Available online at www.sciencedirect.

com

Addictive Behaviors 33 (2008) 515 – 527

Metacognitive beliefs about alcohol use: Development and

validation of two self-report scales
Marcantonio M. Spada a,⁎, Adrian Wells b,c
a
School of Human and Life Sciences, Roehampton University, Whitelands College,
Holybourne Avenue, London SW15 4JD, UK
b
University of Manchester, UK
c
Norwegian University of Science and Technology, Norway

Abstract

The goal of this research was to develop clinical assessment tools of positive and negative metacognitive beliefs
about alcohol use. In Study 1 we constructed two scales and conducted preliminary factor analyses. Studies 2 and 3
investigated the predictive validity and temporal stability of the scales. Study 4 examined the factor structure,
predictive validity and classification accuracy of the scales in a clinical sample. The Positive Alcohol Metacognitions
Scale (PAMS) and the Negative Alcohol Metacognitions Scale (NAMS) were shown to possess good psychometric
properties, as well as predictive validity and classification accuracy, in both clinical and community populations. The
scales may aid future research into problem drinking and facilitate clinical assessment and case formulation.
© 2007 Elsevier Ltd. All rights reserved.

Keywords: Alcohol outcome expectancies; Alcohol use; Metacognition; Metacognitive beliefs about alcohol use; Negative
emotions; Problem drinking

1. Introduction

Metacognition refers to the psychological structures, beliefs, events and processes that are involved in
the control, modification and interpretation of thinking (Flavell, 1979; Moses & Baird, 1999; Wells,
2000). The great majority of theorists would agree in drawing a distinction between two basic aspects of
metacognition (Flavell, 1979; Wells, 2000; Yussen, 1985): metacognitive regulation and metacognitive

⁎ Corresponding author. Tel.: +44 20 8392 3559.

E-mail address: M.Spada@roehampton.ac.uk (M.M. Spada).

0306-4603/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.addbeh.2007.10.011
516 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527

knowledge (or metacognitive beliefs). Metacognitive regulation refers to a broad spectrum of executive
functions, such as monitoring, planning, checking, attention and detection of errors in performance
(Wells, 2000). Metacognitive knowledge refers to the information individuals hold about their internal
states and about coping strategies that impact on them (Wells, 2000). Examples of metacognitive
knowledge may include beliefs concerning the significance of particular types of thoughts (e.g. “Having
thought X means I am weak”) and emotions (e.g. “I need to control my anxiety at all times”), and beliefs
about cognitive competence (e.g. “I do not trust my problem-solving capabilities”). Examples of
information individuals hold about their own coping strategies that impact on internal states may include
both positive (“Ruminating will help me find a solution”) and negative (“My checking behaviour is
making me lose my mind”) beliefs. In the metacognitive conceptualization of psychological dysfunction
(Wells & Matthews, 1994; Wells, 2000) all the above constructs interact in maintaining maladaptive
behavior.
The Self-Regulatory Executive function (S-REF: Wells & Matthews, 1994, 1996) theory was the first
to conceptualise the role of metacognition in the etiology and maintenance of psychological disturbance.
In this theory Wells and Matthews (1994, 1996) argue that a common style of thinking across
psychological disorders leads to dysfunction. They propose that psychological disturbance is maintained
by a combination of perseverative thinking styles, maladaptive attentional routines, and dysfunctional
behaviors. This array of factors constitutes a cognitive-attentional syndrome (CAS; Wells, 2000). The
CAS is derived from the individual's metacognitive knowledge (or metacognitive beliefs), which is
activated in problematic situations and drivers coping (such as alcohol use) (Wells & Matthews, 1994,
1996; Wells, 2000).
The S-REF theory has led to the development of disorder-specific models of depression (Papageorgiou
& Wells, 2003), generalised anxiety disorder (Wells & Matthews, 1994; Wells, 2000), obsessive–
compulsive disorder (Wells & Matthews, 1994; Wells, 2000), post-traumatic stress disorder (Wells, 2000)
and social phobia (Clark & Wells, 1995). Metacognitive beliefs have been found to be positively associated
with depression (Papageorgiou & Wells, 2003), hypochondriasis (Bouman & Meijer, 1999), obsessive–
compulsive symptoms (Emmelkamp & Aardema, 1999; Hermans, Martens, De Cort, Pieters, & Eelen,
2003; Myers & Wells, 2005; Wells & Papageorgiou, 1998), pathological procrastination (Spada, Hiou, &
Nikčević, 2006), pathological worry (Wells & Papageorgiou, 1998), post-traumatic stress disorder
(Roussis & Wells, 2006), predisposition to auditory hallucinations (Baker & Morrison, 1998; Morrison,
Wells, & Nothard, 2000), psychosis (Morrison, French, & Wells, 2007), smoking dependence (Spada,
Nikčević, Moneta, & Wells, 2007) and test-anxiety (Matthews, Hillyard, & Campbell, 1999; Spada,
Nikčević, Moneta, & Ireson, 2006).
Whilst the S-REF theory was initially intended to account for emotional disorders, recent work has
examined its application in predicting alcohol use. In a series of preliminary investigations (Spada &
Wells, 2005; Spada, Zandvoort, & Wells, 2007) evidence was found of: (1) a positive association between
a general dimension of metacognition (beliefs about the need to control thoughts) and alcohol use that is
independent of negative emotions; and (2) an independent contribution (over negative emotions) of
general dimensions of metacognition (beliefs about the need to control thoughts and low cognitive
confidence) towards category membership as a problem drinker.
Further research undertaken by Spada and Wells (2006) has identified the existence of specific positive
and negative metacognitive beliefs about alcohol use in problem drinkers. Positive metacognitive beliefs
about alcohol use can be conceptualised as a specific form of outcome expectancy relating to the use of
alcohol as a means of controlling cognition and emotion. From a metacognitive standpoint such beliefs
 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527 517

