used a T score ⱖ 80 to define an simulated random responding.
elevation for TRIN-r. The rationale for the slightly lower cutoff In the MMPI-2-RF normative sample, SEMs range from 3 T-score
score for TRIN is that a TRIN raw score of 5 corresponds to a T points for RCd (demoralization) to 6 points for RC2 (low-positive
score of 79F, whereas a raw score of 4 corresponds to a T score of emotions), RC6 (ideas of persecution), and RC8 (aberrant experi-
87F. Therefore, defining an elevated case as one with a T score ⱖ ences) for men (Tellegen & Ben-Porath, 2008). For women, SEMs
80F for TRIN would result in a de facto cutoff score of 87F. also range from 3 T-score points (RCd) to 6 T-score points (RC2 and
However, this phenomenon does not occur for TRIN-r for which a RC6). We used these SEM values as benchmarks for substantive
raw score of 7 corresponds to a T score of 80F. As seen in Table deviations from mean scores under various conditions of simulated
3, TRIN-r tended to produce greater mean scores than did TRIN, non-content-based responding. When there was a difference in SEM
and this discrepancy became more pronounced as increasing de- values between genders, we used the smaller of the two values. In
grees of simulated counter-acquiescence were introduced into pro- Table 4, superscripts are used to designate mean T scores that deviated
tocols. Furthermore, TRIN-r consistently produced a greater per- from the baseline mean RC scores (i.e., 0% response insertion) by at
centage of cases with T scores ⱖ 80. least one, two, or three times the SEM for each scale (these values are
Next, we evaluated the effects of introducing varying degrees of denoted by superscripts a, b, and c, respectively).
simulated non-content-based responding on the RC scale scores. RC6 was the most sensitive scale to simulated random responding
This series of analyses bears on the issue of the conditions under followed closely by RC8. For RC6, the mean score deviated from the
which interpretation of individual RC scores may become prob- baseline score by at least twice the SEM after only 20% simulated
lematic due to score distortion owing to non-content-based re- random responding was introduced. Similarly, the mean RC8 score
sponding. Table 4 provides the results of these analyses. In this deviated from baseline by at least twice the SEM after 30% simulated
table, the percentage of randomly selected items is listed in the first random responding was introduced. RCd and RC1 were the next scales

Table 3
MMPI-2 TRIN and MMPI-2-RF TRIN-r Mean T-Scores and Percentage of Elevated Cases for
Varying Degrees of False-Response Insertion—Normative Sample (n ⫽ 2,130)

TRIN TRIN-r
False-insertion Percentage ⱖ Percentage ⱖ
percentage M SD T-score of 79F M SD T-score of 80F

0% 50.2F 9.4 0.5 50.2F 9.3 0.6


10% 56.4F 10.4 3.1 57.3F 10.8 4.0
20% 62.2F 11.3 11.1 64.7F 12.4 16.5
30% 68.9F 11.6 27.5 72.0F 12.6 36.0
40% 75.3F 11.6 49.5 80.5F 13.6 62.3
50% 81.9F 12.0 70.2 88.5F 13.5 81.1
60% 88.8F 11.5 87.5 96.7F 12.9 94.0
70% 95.5F 10.3 96.9 105.1F 12.2 99.1

Note. MMPI-2 ⫽ Minnesota Multiphasic Personality Inventory-2; TRIN ⫽ True Response Inconsistency;
MMPI-2-RF ⫽ Minnesota Multiphasic Personality Inventory-2-Restructured Form; TRIN-r ⫽ True Response
Inconsistency-Revised; T ⫽ True; F ⫽ False.
92 HANDEL, BEN-PORATH, TELLEGEN, AND ARCHER

Table 4
The Effects of Varying Degrees of Random, True-, and False-Response Insertion on Mean RC Scale T-Scores

Response RCd dem RC1 som RC2 lpe RC3 cyn RC4 asb RC6 per RC7 dne RC8 abx RC9 hpm
insertion
percentage M SD M SD M SD M SD M SD M SD M SD M SD M SD

