The Oxford–Liverpool Inventory of Feelings and Experiences short version: Further validation

Highlights

• The sO-LIFE scores showed adequate psychometric properties.

• Three- and four-factor models of schizotypy displayed better goodness-of-fit indices.

• Multi-group ESEM showed that the three-factor model had partial strong measurement invariance across gender.

• The sO-LIFE scores were associated with schizotypal personality traits and hedonic capacity.

Abstract

The main goal of the present study was to examine the dimensional structure and gather new sources of validity evidence of the Oxford–Liverpool Inventory of Feelings and Experiences short version (sO-LIFE) in a large sample of young adults. The sample was made up of 1002 college students (M = 21.11 years; SD = 3.92). The study of the internal structure, using confirmatory factor analysis, revealed that both three and four-factor solutions fitted well to the data. Furthermore, new measurement models, such us Exploratory Structural Equation Modelling (ESEM), showed that the hypothetical three-factor model displayed better goodness-of-fit indices than the other competing models tested. Multi-group ESEM showed that the three-factor model had partially strong measurement invariance across gender. The reliability of the scores ranged from 0.78 to 0.87. The sO-LIFE scores showed good convergent and discriminant validity with other measures of schizotypal personality traits and hedonic capacity. These results provide new information about the factor structure of schizotypy in non-clinical samples and support the use of sO-LIFE as a measure of schizotypy in nonclinical samples.

Introduction

Schizotypy is a latent personality organization that harbours the liability for psychosis in general, and schizophrenia in particular (Lenzenweger, 2010, Meehl, 1990). This is a complex construct that captures the expression of psychosis symptoms and impairment from non-clinical and subclinical levels to full-blown psychosis (Kwapil & Barrantes-Vidal, 2015). Empirical evidence indicates that individuals with high scores on schizotypy measures are at heightened risk for the later development of psychosis (Debbané et al., 2015) and show similar deficit to those found in patient with schizophrenia and individuals at high risk for psychosis (Cohen et al., 2015, Ettinger et al., 2014).

Several measuring instruments have been developed with the aim of detecting the latent vulnerability to psychosis of individuals at-risk for psychosis (Mason, 2015). The Oxford–Liverpool Inventory of Feelings and Experiences (O-LIFE) (Mason, Claridge, & Jackson, 1995), or its short version (sO-LIFE) (Mason, Linney, & Claridge, 2005), are one of the most well-known measures. The O-LIFE is a tool with 104 items (Yes/No response format) and four subscales empirically derived called: Positive Schizotypy (i.e., Unusual perceptual experiences), Cognitive Disorganisation, Introvertive Anhedonia, and Impulsive Nonconformity. Specifically, its psychometric properties have been analysed previously (Burch et al., 1998, Mason, 1995, Mason, 2015, Mason and Claridge, 2006). Nevertheless, according to Mason et al. (2005), the full scales are arguably cumbersome and a shortened form is preferable particularly when used alongside other instruments. The shortened form aims to measure the same constructs reliably in an efficient manner as possible for use in large scale genetic or screening settings as well as traditional experimental research.

The sO-LIFE is a brief tool composed of 43 items covering the same four subscales. Currently, few studies have tested its psychometric properties (Cella et al., 2013, Lin et al., 2013, Sierro et al., in press). Moreover, from a clinical and psychometric point of view, several limitations have been found: a) lack of factorial consistency and validity of the Impulsive Nonconformity (Lin et al., 2013); b) different methods of estimation to analyse the underlying structure for dichotomous data (e.g., ML, WLSMV); and c) low levels of reliability of the scores (Cella et al., 2013). Previous factorial models tested, using Confirmatory Factor Analysis (CFA), have demonstrated that both three- (without Impulsive Nonconformity dimension) and four-factor models fitted well to the data; however, Sierro et al. (in press) comparing both factorial models found that the three-factor solution was superior in terms of goodness-of-fit indices than the four-factor solution.

As has been seen, the internal structure of the sO-LIFE has not been clearly established and there is need for further replication. Likewise, new measurement models, such as bifactor models (Reise, 2012) or Exploratory Structural Equation Modelling (ESEM) (Marsh, Morin, Parker, & Kaur, 2014), have yet to be explored in depth. These new measurement models should be conducted in order to advance in the underlying factor structure of this measure of schizotypy as well as in personality assessment. For instance, bifactor models are useful when researchers typically write self-report items to assess a single construct (e.g., schizotypy) and they also recognize that the facets of the construct are diverse and substantively complex (e.g., four schizotypy dimensions). ESEM approach allow us to test less restrictive measurement models than those used in the traditional CFA models, e.g., where all cross-loadings constrained to zero. ESEM relaxes this restriction and factor loadings in all factors are estimated for each item, obtaining parameter estimates, standard errors, and goodness-of-fit indices usually associated with CFA. Consequently, to validly represent the construct, new measurement models need to be tested in this field.

Within this research context, the main goal of the present study was to analyse the dimensional structure and gather new sources of validity evidence of the sO-LIFE scores in a large sample of college students. Deriving from this general goal are the following specific objectives: a) to examine the internal structure of the sO-LIFE scores using CFAs and ESEM approach; b) to test the measurement invariance of the sO-LIFE scores across gender; c) to estimate the internal consistency with Ordinal alpha; and d) to analyse the relationship between sO-LIFE scores with schizotypal traits and hedonic capacity measures. We hypothesized that a four- and three-factor model would be more adequate than the other competing models. Moreover, it is possible that ESEM and bifactor models would show good fit to the data. In addition, we hypothesized that the factor structure of the measure would be equivalent across gender. Finally, it is also hypothesized that sO-LIFE scores would show adequate internal consistency values and associations with measures of schizotypal personality and hedonic capacity would be found.