are thought to play a central role in motivating individuals to engage in alcohol use as a means of
cognitive-emotional regulation (Spada & Wells, 2006). Examples of positive metacognitive beliefs about
alcohol use may include: “Drinking makes me think more clearly” (problem-solving), “Drinking helps me
to control my thoughts” (thought control), “Drinking helps me focus my mind” (attention regulation),
“Drinking reduces my self-consciousness” (self-image regulation), “Drinking reduces my anxious
feelings” (emotion regulation). Negative metacognitive beliefs about alcohol use concern the perception
of lack of executive control over alcohol use (e.g. “My drinking persists no matter how I try to control it”),
and the evaluation of the negative impact of alcohol use on cognitive functioning (e.g. “Drinking will
damage my mind”). From a metacognitive standpoint such beliefs are thought to play a crucial role in the
perpetuation of alcohol use by becoming activated during and following a drinking episode, and
triggering negative emotional states that compel a person to drink more (Spada & Wells, 2006).
Positive metacognitive beliefs about alcohol use share similarities, but also fundamental differences,
with a crucial cognitive variable involved in the initiation and maintenance of alcohol use: positive
alcohol outcome expectancies. Positive alcohol outcome expectancies refer to the drinker's perception of
the positive outcomes of drinking, and have been shown, by enlarge, to be associated to alcohol use
(Brown, Christiansen, & Goldman, 1987; Christiansen, Smith, Roehling, & Goldman, 1989; Goldman,
Del Boca & Darkes, 1999; Leigh, 1989; Maisto, Connors, & Sachs, 1981). According to Spada and
colleagues (Spada, Moneta, & Wells, 2007) the key similarity between positive metacognitive beliefs
about alcohol use and positive alcohol outcome expectancies is that both constructs capture motivations
for alcohol use. As such there is a degree of overlap between positive metacognitive beliefs about alcohol
use pertaining to emotion regulation and positive alcohol outcome expectancies pertaining to tension
reduction and the modulation of negative affect. However, these constructs are not identical as correlation
coefficients between them of around .50 attest to (Spada, et al., 2007). A key difference between positive
metacognitive beliefs about alcohol use and positive alcohol outcome expectancies is that items
pertaining to the former construct also tap into the effects of alcohol use on cognition (problem-solving,
thought control, attention regulation, and self-image regulation). This particular domain is largely
overlooked by current measures of positive alcohol outcome expectancies.
Negative metacognitive beliefs about alcohol use concerning the perception of lack of executive
control over alcohol use assess cognitive confidence in regulating alcohol use and can thus be
conceptualized as a specific form of cognitive self-efficacy belief. Negative metacognitive beliefs
concerning the impact of alcohol use on cognitive functioning are evaluations of the cognitive costs of
drinking. These beliefs are related to negative alcohol outcome expectancies (Christiansen et al., 1989;
Jones, Corbin, & Fromme, 2001; Stacy, Widaman, & Marlatt, 1990) which assesses an individual's
estimation that a given behavior will lead to specific negative outcomes (Bandura, 1997). However, they
extend current measures of negative outcome expectancies by focussing specifically on alcohol's
detrimental effect on cognitive functioning.
The distinctions between alcohol outcome expectancies and metacognitive beliefs about alcohol use
are important because according to the metacognitive theory of psychopathology the key markers of
dysfunction are beliefs pertaining to the metacognitive rather than cognitive domain (Wells, 2000). In
support of the importance of differentiating between alcohol outcome expectancies and metacognitive
beliefs about alcohol use a recent study by Spada and colleagues (Spada et al., 2007) employing the
Positive Alcohol Metacognitions Scale (PAMS) and the Negative Alcohol Metacognitions Scale (NAMS)
in a community sample of 355 individuals revealed that three of the four facets of metacognitive beliefs
about alcohol use were an independent contributor to drinking behavior over and above alcohol outcome
518 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527

expectancies. Furthermore, when controlling for metacognitive beliefs, only one sub-facet of negative
alcohol outcome expectancies (Negative Social Performance) explained additional variance in drinking
behavior.
In view of the recent findings indicating a possible role of metacognitive beliefs in problem drinking
and their utility in extending the alcohol outcome expectancy construct, the purpose of this research is to
report four studies on the development of PAMS and NAMS.

