Factor structure and response bias of the Obsessive-Compulsive Inventory-Revised (OCI-R) in a female undergraduate sample from New Zealand.(Report)

As self-report measures of clinical constructs are increasingly administered to student populations, it is important that the psychometric properties of such measures are investigated. Additionally, the response bias intrinsic in self-report responses requires further understanding. The factor structure and response bias of the Obsessive-Compulsive Inventory-Revised (OCI-R) is investigated in a subclinical sample of 282 female students from New Zealand. The present study adds to previous research by using not only standard confirmatory factor analysis but also hierarchical confirmatory factor analysis, in parallel with an investigation of response bias. The six-factor model provided the most appropriate fit to the data, with a single latent factor driving the six differential factors. Further support is provided for strong internal reliability of the OCI-R. Overall, subscales of the OCI-R were robustly unrelated to the response bias of impression management.

**********

As research on Obsessive-Compulsive Disorder (OCD) continues to diversify, subclinical (particularly student) samples are being drawn on more frequently to inform our knowledge of this pathology. It has been found that subclinical OCD exhibits very similar, albeit lessened, symptoms to that of clinical OCD (Gibbs, 1996). Self-report measures provide a fast and convenient data collection method, and are therefore particularly useful in subclinical research. It is however important that the psychometric properties of employed measures are understood in this population, so that we can be confident of the validity of findings from this more readily accessible group. The aim of this study was to evaluate the psychometric properties and structure of a common measure of OCD.

Obsessive-Compulsive Disorder is diagnostically characterised by the presence of obsessions such as pathological fear of harm associated with germs and contamination or violation of rules of symmetry, and of compulsive behaviours and mental acts that may include excessive washing or touching, checking and ordering (e.g., APA, 2000), and affects up to 3% of the population (Weissman, Bland, Canino, Greenwald, Hwu, & Lee, 1994). At the same time, OCD-like characteristics are also relatively common among the general population, meaning that individuals may display some (but not all) potentially debilitating symptoms characteristic of OCD without meeting the full criteria for a diagnosis of clinical OCD (see Gibbs, 1996, for a detailed review of research on OCD in nonclinical populations). Gibbs (1996) noted that a significant minority of the general population might be considered to display subclinical OCD, and that people who do so tend also to show elevated levels of anxiety and depression relative to normal controls (but lower than clinical OCD controls), as well as using qualitatively similar strategies for dealing with obsessive-compulsive symptoms. At the same time, an important difference in the symptom profile was the relative rarity of co-occurring obsessive and compulsive symptoms in comparison with samples with a diagnosis of clinical OCD (Gibbs, 1996). It has been argued that being able to identify such examples of subclinical OCD would allow early identification of those who might go on to develop full-blown OCD, and at the very least to treat symptoms even in those who do not (e.g., Morris, Morris, Blashfield, Rankupalli, Bradley, & Goodman, 1996/97; Zucker, Craske, Blackmore, & Nitz, 2006).

The Obsessive-Compulsive Inventory-Revised (OCI-R; Foa et al., 2002) is an 18-item self-report measure designed to assess six subtypes of OCD; Checking, Washing, Hoarding, Obsessing, Cleaning, and Neutralising. The OCI-R was developed from the original 42-item Obsessive-Compulsive Inventory (OCI; Foa, Kozak, Salkovskis, Coles, & Amir, 1998) by way of exploratory factor analysis (Foa et al., 2002). The OCI-R improves on the OCI by eliminating the seemingly redundant 'frequency scale', and discarding 24 items. Not only does this make the questionnaire significantly shorter, but cross loadings on multiple factors are eliminated, and an equal number of items in each subscale (three) simplifies scoring.

Foa et al. (2002) found good to excellent internal reliability ([alpha] > 0.70) among a non-anxious control group (NAC) of 74 undergraduate psychology students for both the total OCI-R score and all subscales but Neutralising ([alpha] = 0.34). In a more comprehensive study of subclinical OCI-R psychometrics, the same group (Hajcak, Huppert, Simons, & Foe, 2004) also found good to excellent internal reliability, again for all subscales except for Neutralising ([alpha] = 0.61) in a majority Caucasian college sample of 395 students. Test-retest reliability and convergent/divergent validity were found to be good to excellent. Hajcak et el. also reported evidence for the six-factor model of the OCI-R using confirmatory factor analysis. The OCI-R has been further validated in both clinical and subclinical populations ...