0% (R) 49.5 9.6 49.5 9.6 49.8 9.7 49.5 9.7 49.6 9.8 49.2 9.3 49.6 9.5 49.3 9.5 49.7 9.6
0% (T or F) 49.6 9.6 49.5 9.6 49.8 9.7 49.6 9.8 49.7 9.8 49.3 9.4 49.6 9.6 49.4 9.6 49.8 9.7
10% (R) 51.9 8.2 53.3 8.5 51.5 9.2 50.0 8.8 51.6 9.1 56.1 9.4a 51.0 8.7 53.0 8.8 50.1 9.0
10% (T) 54.1 7.8a 52.4 8.6 48.3 9.1 52.1 9.5 53.1 9.1 60.8 9.4a 53.7 8.8a 56.8 8.5a 52.8 9.4
10% (F) 49.2 8.9 53.8 8.3 54.8 9.4 47.8 8.4 50.3 9.0 49.8 9.2 48.2 8.7 49.0 9.1 47.6 8.6
20% (R) 53.6 7.3a 56.4 7.6a 53.3 8.9 50.3 8.0 53.6 8.5 61.4 9.2b 52.4 7.9 56.1 8.2a 50.4 8.1
20% (T) 57.3 6.8b 54.9 7.7a 46.7 8.4 54.6 9.4a 56.4 8.4a 69.0 9.4c 57.7 8.5b 62.2 7.8b 56.1 9.3a
20% (F) 48.9 8.3 57.6 7.4a 59.6 9.4a 46.4 7.5 50.7 8.4 50.4 8.8 46.8 8.0 49.0 8.6 45.5 7.7a
30% (R) 55.1 6.2 58.8 7.2 54.9 8.7 50.5 7.5 55.4 8.0a 65.6 8.8 53.5 7.3 59.4 7.6b 50.9 7.7
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

30% (T) 60.1 6.6c 57.1 6.9 45.3 7.9 57.0 9.2 59.6 8.0 76.5 9.2 61.4 8.4 67.5 7.9c 59.4 9.3b
This document is copyrighted by the American Psychological Association or one of its allied publishers.

30% (F) 48.7 7.4 61.0 6.4b 64.7 9.1b 45.0 6.8 51.1 7.7 50.9 8.5 45.5 7.4a 48.6 8.0 43.5 6.6
40% (R) 56.3 5.5b 61.5 6.6b 56.8 8.3a 51.0 6.8 57.3 7.5 69.7 8.6c 54.6 6.9a 61.8 7.2 51.3 7.1
40% (T) 62.9 6.1 59.1 6.2 43.8 7.2a 59.8 9.1b 62.8 7.2b 83.5 9.3 65.9 8.2c 72.8 7.5 62.9 9.0c
40% (F) 48.3 6.6 63.8 6.0 69.3 9.0c 43.5 5.8a 51.7 6.7 51.2 8.2 44.2 6.8 48.3 7.4 41.6 6.1b
50% (R) 57.4 5.1 63.8 6.5 58.5 8.2 51.4 6.5 59.1 7.2 73.4 8.8 55.8 6.5 64.3 7.4c 51.7 6.6
50% (T) 66.2 5.7 61.1 5.5b 42.2 6.5 63.2 9.0 66.0 6.9c 90.5 9.5 70.5 7.7 78.0 7.4 66.8 8.2
50% (F) 47.9 5.9 66.9 5.6c 74.5 8.2 42.2 5.4 52.2 6.0 52.0 7.7 42.7 6.1 48.1 6.8 39.8 5.5
60% (R) 58.5 4.7c 66.3 6.1c 60.1 7.9 51.7 6.1 60.9 6.9b 76.9 9.0 57.3 6.4 66.8 7.2 51.9 6.2
60% (T) 69.3 5.1 62.8 5.0 40.6 5.9 66.9 8.3c 68.9 6.4 97.5 8.9 75.1 6.8 82.5 6.9 71.0 7.5
60% (F) 47.6 5.1 69.8 5.0 79.5 7.6 40.8 4.9 52.6 5.2 52.4 7.2 41.2 5.5b 47.8 6.3 37.7 4.8c
70% (R) 59.9 4.4 68.5 6.1 62.0 8.0b 52.0 5.6 62.8 6.9 80.5 9.1 58.4 6.0b 68.9 7.0 52.5 5.9
70% (T) 72.8 4.6 64.8 4.3c 39.0 5.0 70.8 7.6 72.1 5.4 105.1 8.3 79.6 6.0 87.9 6.0 75.4 6.5
70% (F) 47.3 4.5 72.5 4.4 84.2 7.0 39.2 4.3b 53.0 4.4 53.4 6.2 39.5 4.9 47.4 5.3 35.7 4.4