2. Study 1: Construction and preliminary factor analyses of PAMS and NAMS

2.1. Method

2.1.1. Participants
A community sample of 261 individuals (121 females and 140 males) agreed to take part in the study
which was approved by an ethics committee at a London University. For purposes of inclusion in the
study the participants were required to speak English, be at least 18 years of age and have consumed
alcohol over the last week. The mean age for the total sample, which consisted primarily of Caucasian
university students, was 22.1 years (SD = 3.5 years) and the age range was 18–49 years.

2.1.2. Measures
In order to achieve adequate face validity, a pool of items used to construct PAMS and NAMS was
derived from the data obtained in an earlier semi-structured interview study (Spada & Wells, 2006) and
transcripts of therapy sessions.
In total, 15 items were selected for PAMS and 6 items for NAMS. Responses to each item were
required on a 4-point rating scale as follows: 1 (do not agree), 2 (agree slightly), 3 (agree moderately) and
4 (agree very much). Examples of positive metacognitive beliefs about alcohol use included: “Drinking
makes my negative thoughts more bearable” and “Drinking helps me to control my thoughts”. Examples
of negative metacognitive beliefs about alcohol use included: “If I cannot control my drinking I will cease
to function” and “Drinking will make me lose control”.

2.1.3. Procedure
Participants were informed that the purpose of the study was to examine beliefs related to alcohol use.
After giving informed consent participants were instructed to provide demographic details and complete
the questionnaires. No opportunity was given for response correction. All participants were debriefed
following completion of the questionnaires.

2.2. Results

A principal components method of factor extraction was performed on data from PAMS and a two-
factor solution (eigenvalues of 6.15 and 1.72 respectively) based on a scree test was subjected to oblique
rotation. Oblique rotation was selected because we expected factors to be correlated. This type of rotation
produces more accurate estimates of the true factors and a better simple structure than orthogonal rotation
(Fabrigar, Wegener, MacCallum, & Strahan, 1999).
The two factors identified (accounting for 52.5% of variance) reflected the following domains: (1)
positive metacognitive beliefs about emotional self-regulation; and (2) positive metacognitive beliefs
 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527 519

Table 1
Factor loadings and communalities for individual items of PAMS
Factor 1 Factor 2 Communality
Factor 1: positive metacognitive beliefs about emotional self-regulation
(1) Drinking makes me more affectionate .61 .26 .38
(2) Drinking makes me more confident .81 .24 .68
(3) Drinking makes me feel more relaxed .71 .23 .50
(4) Drinking reduces my anxious feelings .75 .54 .62
(5) Drinking makes me more sociable .79 .27 .63
(6) Drinking reduces my self-consciousness .74 .38 .55
(7) Drinking makes me feel happy .60 .28 .35
(8) Drinking helps me fit in socially .73 .45 .55

Factor 2: positive metacognitive beliefs about cognitive self-regulation

(1) Drinking makes me think more clearly .34 .80 .64
(2) Drinking helps me to control my thoughts .32 .83 .70
(3) Drinking makes my negative thoughts more bearable .51 .74 .59
(4) Drinking helps me focus my mind .30 .80 .64

about cognitive self-regulation. The item pool was revised as some items loaded on more than one factor.
These redundant items, and those which loaded less than a conservative .40 on their factor, were discarded.
The revised scale consisted of 12 items. Individual items of PAMS and their factor loadings are displayed in
Table 1.
A principal components method of factor extraction was also performed on data from NAMS and a
two-factor solution (eigenvalues of 2.73 and 1.12 respectively) based on a scree test was subjected to
oblique rotation. The two factors identified (accounting for 64.4% of variance) reflected the following
domains: (1) negative metacognitive beliefs relating to lack of executive control over alcohol use; and (2)
negative metacognitive beliefs relating to the negative impact of alcohol use on cognitive functioning.
The item pool was not revised as none of the items loaded on more than one factor and all items
loaded more than .40 on their factor. Individual items of NAMS and their factor loadings are displayed
in Table 2.

Table 2
Factor loadings and communalities for individual items of NAMS
Factor 1 Factor 2 Communality
Factor 1: negative metacognitive beliefs about uncontrollability
(1) I have no control over my drinking .76 .16 .62
(2) My drinking persists no matter how I try to control it .80 .41 .66
(3) Drinking controls my life .78 .33 .61

Factor 2: negative metacognitive beliefs about cognitive harm

(1) If I cannot control my drinking I will cease to exist .83 .16 .71
(2) Drinking will damage my mind .81 .34 .66
(3) Drinking will make me lose control .76 .47 .63
520 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527

The internal consistencies (homogeneity) of PAMS and NAMS were determined by computing the
Cronbach coefficient alpha. This coefficient was .88 for PAMS and .74 for NAMS. Alpha coefficients were
also calculated for each factor. These were .81 for PAMS Factor 1, .87 for PAMS Factor 2, .68 for NAMS
Factor 1 and .72 for NAMS Factor 2. Communalities for both measures are shown in Tables 1 and 2. Inter-
item correlations ranged from .17 to .63 for PAMS and from .21 to .51 for NAMS showing that redundancy
of items was not problematic.
The final names assigned to each factor were determined by its item content. For PAMS, Factor 1 concerned
positive metacognitive beliefs about emotional self-regulation. Factor 2 concerned positive metacognitive
beliefs about cognitive self-regulation. For NAMS, Factor 1 consisted of negative metacognitive beliefs about
uncontrollability. Factor 2 consisted of negative metacognitive beliefs about cognitive harm.

