Psychometric properties of the Inventory of Interpersonal Problems (IIP-C) used with a clinical sample of adolescents: a preliminary study

IIP background and adaptations

The IIP was developed in the crucible of clinical work with people seeking psychotherapy. Horowitz et al. (7) noted that people’s interpersonal complaints dominated the content and themes of their therapy. This was different from the symptom focus of assessments and outcome measures. This gave impetus to develop the IIP, which consisted of 127 items and had six scales. The alpha coefficients ranged from .82 to .93, and test-retest reliability ranged between .80 to .90 over a 10-week period. Further, the IIP displayed high sensitivity to therapeutic change, and its usefulness as a clinical tool gained popularity.

Over the years, researchers keen on using the measure created various versions of the IIP to best suit their needs. One line of development was motivated by a need for a clinic-friendly measure that was short and brief (8–10). In a review, Hughes & Barkham (11) reported ten shorter versions of IIP. The IIP short version’s popularity has grown globally and it is available in various languages and versions (12). Additionally, studies using IIP have been reported from Italy (13), Turkey (14), and also international students (Chinese) in USA (15). Cronbach alpha for the short form IIP’s ranged from .68 to .92 (10, 16).

Another line of development was influenced by the research-based traditions of interpersonal relationships and the circumplex model (17, 18). Circumplex models emerge from interpersonal theories and provide a framework to organize interpersonal content along two bipolar dimensions of dominance and love. The connotation of dominance is personal control and agency, while love alludes to affiliation, closeness and friendliness (19). The interpersonal problems are organized in a circular fashion, and each of the eight scales is a blend of varying amount of dominance and love. Horowitz et al. (10) postulated that there was a common complaint factor, a general tendency to report distress that varied from patient to patient. However, removing the distress factors by ipsatizing the items and subjecting the scales to PCA, a two-factor solution emerged, reminiscent of the dominant-submissive and hostile-friendly dimensions derived from interpersonal circumplex tradition. Inspired by this discovery and possibility of multiple uses, Alden et al. (20) constructed circumplex scales for the IIP. This work yielded a measure with 64 items and eight scales that were labelled following the Sullivan and Leary tradition (21, 22): Domineering (PA), Vindictive (BC), Cold (DE), Socially avoidant (FG), Nonassertive (HI), Exploitable (JK), Overly Nurturant (LM), & Intrusive (NO). Alden et al. (20) reported alpha coefficients in the range of .72 -.85 for the eight scales. Other studies using IIP-C have also reported alpha coefficients in the same range (23, 24). Additionally, some studies examined the expression of IIP-C compared to the hypothesized circumplex structure. One study based on a community sample reported a bifactor model with an acceptable fit (CFI= .97; RMSEA= .078) (25). Citing previous studies that conducted a formal test of the model (26–28), Monsen et al. (29) conducted confirmatory factor analysis based on a mixed sample of a clinical (n = 374) and a normal reference sample (n = 355). The baseline result prior to applying modification indexes was not acceptable (TLI = .928; CFI = .933; RMSEA = .114). However, after relaxing constraints on three correlations in the normal sample and seven correlations in the clinical sample, the model achieved acceptable fit to the data (TLI = .946; CFI = .977; RMSEA =.078).

Measuring interpersonal problems in adolescents

Two sources provide guidelines for what psychometric instruments are recommended for CAMHS in Norway: the Norwegian Directorate of Health, and the Norwegian Association for Child and Adolescent Psychiatry. Guidelines propose assessing a wide range of factors that cover anamnestic information, symptoms, somatic functioning, cognitive abilities, i.e., intelligence, family and social network, and functioning (30). The Youth Self Report - YSR (31) is widely endorsed for use in CAMHS. While the YSR provides comprehensive information, only two scales address interpersonal issues. The narrow-band competency scales assess a combination of pro-social activities and global questions regarding how well the child/youth gets along with family and friends. The syndrome scale called social problems consists of 11 items of which six items directly address problematic interpersonal problems with peers. Other recommended instruments are the Strengths and Difficulties Questionnaire (SDQ; 32) and HoNOSCA (33), which are primarily screening instruments. Some other instruments target adolescents’ relationships, like The Inventory of Parent and Peer Relationship (IPPA; 34), which measures adolescents’ trust, communication and alienation. While these instruments provide information that aims to aid in well-informed diagnostics and treatment choices, none of them offers a comprehensive assessment of the adolescents’ interpersonal problems.

The IIP is clinically relevant and robust across the wide variety of versions, translations and populations. Further, it found sound footing in the research-based traditions of the interpersonal circumplex. The potential of this combination was highlighted by Wiggins: Horowtiz’ IIP “opened promising new directions of research in such areas as attachment theory, clinical judgement, depression, and psychotherapy” (21, p. 226). Encouraged by these accolades, we wondered if the IIP was useful also with adolescents. A search of the literature showed IIP being used with adolescents (35–37). However, these studies are few, and to our knowledge, there are no studies that have examined the psychometric properties of the IIP-C in a clinical adolescent population.