Note. n ⫽ 2,109 for random response insertion and n ⫽ 2,130 for true- and false-response insertion. RC ⫽ Restructured Clinical scale; R ⫽ random; T ⫽
true; F ⫽ false; dem ⫽ Demoralization; som ⫽ Somatic Complaints; lpe ⫽ Low Positive Emotions; cyn ⫽ Cynicism; asb ⫽ Antisocial Behavior; per ⫽
Ideas of Persecution; dne ⫽ Dysfunctional Negative Emotions; abx ⫽ Aberrant Experiences; hpm ⫽ Hypomanic Activation.
a
The mean score deviates from the baseline mean score (i.e., 0% random, true-, or false-insertion) by one times the standard error of measurement for this
scale. b The mean score deviates from the baseline mean score by two times the standard error of measurement for this scale. c The mean score deviates
from the baseline mean score by three times the standard error of measurement for this scale. All standard errors of measurement are based on the Minnesota
Multiphasic Personality Inventory-2-Restructured Form normative sample as reported in Tellegen and Ben-Porath (2008). In cases in which there were
gender differences in standard-error-of-measurement values, the lower of the two values was used.

most susceptible to distortion from this response set, reaching twice the RC6, RC7, and RC8). The BPRS items used for these analyses are
SEM from baseline after 40% random response insertion. indicated at the bottom of Table 5. Baseline correlations (i.e., 0%
For simulated acquiescence, RCd, RC6, RC7, and RC8 mean response insertion) between the RC scales and the BPRS items were
scores all deviated by at least one SEM after 10% true-response originally published by Handel and Archer (2008). Some correlation
insertion and two times the SEM after 20% true-response insertion coefficients presented in Table 5 are slightly different from those
(three times the SEM in the case of RC6). RC4 and RC9 mean presented in Handel and Archer (2008) because the two studies did
scores reached one SEM from the baseline after 20% true-response not use the same protocol validity criteria.
insertion and at least twice the SEM after 30% true-response As seen in Table 5, all RC validity coefficients were found to be
insertion for RC9 and 40% true-response insertion for RC4. impervious to simulated random responding rates of less than 30%.
RC1 and RC2 mean scores were most sensitive to simulated To evaluate the degradation in validity coefficients, we squared the
counter-acquiescence deviating by at least one SEM after we Pearson r values in Table 5 and subtracted these from their baseline
introduced 20% false responses and by two times the SEM after values. At the 30% level of simulated random responding, we noted
we introduced 30% false responses. RC9 also deviated by at least some degradation in validity coefficients, particularly for RC1 (ac-
one times the SEM after we introduced 20% false responses but by counting for 4% less variance than at baseline), RC2 (5% less vari-
two times the SEM when we inserted 40% false responses. RC3 ance), and RC8 (5% less variance). When we inserted 70% random-
deviated by at least one times the SEM at 40% simulated counter- ness, the reduction in percentage of variance accounted for ranged
acquiescence and by two times the SEM at 70% false-response from 4% for RC7 to 11% for RC2. As seen in Table 5, and as
insertion. expected, simulated randomness had no large or systematic effect on
Finally, we evaluated the effects of increasing degrees of non- discriminant validity coefficients. We found similar results for simu-
content-based responding on RC convergent and discriminant validity lated acquiescent responding: The introduction of random acquies-
coefficients. For these analyses (presented in Table 5), we only cence of less than 30% had minimal effect on validity coefficients. At
evaluated RC scales that measure constructs with clear conceptual 30%, validity coefficients for RC1 (5%) and RC6 (5%) were most
counterparts in the BPRS (Overall & Gorham, 1988). We rationally affected. At 70% simulated acquiescence, the degradation of validity
selected one representative convergent correlation and one represen- coefficients ranged from 3% for RC8 to 7% for RC2. Similar to the
tative discriminant correlation for each scale (i.e., RCd, RC1, RC2, findings for simulated random responding, the introduction of ran-
VRIN-r AND TRIN-r SCALES 93