3. Study 2: Preliminary examination of the predictive validity of PAMS and NAMS

3.1. Method

3.1.1. Participants
A community sample of 138 individuals (77 females and 61 males) agreed to take part in the study
which was approved by an ethics committee at a London University. For purposes of inclusion in the
study the participants were required to speak English, be at least 18 years of age and have consumed
alcohol over the last month. The mean age for the total sample, which consisted primarily of Caucasian
professionals, was 30.0 years (SD = 9.3 years) and the age range was 18 to 60 years.

3.1.2. Measures
To determine the predictive validity of the final versions of PAMS and NAMS, two measures of
drinking behavior were selected and administered alongside them:
The Quantity Frequency Scale (QFS; Cahalan, Cisin, & Crossley, 1969). QFS is a measure of alcohol
consumption levels, with items assessing the dimensions of quantity and frequency of alcohol beverages
consumed over a period of 30 days. This scale consists of three questions (“have you been drinking any
beer/wine/spirits over the last 30 days?”; “about how often do you consume beer/wine/spirits?”; and
“about how much beer/wine/spirits did you drink on a typical day when you drink beer/wine/spirits?”).
These are repeated for each of the major alcohol beverage categories (beer, wine and distilled spirits). The
total scores from the different alcohol beverage categories are then added together and an estimated daily
(or weekly) level of alcohol consumption can be computed. This instrument has been extensively used
and possesses good validity and reliability (Hester & Miller, 1995).
The Alcohol Use Disorders Identification Test (AUDIT; Babor, de la Fuente, Saunders, & Grant, 1992).
AUDIT was developed as a screening tool by the World Health Organisation (WHO) for early
identification of problem drinkers. AUDIT consists of 10 questions regarding recent alcohol consumption,
alcohol dependence symptoms and alcohol-related problems. Respondents are asked to choose one of 5
statements (per question) that most applies to their use of alcohol beverages over the past year. Responses
are scored from 0 to 4 in the direction of problem drinking. The summary score for the total AUDIT
ranges from 0, indicating no presence of problem drinking behavior, to 40 indicating marked levels of
problem drinking behavior and alcohol dependence. The threshold for indicating possible problem
drinking pathology is a score of 8. This instrument has been extensively used and possesses good validity
and reliability (Hester & Miller, 1995).
 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527 521

Table 3
Means, standard deviations, ranges and intercorrelations of variables
Mean SD Range 2. 3. 4. 5. 6.
1. QFSa 22.6 19.3 0–80 .82⁎⁎ .27⁎⁎ .26⁎⁎ .47⁎⁎ .14
2. AUDITb 7.6 5.2 0–30 – .39⁎⁎ .23⁎⁎ .47⁎⁎ .14
3. PAMS — Beliefs about emotional self-regulationc 18.6 5.7 8–31 – – .51⁎⁎ .14 .20⁎
4. PAMS — Beliefs about cognitive self-regulationd 5.2 1.5 4–13 – – – .22⁎ .17⁎
5. NAMS — Beliefs about uncontrollabilitye 3.6 1.4 3–11 – – – – .24⁎⁎
6. NAMS — Beliefs about cognitive harmf 5.2 2.5 3–12 – – – – –
Note: n = 138. aUnits per week. bRange of scale: 0–42. cRange of scale: 8–32. dRange of scale: 4–16. eRange of scale: 3–12.
f
Range of scale: 3–12.
⁎p b 0.05; ⁎⁎p b 0.01.

3.1.3. Procedure
Participants were informed that the purpose of the study was to examine beliefs related to alcohol use.
After giving informed consent participants were instructed to provide demographic details and complete
the questionnaires. No opportunity was given for response correction. All participants were debriefed
following completion of the questionnaires.

3.2. Results

Descriptive statistics and intercorrelations for all measures are shown in Table 3. An inspection of skewness
coefficients showed that several measures were asymmetrically distributed. Two-tailed Spearman correlations
revealed that both factors 1 and 2 of PAMS correlated positively and significantly with alcohol use and problem
drinking. Factor 1 of NAMS correlated positively and significantly with alcohol use and problem drinking.

4. Study 3: Test–retest reliability of PAMS and NAMS

4.1. Method

4.1.1. Participants
A community sample of 53 individuals (33 females and 20 males) agreed to take part in the study
which was approved by an ethics committee at a London University. For purposes of inclusion in the
study the participants were required to speak English, be at least 18 years of age and have consumed
alcohol over the last week. The mean age for the total sample, which consisted primarily of Caucasian
professionals, was 34.1 years (SD = 13.5 years) and the age range was 18 to 61 years.

