Participants

A fresh sample of 73 people with drug-resistant auditory hallucinations participated in the study: 41 men and 32 women, with a mean age of 40 years (s.d.=10.91); all had heard auditory hallucinations for at least 2 years. All had hospital diagnoses of schizophrenia, schizoaffective disorder or psychotic depression. Approximately 60% of the participants were referred for a psychological assessment for cognitive behaviour therapy, and completed the BAVQ—R as part of this assessment; all the other participants had volunteered to complete the measure and were then asked if they wished to be referred for a psychological assessment of suitability for cognitive behaviour therapy.

Measures

Statistical analysis

For each scale of the BAVQ—R the α coefficient was calculated, thus giving a measure of reliability drawn from a single administration of the measure. We present descriptive statistical analysis of the distribution of scores on all sub-scales (including skewness), and endorsement of each omnipotence item. Pearson correlations are calculated for relationships among the different BAVQ—R sub-scales, and between these and the HADS. All analyses were carried out using SPSS 8 for Windows.

Descriptive statistics

In the present study participants used the full range of scores for the sub-scales measuring malevolence, resistance and engagement; no one scored above 16 on the benevolence sub-scale, and no one scored zero on the omnipotence sub-scale. Differences were seen in the distributions of scores, and these differences replicated our previous findings (Reference Chadwick and BirchwoodChadwick & Birchwood, 1995). Skewness statistics for the different sub-scales were as follows: malevolence ‒0.452, resistance ‒1.231, engagement 1.274, benevolence 1.113, and omnipotence ‒0.180. Sub-scale means (with standard deviations) and medians were as follows: malevolence, mean 10.1 (s.d.=5.5), median 12; benevolence, mean 3.7 (s.d.=4.7), median 1; omnipotence, mean 11.1 (s.d.=4.5), median 11; engagement, mean 5 (s.d.=5.7), median 2; resistance, mean 19.4 (s.d.=6.5), median 21. (Both means and medians are given, because not all sub-scales showed a normal distribution.)

Reliability and validity of the BAVQ-R

For each sub-scale of the BAVQ—R the α coefficient was calculated, thus giving a measure of reliability drawn from a single measure. Cronbach's α scores for each sub-scale, including the new omnipotence sub-scale, were uniformly high. Furthermore, comparison with data from our previous study (Reference Chadwick and BirchwoodChadwick & Birchwood, 1995) showed that for malevolence, benevolence and resistance, Cronbach's α scores were all higher on the BAVQ—R. For engagement, the Cronbach's α scores were identical. It therefore appears that the BAVQ—R continues to measure clear and stable aspects of individuals' relationships with their auditory hallucinations.

Construct validity

The correlations between the different subscales of the BAVQ—R were examined. As in previous studies (Chadwick & Birchwood, Reference Chadwick and Birchwood1994, Reference Chadwick and Birchwood1995; Reference Birchwood and ChadwickBirchwood & Chadwick, 1997), we found a strong relationship between malevolence and resistance (r=0.68, d.f.=69, P<0.01) and benevolence and engagement (r=0.80, d.f.=69, P<0.01), with all other correlations between these sub-scales being strongly negative.

Omnipotence scale

The descriptive statistics that follow take a conservative view of endorsement of an item, as ‘Agree slightly’ or ‘Agree strongly’. The percentage level of endorsement of each item on the omnipotence sub-scale is: ‘My voice is very powerful’ (86%), ‘My voice seems to know everything about me’ (79%), ‘I cannot control my voice’ (75%), ‘My voice rules my life’ (63%), ‘My voice makes me do things I really don’t want to do' (47%), and ‘My voice will harm or kill me if I do disobey or resist it’ (38%). In relation to the item—total correlations (Table 1), the score for one item (‘I cannot control my voices’) is a little low (0.26). However, this item is retained in the measure because it is of critical importance clinically, and is one of the most highly endorsed on the entire measure (mean score 2.2).

Table 1 Psychometric properties of the revised Beliefs About Voices Questionnaire (BAVQ-R) (n=71)

	Malevolence	Benevolence	Omnipotence	Resistance	Engagement
No. of items	6	6	6	9	8
Range of scores	0-18	0-18	0-18	0-27	0-24
Range of item-item correlations	0.29-0.69	0.35-0.67	0.07-0.68	0.16-0.75	0.2-0.63
Range of item-total correlations	0.47-0.83	0.55-0.78	0.26-0.6	0.42-0.69	0.51-0.74
Cronbach's α	0.84	0.88	0.74	0.85	0.87

A two-tailed examination was made of the new omnipotence sub-scale in relation to the other sub-scales of the BAVQ. The relationship between omnipotence and malevolence was found to be strongly positive (r=0.70, d.f.=69, P<0.01), as was the relationship between omnipotence and resistance (r=0.50, d.f.=69, P<0.01). The sub-scales of omnipotence and engagement had a negative relationship (r=-0.26, d.f.=69, P<0.05). There was no significant relationship between the omnipotence and benevolence sub-scales.