Table 5
The Effects of Varying Degrees of True-, False-, and Random Response Insertion on RC Scale Correlations With Variables From the
Brief Psychiatric Rating Scale

Response insertion
percentage RCda dem RC1b som RC2a lpe RC6c per RC7d dne RC8e abx

0% (R) .38/.08 .40/.09 .41/.06 .29/.02 .27/.03 .32/.00


0% (T or F) .38/.10 .40/.09 .40/.08 .30/.02 .27/.04 .32/.00
10% (R) .37/.08 .37/.09 .41/.06 .27/.03 .25/.01 .30/⫺.01
10% (T) .39/.10 .38/.09 .39/.07 .28/.01 .26/.04 .30/⫺.02
10% (F) .38/.10 .39/.07 .40/.09 .29/.01 .27/.03 .32/.01
20% (R) .37/.08 .37/.08 .41/.06 .30/.06 .26/.00 .29/⫺.02
20% (T) .37/.11 .36/.08 .39/.06 .24/.06 .25/.02 .29/⫺.01
20% (F) .38/.07 .36/.07 .38/.07 .29/.01 .28/.03 .33/.00
30% (R) .34/.07 .34/.06 .34/.07 .25/.01 .26/.02 .22/.00
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

30% (T) .38/.09 .33/.09 .38/.07 .21/.00 .26/.04 .28/⫺.03


This document is copyrighted by the American Psychological Association or one of its allied publishers.

30% (F) .34/.10 .36/.11 .35/.05 .31/.02 .25/.03 .30/.01


40% (R) .34/.08 .34/.08 .36/.04 .21/⫺.03 .27/.05 .25/.00
40% (T) .34/.08 .36/.07 .34/.07 .32/.03 .23/.03 .26/.04
40% (F) .34/.09 .37/.07 .35/.06 .30/.01 .24/.03 .26/⫺.02
50% (R) .33/.08 .34/.07 .36/.01 .20/⫺.03 .18/.01 .20/.04
50% (T) .33/.08 .36/.08 .26/.08 .19/.03 .24/.05 .26/.01
50% (F) .31/.12 .37/.12 .31/.06 .26/.02 .22/⫺.01 .28/.00
60% (R) .30/.09 .23/.11 .23/.07 .10/⫺.03 .19/.03 .18/.06
60% (T) .31/.07 .34/.02 .33/.05 .18/⫺.01 .22/⫺.02 .20/⫺.05
60% (F) .34/.13 .29/.09 .32/.03 .24/.03 .24/.00 .24/.00
70% (R) .23/.10 .28/.05 .24/.05 .12/.02 .19/.01 .20/.04
70% (T) .32/.03 .33/.07 .30/.08 .17/⫺.02 .19/⫺.01 .26/.03
70% (F) .31/.07 .31/.07 .25/.04 .21/.02 .19/.01 .22/⫺.02

Note. N ranges from 442 to 446 for random response insertion. N ranges from 459 to 463 for true- and false-response insertion. RC ⫽ Restructured
Clinical scale; R ⫽ random response insertion; T ⫽ true-response insertion; F ⫽ false-response insertion; dem ⫽ Demoralization; som ⫽ Somatic
Complaints; lpe ⫽ Low Positive Emotions; per ⫽ Ideas of Persecution; dne ⫽ Dysfunctional Negative Emotions; abx ⫽ Aberrant Experiences. In each
cell, the correlation coefficient to the left of the forward slash is the convergent correlation, and the correlation coefficient to the right of the forward slash
is the discriminant correlation.
For these analyses, convergent and discriminant correlations were assigned as follows (Convergent Correlation/Discriminant Correlation): a Depressive
Mood/Hallucinatory Behavior. b Somatic Concern/Suspiciousness. c Suspiciousness/Depressive Mood. d Anxiety/Uncooperativeness. e Hallucina-
tory Behavior/Guilt Feelings.