4.1.2. Measures
In order to assess the temporal stability of PAMS and NAMS the final versions of these scales were used.

4.1.3. Procedure
Participants were informed that the purpose of the study was to examine beliefs related to alcohol use.
After giving informed consent participants were instructed to provide demographic details and complete
the questionnaires on two occasions (the time interval between testing and retesting was approximately
522 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527

8 weeks). No opportunity was given for response correction. All participants were debriefed following
completion of the questionnaires.

4.2. Results

An inspection of skewness coefficients did not support the use of parametric statistics for both PAMS
and NAMS. Mean PAMS scores at testing and retesting were 18.8 (SD = 5.7) and 18.5 (SD = 5.8)
respectively for Factor 1, and 5.0 (SD = 1.5) and 5.2 (SD = 1.6) respectively for Factor 2. A Wilcoxon
Signed Ranks test indicated that the mean scores for both factors did not change over the 8-week interval
(Factor 1: z = −.46, p = .64; Factor 2: z = −.84, p = .40). Spearman coefficients (Factor 1: rho = .75,
p b .0005; Factor 2: rho = .65, p b .0005) showed acceptable test–retest reliability for both factors,
suggesting that they possesses relatively stable characteristics.
Mean NAMS scores at testing and retesting were 3.3 (SD = .7) and 3.2 (SD = .5) respectively for Factor
1, and 5.9 (SD = 2.7) and 5.6 (SD = 2.6) respectively for Factor 2. A Wilcoxon Signed Ranks test indicated
that the mean scores for both factors did not change over the 8-week interval (Factor 1: z = −.74, p = .46;
Factor 2: z = −1.00, p = .32). Spearman coefficients (Factor 1: rho = .42, p = .001; Factor 2: rho = .68,
p b .0005) showed acceptable test–retest reliability for Factor 2 suggesting that it possesses relatively
stable characteristics, but poor test–retest reliability for Factor 1 suggesting that the measure is not stable
over time.

5. Study 4: Preliminary factor analyses, predictive validity and classification accuracy of PAMS
and NAMS in a clinical sample

5.1. Method

5.1.1. Participants
The clinical sample consisted of 80 problem drinkers (15 females and 65 males) who were referred or
self-referred for treatment to a variety of alcohol services in the south of the United Kingdom. Permission
for running the study was granted by an ethics committee at a London University. All patients identified
problem drinking as their primary problem for which they were seeking psychological treatment. The
mean age for the total sample, which consisted primarily of Caucasian professionals, was 32.2 years
(SD = 12.4 years) and the age range was 18–62 years. The mean alcohol consumption level for the total
sample was 101 units per week (SD = 32 units) and the mean problem drinking score was 22.9 (SD = 9.5).
A community sample of 83 individuals (18 females and 65 males), consisting mainly of Caucasian
professionals, was also recruited as a comparison group. For purposes of inclusion in the study the
participants were required to speak English, be at least 18 years of age and have consumed alcohol over
the last week. Their total mean age was 31.7 years (SD = 12.8 years) and the age range was 18 to 65 years.

5.1.2. Measures
To determine the predictive validity of PAMS and NAMS, the following measures were selected and
administered alongside them:
Hospital Anxiety and Depression Scale (HADS; Zigmond & Snaith, 1983). This scale consists of 14
items, 7 assessing anxiety and 7 assessing depression. Higher scores represent higher levels of anxiety and
depression. HADS has been extensively used with both clinical and non-clinical samples. The scale
 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527 523

possesses good validity and reliability (Caci, Bayle, Mattei, Dossios, Philippe, & Boyer, 2003; Mykletun,
Stordal, & Dahl, 2001; Zigmond & Snaith, 1983).
The Quantity Frequency Scale (QFS; Cahalan et al., 1969). See Study 2 for full description.
The Alcohol Use Disorders Identification Test (AUDIT; Babor et al., 1992). See Study 2 for full
description.

5.1.3. Procedure
Participants were informed that the purpose of the study was to examine factors related to alcohol use.
After giving informed consent participants were instructed to provide demographic details and complete
the questionnaires. No opportunity was given for response correction. All participants were debriefed
following the completion of the questionnaires.