The aims of the study

The primary aim of this study is to assess the psychometric properties of the IIP-C used with a clinical sample of adolescents (14-18 years) treated in CAMHS. We are primarily interested in examining if the IIP-C results from an adolescent population yield the same factor structure as we see in an adult clinical population. To establish validity, we use a commonly used tool in CAMHS - the Youth Self Report (YSR), and its two broadband syndrome scales of internalizing and externalizing problems. We expect to see a correlation pattern where IIP-C dominant scales will be significantly correlated with the externalizing scale. Conversely, IIP-C submissive scales will be significantly correlated with the internalizing scale. Additionally, the YSR scale “social problems” was of particular interest due to its implicit concordance with interpersonal content. We expected social problems to have high correlation with all of the IIP-C scales.

Procedures

The study was part of a larger study about parent involvement in usual clinical practice. The regional ethics committee (REK-III), University of Bergen approved the project. All the participants and parents of the participants under 16 years of age consented to participate in the study. All new referrals to the clinics received printed information regarding the research study either before or shortly after the first consultation at the clinic. The primary therapist was present during the assessments and answered questions the adolescents may have had. Since the IIP-C was not used in adolescents before, we first asked a panel of six therapists and another panel of five adolescents who were currently receiving in-patient treatment, to provide feedback regarding the instrument. We were interested in their assessment of the comprehension level, instructions, burden of time to complete the assessments and its level of difficulty. Based on their feedback and recommendation, no changes were made to the Norwegian translation.

Participants

A total of 62 adolescents between 14-18 years of age participated in the study. The participants were referred to and treated in outpatient clinics in CAMHS. Information regarding referral reason was available for 70% (n = 44) of the clinical sample. Forty percent (n = 18) was referred due to being sad/depressed 16 % (n = 7) for behavioral problems. The others were spread over a variety of referral reasons like anxiety/phobia (2%), school problems (7%), hyperactivity (9%), eating disorders (9%), autistic symptoms (7%) and (10%) were referred for various other reasons. We had only two exclusion criteria - psychosis and clinician evaluated developmental disorder, i.e., IQ < 70. The mean age of the sample was 15.7 years (SD= 1.04), and 69 % were girls. Fifty-six per cent of the adolescents lived with both parents, 32 % had permanent residence with their mothers, and a small group (2.4 %) commuted between the mother and father. The rest (9.6 %) lived in adoptive homes and residential care facilities.

Inventory of Interpersonal Problems (IIP-C)

The IIP-C consists of 64 problem statements assessed in relation to a significant person and indicated on a Likert scale based on agreement (0 = not at all - 4 = extremely). The first 39 items begin with “It is hard for me to (for example) trust other people and keep things private from other people.” The second 25 items start with “things I do too much” (for example),”I fight with other people too much,” and “I am too sensitive to criticism”. The items form eight scales that are projected on to the interpersonal circumplex. The IIP-C scales have their own abbreviations which are as follows: Domineering/Controlling (PA), Vindictive/Self-centered (BC), Cold/Distant (DE), Socially inhibited (FG), Nonassertive (HI), Exploitable (JK), Overly Nurturant (LM), and Intrusive/Needy (NO). The IIP-C was translated into Norwegian (38) and reported good reliability with alpha coefficients in the range of .67 – .87 (39). This is the version of the translation that we used in the current study.

The Youth Self Report (YSR)

The YSR questionnaire consists of 112 items that adolescents report on a 3-point Likert scale (0 = Not true, 1 = somewhat or sometimes true, and 3 = very true or often true). There are three scales assessing the competence (Activities, Social, and School), eight narrow-band syndrome scales (Anxious/Depressed; Withdrawn/Depressed; Somatic Complaints; Social Problems; Thought Problems; Attention Problems; Rule-Breaking Behavior and Aggressive Behavior). The two broadband scales are made up of Anxious/Depressed, Withdrawn/Depressed and Somatic Complaints (Internalizing Problems) and Rule-breaking behavior and Aggressive behavior (Externalizing problems). The alpha coefficients for the syndrome scales range from .71 - .95 (40). The Norwegian version was found to have acceptable psychometric properties (41).

Confirmatory factor analyses

The fit indices were χ ² ₂₅ = 51.073; RMSEA = .131 (.079 - .182); CFI = .918; TLI = .908; SRMR= 0.281. The RMSEA exceeded the acceptable level of .08, implying that the estimated model was significantly different from the theoretical model. However, CFI and TLI were above > .90 and SRMR was below .80, indicating that the model may hold some promise.

Reliability of the IIP-C scales

Table 1 shows Cronbach’s alpha for the total score (.93) and the eight scales (range = .69 -.84). The highest α coefficient was for HI and lowest for PA. The item-total correlation had a range between .08 and .72. Two items had an item-total correlation below .20. They were item 44, I am too independent, and item 29, it is hard for me to put another's needs before my own. Deleting these items did not affect the alpha coefficient of the total IIP-C scale. However, removing item 44 from the subscale increased the α coefficient of PA by .04, and removing item 29 increased the α coefficient of BC by .02, which are small-to-medium changes and therefore, the items were retained in subsequent analyses.