domly inserted true responses had little effect on discriminant validity protocols. The MMPI-2 VRIN scale produced a slightly greater num-
coefficients. We obtained similar results for simulated counter- ber of elevated cases (i.e., T score ⱖ 80) in most conditions.
acquiescence. At the 30% level of simulated counter-acquiescence, Similarly, the introduction of simulated acquiescent responding
the difference in variance accounted for in comparison to baseline was into existing protocols results in comparable mean T scores for
4% for RC2, 3% for RCd, and 3% for RC1. At 70% simulated TRIN and TRIN-r as well as similar percentages of elevated cases.
counter-acquiescence, the RC2 validity coefficient accounted for 10% However, TRIN-r produced greater mean T scores than did TRIN
less variance than at baseline. The degradation of other validity when counter-acquiescence was simulated, and this phenomenon
coefficients ranged from 4% for RC7 to 6% for RC1. Discriminant becomes more pronounced as greater degrees of false responses
validity coefficients showed little change (similar to the findings for are inserted into protocols. Not surprisingly, TRIN-r also is more
simulated randomness and acquiescence). sensitive to counter-acquiescence than TRIN as evidenced by more
than a 10% difference in the number of elevated cases for the 40%
Discussion and 50% response insertion conditions.
In the present study, we evaluated the psychometric properties The present study allows for several practical clinical inter-
of scores on the new MMPI-2-RF (Ben-Porath & Tellegen, 2008) pretive recommendations. Ben-Porath and Tellegen (2008) rec-
VRIN-r and TRIN-r scales under various conditions of simulated ommended that profiles with T scores ⱖ 70 on VRIN-r or on
inconsistent responding. We also explored the effects of varying TRIN-r be interpreted cautiously. Furthermore, T scores ⱖ 80
degrees and kinds of inconsistency on mean T scores from the RC on either VRIN-r or TRIN-r are considered to invalidate a
scales (Tellegen et al., 2003). Finally, we investigated the extent to profile. Because we used a subset of the MMPI-2-RF normative
which RC validity coefficients are degraded when subjected to sample in the present study, and therefore the mean T scores of
increasing degrees of simulated inconsistent responding. the RC scales are near 50 prior to non-content-based response
The results of the present study indicate that the MMPI-2 VRIN insertion, we were able to examine directly the effects of
and the MMPI-2-RF VRIN-r scales produce nearly identical mean simulated inconsistent responding on the mean T scores of each
linear T scores when random responses are introduced into existing of the individual scales. A mean VRIN-r T score of 70 occurs
94 HANDEL, BEN-PORATH, TELLEGEN, AND ARCHER