5.2. Results

5.2.1. Factor analysis

In order to determine the optimum number of factors and the item compositions of PAMS and NAMS
in the clinical sample, principal components factor analyses were preferred to confirmatory factor
analyses because of the limited sample size and the preliminary nature of this investigation (Kline, 1998).
A principal components method of factor extraction was performed on data from PAMS and a two-
factor solution (eigenvalues of 4.40 and 1.70 respectively) was suggested by the scree test. This was
subjected to oblique rotation. Two factors were identified (accounting for 50.9% of variance) that
appeared to have a similar item composition as that obtained in the first study. The only difference being
that the item “Drinking reduces my anxious feelings” moved from Factor 1 to Factor 2.
A principal components method of factor extraction was also performed on data from NAMS and a
two-factor solution (eigenvalues of 3.62 and 1.10 respectively) based on a scree test was subjected to
oblique rotation. The two factors identified (accounting for 78.8% of variance) had an identical item
composition as those obtained in the first exploratory study.
The internal consistencies (homogeneity) of PAMS and NAMS were determined by computing the
Cronbach coefficient alpha for each measure. This coefficient was .84 for PAMS and .87 for NAMS.
Alpha coefficients were also calculated for each factor. These were .77 for PAMS Factor 1, .81 for PAMS
Factor 2, .87 for NAMS Factor 1 and .83 for NAMS Factor 2. Inter-item correlations ranged from .15 to
.64 for PAMS and from .22 to .71 for NAMS showing that redundancy of items was not problematic.

5.2.2. Predictive validity

Descriptive statistics and intercorrelations for all measures from the clinical sample are shown in Table 4.
An inspection of skewness coefficients showed that several measures were asymmetrically distributed. Two-
tailed Spearman correlations showed that factors 1 and 2 of PAMS did not correlate significantly with alcohol
use. Both factors, however, were found to correlate positively and significantly with problem drinking.
Additionally both factors correlated positively and significantly with anxiety and depression. Factors 1 and 2
of NAMS correlated positively and significantly with alcohol use and problem drinking. Additionally both
factors correlated positively and significantly with anxiety and depression.
To evaluate whether PAMS and NAMS predicted problem drinking independently of negative
emotions, a series of multiple regression analyses were run in which anxiety and depression scores were
controlled and each of the PAMS and NAMS factors that were significant in the bivariate analysis were
524 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527

Table 4
Means, standard deviations, ranges and intercorrelations of clinical population
Mean SD Range 2. 3. 4. 5. 6. 7. 8.
1. QFSa 100.9 31.9 40–162 .34⁎⁎ .01 .19 −.02 .06 .24⁎ .24⁎
2. AUDITb 22.9 9.5 5–39 – .50⁎⁎ .65⁎⁎ .32⁎⁎ .39⁎⁎ .69⁎⁎ .59⁎⁎
3. HADS — Anxietyc 9.5 5.1 0–19 – – .74⁎⁎ .28⁎⁎ .36⁎⁎ .48⁎⁎ .47⁎⁎
4. HADS — Depressiond 6.6 5.3 0–21 – – – .25⁎⁎ .43⁎⁎ .50⁎⁎ .50⁎⁎
5. PAMS — Beliefs about emotional self–regulatione 22.0 5.8 11–32 – – – – .47⁎⁎ .14 .24⁎⁎
6. PAMS — Beliefs about cognitive self–regulationf 8.6 3.4 4–16 – – – – – .36⁎⁎ .32⁎⁎
7. NAMS — Beliefs about uncontrollabilityg 6.7 3.3 3–12 – – – – – .57⁎⁎
8. NAMS — Beliefs about cognitive harmh 7.5 3.0 3–12 – – – – – – –
Note: n = 80. aUnits per week. bRange of scale: 0–42. cRange of scale: 0–21. dRange of scale: 0–21. eRange of scale: 8–32.
f
Range of scale: 4–16. gRange of scale: 3–12. hRange of scale: 3–12.
⁎p b 0.05; ⁎⁎p b 0.01.

entered in turn. PAMS Factor 2 and NAMS Factor 2 were found to account for no significant variance in
addition to the variance explained by negative emotions. PAMS Factor 1 and NAMS Factor 1 were found
to account respectively for a significant 2.9% ( β = .18, t = 1.96, p = .05) and 18.4% ( β = .50, t = 5.80,
p b .0005) of variance in problem drinking, in addition to 40.1% variance accounted for by negative
emotions (anxiety β = .07, t = .49, p = .62; depression β = .58, t = 4.42, p b .0005).

5.2.3. Classification accuracy

Subsequently a binary logistic regression analysis was employed to determine the odds of being in one
category or another (in this study community sample or problem drinker). This is represented by the
exponential of the coefficient (Exp B), which indicates the change in the dependent variable given a one-
unit change in the predictor variable. Anxiety and depression were entered as control variables in the
initial step. These control variables were considered on an a priori basis of having a high probability of
being associated with the dependent variables. On the next step, the contribution of the hypothesized sets
of predictors was tested with PAMS factors 1 and 2, and NAMS factors 1 and 2 using forward (Wald)
selection. Negative emotions were found to be a significant predictor of problem drinking. The Chi-
square for improvement was 34.5 and significant ( p b 0.0005). On step two of the analysis the Chi-square
for improvement was 34.0 and significant ( p b 0.0005), meaning that both PAMS and NAMS factors
contribute significantly to the prediction of problem drinking caseness over and above negative emotions.

