My Voice, My Life: a measure based on the consumer model of recovery

Abstract

We note the Forum on “Consumer models of recovery: issues and perspectives” in the October 2012 issue of World Psychiatry, which comments on the paucity of robust, psychometrically sound measures of recovery. We wish to highlight such a measure, “My Voice, My Life”, formulated using a systematic psychometric process of scale development (1), as advocated by Bellack and Drapalski (2), while based on the consumer model of recovery and utilizing the consumer-led model of development promoted by Rose et al (3), as advocated by Callard (4). The process of development of the measure began with a deliberately over-inclusive preliminary version consisting of 127 items, based on 12 presumptive domains derived from the recovery literature and consumer consultation, which was piloted with 504 mental health consumers. The participant data set was randomly split into two discrete sets, one for the initial exploratory factor analysis and the other for the subsequent independent confirmatory factor analysis and reliability estimation. These analyses identified and confirmed (using the separate data sets) a robust factor structure, with 11 distinct and relatively independent factors (relationships; day-to-day life; culture; physical health; quality of life; mental health; recovery; hope and empowerment; spirituality; resources; and satisfaction with services) underlying one substantial principal construct (that we refer to as “consumer recovery”). The measure was then refined to 65 items, between three and ten items for each of the 11 domains, with uniformly high reliabilities (1). These 11 psychometrically discrete domains may be seen as a significant verification of the consumer-driven theory of recovery, based on, and informed by, first-hand experience. Such results provide empirical support for the theoretical validity of the consumer recovery construct in its own right, rather than as a derivative of a social cognitive model developed within an earlier construct of mental illness, as proposed by Bellack and Drapalski (2). At 65 items, the measure is longer than many routinely used “outcome” measures. However, if it is to adequately measure the 11 factors identified and confirmed in the factor analysis, this is perhaps inevitable. Maintaining domain coverage was considered crucial by our consumer reference group and it was this, as much as the psychometric issues, that determined our decision not to condense the measure at this stage. The fact that our process of scale development was consumer led ensured that matters of significance to consumers generally were prioritized. Development processes which commence with and/or insist on a small number of domains and items, such as the Maryland Assessment of Recovery in People with Serious Mental Illness (MARS, 2), run the risk of neglecting constructs and consequently being criticized for applying a reductionist form of science (5). Our empirical work suggests that the MARS may not be measuring the full range of recovery domains and/or is encapsulating multiple constructs within individual domains. If these measures are to influence services in a manner consistent with the consumer recovery paradigm, they must reflect all its distinct and independent domains. Failure to do this will distort how “recovery” services are developed, risking some key domains being ignored or at least undervalued.