after approximately 30% of MMPI-2-RF items have been mod- Ben-Porath, Y. S., & Forbey, J. D. (2003). Non-gendered norms for the
ified to random responses. At this 30% level of quasi- MMPI-2. Minneapolis: University of Minnesota Press.
randomness, test users should be particularly cautious about Ben-Porath, Y. S., & Tellegen, A. (2008). MMPI-2-RF (Minnesota Mul-
interpreting scores on RC1, RC6, and RC8, as these mean scale tiphasic Personality Inventory-2-Restructured Form): Manual for ad-
scores deviated approximately 10 T-score points from baseline ministration and scoring. Minneapolis: University of Minnesota Press.
Berry, D. T. R., Wetter, M. W., Baer, R. A., Larsen, L., Clark, C., &
for RC1 and RC8 and approximately 15 T-score points for RC6.
Monroe, K. (1992). MMPI-2 random responding indices: Validation
These changes are likely to alter clinical interpretation in many using a self-report methodology. Psychological Assessment, 4, 340 –345.
cases. Whereas 30% random responding raises the RC1, RC6, Berry, D. T. R., Wetter, M. W., Baer, R. A., Widiger, T. A., Sumpter, J. C.,
and RC8 T-score means approximately one standard deviation Reynolds, S. K., & Hallam, R. A. (1991). Detection of random respond-
above baseline, acquiescence has a particularly marked influ- ing on the MMPI–2: Utility of F, back F, and VRIN scales. Psycholog-
ence on RC6 and RC8. The mean TRIN-r T score reaches ical Assessment: A Journal of Consulting and Clinical Psychology, 3,
approximately 70 after 20% of items have been randomly 418 – 423.
selected and modified to “true” responses. At this relatively Butcher, J. N., Dahlstrom, W. G., Graham, J. R., Tellegen, A., & Kaemmer, B.
modest level of simulated acquiescence, RC6 scores are dis- (1989). Minnesota Multiphasic Personality Inventory-2 (MMPI-2): Manual
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

for administration and scoring. Minneapolis: University of Minnesota


This document is copyrighted by the American Psychological Association or one of its allied publishers.