Table 5
Summary statistics for the logistic regression equation predicting problem drinking
B S.E. Wald df Sig. Exp(B)
HADS — Anxiety −.57 .08 .54 1 .46 .94
HADS — Depression −.67 .20 .41 1 .52 .93
PAMS — Beliefs about emotional self-regulation .02 .05 .15 1 .70 1.0
PAMS — Beliefs about cognitive self-regulation −.42 .25 7.81 1 .005 .65
NAMS — Beliefs about uncontrollability −.27 .24 4.12 1 .04 .76
NAMS — Beliefs about cognitive harm −.00 .20 .01 1 .94 .99
Constant 3.97 .94 18.04 1 .00 52.0
 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527 525

A look at the final equation in Table 5 shows that only two variables were significant independent
predictors of problem drinking: PAMS Factor 2 and NAMS Factor 1. The overall statistics for the final
equation were as follows: c2 = 68.7, df = 6, p b 0.0005, with 80.6% of cases correctly classified.

6. Discussion

Extrapolating from Wells and Matthews' (1994, 1996) metacognitive theory of emotional disorders,
and recent evidence supporting the existence of specific metacognitive beliefs about alcohol use in
problem drinkers (Spada & Wells, 2006), we conducted four studies aimed at developing and validating
two self-report scales of positive and negative metacognitive beliefs about alcohol use. Results from the
first, second and fourth study suggest that both scales are dimensional. A similar factor structure was
obtained in non-clinical and clinical samples. PAMS comprises items relating to two sets of metacognitive
beliefs that alcohol use is a helpful strategy when dealing with negative emotions and unwanted thoughts.
The scale possesses high internal consistency, and the magnitude of the inter-item correlations is
suggestive of a low to moderate overlap indicating low item redundancy. NAMS comprises items relating
to two sets of metacognitive beliefs that alcohol use is uncontrollable, and harmful to cognitive
functioning. The scale possesses high internal consistency, and the magnitude of the inter-item
correlations is suggestive of a low to moderate overlap indicating low item redundancy. Results from the
third study suggest that both PAMS factors and negative metacognitive beliefs about cognitive harm have
acceptable external reliability with negative metacognitive beliefs about uncontrollability possessing
unstable characteristics. The latter finding may be due to the need of identifying more items for this scale
or to the measure capturing what is fundamentally a situational rather than dispositional construct.
Relationships between both scales and symptom measures were also observed in the clinical sample.
Both PAMS factors are significantly and positively associated with problem drinking with positive
metacognitive beliefs about emotional self-regulation predicting problem drinking independently of
anxiety and depression. Both NAMS factors are significantly and positively associated with alcohol use
and problem drinking and both also predict problem drinking interpedently of anxiety and depression.
Finally, binary logistic regression analysis showed that positive beliefs about cognitive self-regulation and
negative beliefs about uncontrollability are independent predictors of problem drinking classification
when anxiety and depression is controlled. This finding lends further support to the role played by
metacognitive beliefs in predicting alcohol use, especially when considering the distinctiveness of
positive metacognitive beliefs about cognitive self-regulation as a metacognitive construct.
Taken together these results confirm the utility of both PAMS and NAMS. In terms of assessment, for
example, information could be gathered not only in relation to alcohol outcome expectancies, but also
associated metacognitive beliefs about alcohol use. With regards to treatment it is plausible to assume that
the modification of metacognitive beliefs about alcohol use may be a valuable add-on intervention to
restructuring alcohol outcome expectancies. Finally, in case of relapse of problem drinking behavior, it
may be helpful to derive and illustrate the role of metacognitive beliefs about alcohol use in the given
episode together with other relevant cognitive-behavioral constructs.
The present results are preliminary in nature. Clearly future studies are required to further establish the
psychometric properties of both PAMS and NAMS. In particular, it would be necessary to determine the
structure and reliability over time and with other samples. In addition studies are required to examine the
sensitivity of both scales to treatment effects and recovery if the scales are to prove useful treatment
evaluation tools. The role of high levels of positive metacognitive beliefs in predisposing individuals to
526 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527

engage in alcohol use, and of negative metacognitive beliefs in maintaining problematic drinking
behaviors, could also be investigated through longitudinal designs.
Research has already indicated that metacognitive beliefs about alcohol use are construct that signi-
ficantly extends the scope of alcohol outcome expectancies assessment into the domain of metacognition
(Spada et al., 2007), however future research will have to ascertain the relative contribution of meta-
cognitive beliefs about alcohol use and alcohol outcome expectancies to drinking behavior in clinical
samples. It will also be necessary to compare PAMS and NAMS with related measures such as the
Impaired Control Scale (Heather, Tebbutt, Mattick, & Zamir, 1993) the Drinking Motives Questionnaire
(Cooper, 1994), as well as self-efficacy beliefs.
Results of this study must be considered with regard to design limitations. Social desirability, self-
report biases, context effects and poor recall may have contributed to errors in self-report measurements.
However, despite these limitations, we believe that both instruments may already be useful for eliciting
positive and negative metacognitive beliefs in clinical assessment and case formulation, and in
providing a further step towards the development of a metacognitive conceptualisation of problem
drinking.