Correlations between BAVQ-R and HADS scores

Fifty-eight participants completed the HADS. Table 2 shows mean scores on the HADS, and the number of participants falling into the different ranges of severity. Previous findings (Reference Chadwick, Birchwood, Haddock and SladeChadwick & Birchwood, 1996) had led us to expect a relationship between malevolence and depressive symptoms (hereafter called simply ‘depression’). This was found in the present study (r=0.37, d.f.=56, P<0.01). There was also a relationship between depression and both omnipotence (r=0.44, d.f.=56, P<0.01) and resistance (r=0.32, d.f.=56, P<0.05). Depression was negatively associated with engagement (r=-0.42, d.f.=56, P<0.01).

Table 2 Scores on the Hospital Anxiety and Depression Scale (HADS) (58 participants)

	Anxiety	Depression
Mean score (s.d.)	12.97 (5)	10.35 (4.67)
Normal range, % (n)	16 (9)	33 (19)
Mild range, % (n)	17 (10)	19 (11)
Moderate range, % (n)	26 (15)	27 (16)
Severe range, % (n)	41 (24)	21 (12)

Anxious symptomatology (hereafter called simply ‘anxiety’) was related to malevolence (r=0.30, d.f.=56, P<0.05), resistance (r=0.40, d.f.=56, P<0.01) and omnipotence (r=0.33, d.f.=56, P<0.05). There was a negative relationship between anxiety and engagement (r=0.36, d.f.=56, P<0.05). Scores on the HADS showed a significant relationship between anxiety and depression (r=0.59, d.f.=56, P<0.01).

Importance of omnipotence

The central importance of people's perceptions of auditory hallucinations as very powerful was observed first by Bauer (Reference Bauer and Keup1970). In a seminal paper, he reported how auditory hallucinations can be imbued with a “terrifying and compelling quality” and how individuals can feel “caught in a voice's power” (p. 169). The new omnipotence sub-scale may be thought of as an attempt to operationalise this quality. Descriptive data confirm that omnipotence is a vital part of an analysis of these participants' relationships with their auditory hallucinations.

Omnipotence has the highest mean score and lowest standard deviation (11.1, s.d.=4.5) of the three sub-scales which measure beliefs. Indeed, so important do we judge omnipotence to be that we have developed and evaluated a group-based cognitive behaviour therapy which principally targets beliefs about omnipotence (Reference Chadwick, Sambrooke and RaschChadwick et al, 2000).

Defining the concept of omnipotence

The present study helps to define our concept of omnipotence, which has hitherto been used as loosely equivalent to that of ‘power’. Clearly, perceiving an auditory hallucination to be very powerful remains an important attribute of omnipotence. In the present study 86% of the sample agreed with the statement ‘My voice is very powerful’. Yet 75% also endorsed the item ‘I cannot control my voices’, and 79% the item ‘My voices seem to know everything about me’. This makes it clear that the concept of omnipotence implies more than mere power. In future, it may therefore be helpful to refer to power or powerfulness as one specific aspect of omnipotence. The word ‘omnipotence’ would then be reserved for describing the broader concept, as measured by the BAVQ—R sub-scale.

Does a predominant profile emerge for clinical cases of this sort?

Data from this and other studies suggest that a ‘profile’ may be emerging which would describe the majority of patients using psychiatric services who experience auditory hallucinations. First, as in the present study, participants almost invariably perceive their dominant hallucinations as a considerable problem and source of distress, notwithstanding the fact that particular aspects of their relationships with auditory hallucinations may be positive. Second, omnipotence scores are very high (cf. Reference Close and GaretyClose & Garety, 1998). Third, the dominant auditory hallucination is predominantly perceived as malevolent, evoking behavioural resistance and negative affect. Fourth, the person is likely to experience at least a moderate level of anxiety symptoms and mild or moderate depressive symptoms. Yet it is important to note that for many people the experience of auditory hallucinations is a positive one; in a community sample (n=173) of patients and non-patients, Romme & Escher (Reference Romme and Escher1989) found that 15% of participants perceived their auditory hallucinations as positive.

CLINICAL IMPLICATIONS

• ▪ Participants continue to find the measure acceptable and easy to complete.

• ▪ The BAVQ—R is a useful assessment and outcome measure.

• ▪ The measure gives a quick, reliable profile of a person's relationship with an auditory hallucination, which is useful information for psychological therapy.

LIMITATIONS

• ▪ The BAVQ—R assesses relationship with an individual auditory hallucination.

• ▪ The BAVQ—R does not assess form or content.

• ▪ The cognitive model is of maintenance, not origin, of auditory hallucinations.