torted substantially, followed closely by scores on RC8. There-


fore, the results of the present study indicate that test users Press.
Butcher, J. N., Graham, J. R., Ben-Porath, Y. S., Tellegen, A., Dahlstrom,
should be particularly cautious when interpreting RC6 and RC8
W. G., & Kaemmer, B. (2001). MMPI-2 (Minnesota Multiphasic Per-
for cases in which TRIN scores exceed 69 in the acquiescent
sonality Inventory-2): Manual for administration and scoring (rev. ed.).
direction. After 30% false-response insertion, the mean TRIN-r Minneapolis: University of Minnesota Press.
T score reaches approximately 70. Here, RC1 and RC2 scores Clark, M. E., Gironda, R. J., & Young, R. W. (2003). Detection of back
are most likely to be affected, and test users should be partic- random responding: Effectiveness of MMPI-2 and Personality Assessment
ularly cautious in interpreting scores from these two scales. Inventory validity indices. Psychological Assessment, 15, 223–234.
Overall, the results of this investigation support the interpretive Cramer, K. M. (1995). Comparing three new MMPI-2 randomness indices
recommendations for VRIN-r and TRIN-r scores in the MMPI- in a novel procedure for random profile derivation. Journal of Person-
2-RF manual (Ben-Porath & Tellegen, 2008). By following these ality Assessment, 65, 514 –520.
recommendations, MMPI-2-RF users should be able to identify Faustman, W. O., & Overall, J. E. (1999). Brief Psychiatric Rating Scale. In M.
protocols with excessive levels of non-content-based responsive- Maruish (Ed.), The use of psychological testing for treatment planning and
ness. However, future studies need to be conducted to fully eval- outcome assessment (pp. 791– 830). Mahwah, NJ: Erlbaum.
Gallen, R. T., & Berry, D. T. R. (1996). Detection of random responding
uate sensitivity, specificity, positive predictive power, and nega-
in MMPI-2 protocols. Assessment, 3, 171–178.
tive predictive power with varying base rates of randomness,
Handel, R. W., & Archer, R. P. (2008). An investigation of the psycho-
acquiescence, and counter-acquiescence. metric properties of the MMPI-2 Restructured Clinical (RC) Scales with
For the six RC scales with conceptually related BPRS items, mental health inpatients. Journal of Personality Assessment, 90, 239 –
some degradation in validity coefficients occurs with increasing 249.
simulated non-content-based responding. However, all RC validity Handel, R. W., Arnau, R. C., Archer, R. P., & Dandy, K. L. (2006). An
coefficients appear to be relatively robust to modest levels of evaluation of the MMPI-2 and MMPI-A True Response Inconsistency
non-content-based responding. Nevertheless, our findings indicate (TRIN) scales. Assessment, 13, 98 –106.
that at a T-score elevation of 70 on VRIN-r or on TRIN-r, T scores Handel, R. W., Ben-Porath, Y. S., Tellegen, A., & Archer, R. P. (April,
on substantive scales (e.g., the RC scales) may be distorted sig- 2007). Psychometric functioning of the MMPI-2-RF VRIN-r and TRIN-r
nificantly as a result of non-content-based responding, which scales with varying degrees of randomness, acquiescence, and nonac-
would substantially affect their interpretation. quiescence. Paper presented at the 42nd annual meeting on Recent
Developments in MMPI-2 and MMPI-A Research, Fort Lauderdale, FL.
The present study is not without limitations. The degree to
Hathaway, S. R., & McKinley, J. C. (1942). The Minnesota Multiphasic
which our simulation of non-content-based responding approx-
Personality Schedule. Minneapolis: University of Minnesota Press.
imates how actual test-takers may engage in content non- Hedlund, J. L., & Vieweg, B. W. (1980). The Brief Psychiatric Rating
responsiveness is unknown. In our procedure, each item was Scale (BPRS): A comprehensive review. Journal of Operational Psy-
given an equal probability of being selected for response mod- chiatry, 11, 48 – 62.
ification. However, the content non-responsiveness of some test Overall, J. E., & Gorham, D. R. (1962). The brief psychiatric rating scale.
takers may vary across different parts of the test booklet. Psychological Reports, 10, 799 – 812.
Furthermore, we were unable to evaluate the effects of non- Overall, J. E., & Gorham, D. R. (1988). Introduction—The Brief Psychi-
content-based responding on validity coefficients for RC3, atric Rating Scale (BPRS): Recent developments in ascertainment and
RC4, and RC9. Future studies need to address the effects of scaling. Psychopharmacology Bulletin, 24, 97–98.
content nonresponsiveness on these additional RC scales as Paolo, A. M., & Ryan, J. J. (1992). Detection of random response sets on
the MMPI-2. Psychotherapy in Private Practice, 11, 1– 8.
well as on additional scales included in the MMPI-2-RF (Ben-
Tellegen, A. (1982). Brief manual for the Differential Personality Ques-
Porath & Tellegen, 2008).
tionnaire. Unpublished manuscript.
Tellegen, A. (1988). The analysis of consistency in personality assessment.
References Journal of Personality, 56, 621– 663.
American Psychiatric Association. (1987). Diagnostic and statistical man- Tellegen, A., & Ben-Porath, Y. S. (2008). MMPI-2-RF (Minnesota Mul-
ual of mental disorders (3rd ed., revised). Washington, DC: Author. tiphasic Personality Inventory-2-Restructured Form): Technical man-
Archer, R. P., Griffin, R., & Aiduk, R. (1995). MMPI-2 clinical correlates ual. Minneapolis: University of Minnesota Press.
for ten common codes. Journal of Personality Assessment, 65, 391– 407. Tellegen, A., Ben-Porath, Y. S., McNulty, J. L., Arbisi, P. A., Graham,
VRIN-r AND TRIN-r SCALES 95

J. R., & Kaemmer, B. (2003). MMPI-2 Restructured Clinical (RC) Wetter, M. W., Baer, R. A., Berry, D. T. R., Smith, G. T., & Larsen, L. H.
Scales: Development, validation, and interpretation. Minneapolis, MN: (1992). Sensitivity of MMPI-2 validity scales to random responding and
University of Minnesota Press. malingering. Psychological Assessment, 4, 369 –374.
Tellegen, A., & Waller, N. G. (2008). Exploring personality through test
construction: Development of the Multidimensional Personality Ques-
tionnaire. In G. J. Boyle, G. Matthews, and D. H. Saklofske (Eds.), The Received March 3, 2009
sage handbook of personality theory and assessment: Vol. II, Person- Revision received June 1, 2009
ality measurement and testing (Vol. 2, pp. 261–292). London: Sage. Accepted June 2, 2009 䡲
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

E-Mail Notification of Your Latest Issue Online!


Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at http://notify.apa.org/ and you will be
notified by e-mail when issues of interest to you become available!

View publication stats

You might also like