References

Baker, C. A., & Morrison, A. P. (1998). Cognitive processes in auditory hallucinations: Attributional biases and metacognition.
Psychological Medicine, 28, 1199−1208.
Babor, T. F., de la Fuente, J. R., Saunders, J., & Grant, M. (1992). The alcohol use disorders identification test: Guidelines for use
in primary healthcare. Geneva: World Health Organisation.
Bandura, A. (1997). Self-efficacy: The exercise of control. New York, USA: Freeman.
Bouman, T. K., & Meijer, K. J. (1999). A preliminary study of worry and metacognitions in hypochondriasis.Clinical Psychology
and Psychotherapy, 6, 96−101 Special issue, Metacognition and Cognitive Behaviour Therapy. Chichester, UK: Wiley.
Brown, S. A., Christiansen, B. A., & Goldman, M. S. (1987). The alcohol expectancy questionnaire: An instrument for the
assessment of adolescent and adult expectancies. Journal of Studies on Alcohol, 48, 483−491.
Caci, H., Bayle, F. J., Mattei, V., Dossios, C., Philippe, R., & Boyer, P. (2003). How does the hospital anxiety and depression
scale measure anxiety and depression in healthy subjects? Psychiatry Research, 118, 89−99.
Cahalan, D., Cisin, I., & Crossley, H. (1969). American drinking practices: A national survey of drinking behaviours and
attitudes. Monograph no. 6. New Brunswick, NJ: Rutgers Centre for Alcohol Studies.
Christiansen, B. A., Smith, G. T., Roehling, P. V., & Goldman, M. S. (1989). Using alcohol expectancies to predict adolescent
drinking behavior after one year. Journal of Consulting and Clinical Psychology, 57, 93−99.
Clark, D. M., & Wells, A. (1995). A cognitive model of social phobia. In R. Heimber, M. Liebowitz, D. A. Hope, & F. R.
Schneier (Eds.), Social phobia: Diagnosis, assessment and treatment. New York: Guilford Press.
Cooper, M. L. (1994). Motivations for alcohol use among adolescents: Development and validation of a four-factor model.
Psychological Assessment, 6, 117−128.
Emmelkamp, P. M. G., & Aardema, A. (1999). Metacognitions, specific obsessive–compulsive beliefs and obsessive compulsive
behaviour. Clinical Psychology and Psychotherapy, 6, 139−145 Special issue, Metacognition and Cognitive Behaviour
Therapy. Chichester, UK: Wiley.
Fabrigar, L. R., Wegener, D. T., MacCallum, R. C., & Strahan, E. J. (1999). Evaluating the use of exploratory factor analysis in
psychological research. Psychological Methods, 4, 272−299.
Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. American
Psychologist, 34, 906−911.
Goldman, M. S., Del Boca, F. K., & Darkes, J. (1999). Alcohol expectancy theory: The application of cognitive neuroscience. In
H. Blane & K. Leonard (Eds.), Psychological theories of drinking and alcoholism. New York: Guilford Press.
Heather, N., Tebbutt, J. S., Mattick, R. P., & Zamir, R. (1993). Development of a scale for measuring impaired control over
alcohol consumption: a preliminary report. Journal of Studies on Alcohol, 54, 700−709.
 M.M. Spada, A. Wells / Addictive Behaviors 33 (2008) 515–527 527

Hermans, D., Martens, K., De Cort, K., Pieters, G., & eelen, P. (2003). Reality monitoring and metacognitive beliefs related to
cognitive confidence in obsessive–compulsive disorder. Behaviour Research and Therapy, 41, 282−401.
Hester, R. K., & Miller, W. R. (1995). Handbook of alcoholism treatment approaches: Effective alternatives. Needham Heights,
USA: Allyn & Bacon.
Jones, B. T., Corbin, W., & Fromme, K. (2001). A review of expectancy theory and alcohol consumption. Addition, 96, 57−72.
Kline, R. B. (1998). Principles and practice of structural equation modeling. New York: Guilford Press.
Leigh, B. C. (1989). In search of the seven dwarves: Issues of measurement and meaning in alcohol expectancy research.
Psychological Bulletin, 105, 361−373.
Maisto, S. A., Connors, G. J., & Sachs, P. R. (1981). Expectation as a mediator of alcohol intoxication. Cognitive Therapy and
Research, 5, 1−18.
Matthews, G., Hillyard, E. J., & Campbell, S. E. (1999). Metacognition and maladaptive coping as components of test anxiety.
Clinical Psychology and Psychotherapy, 6, 111−125.
Morrison, A. P., French, P., & Wells, A. (2007). Metacognitive beliefs across the continuum of psychosis: Comparisons between
patients with psychotic disorders, patients at ultra-high risk and non-patients. Behaviour Research and Therapy, 45,
2241−2246.
Morrison, A. P., Wells, A., & Nothard, S. (2000). Cognitive factors in predisposition to auditory and visual hallucinations. British
Journal of Clinical Psychology, 39, 67−78.
Moses, L. J., & Baird, J. A. (1999). Metacognition. In R. A. Wilson & F. C. Keil (Eds.), The MIT encyclopedia of the cognitive
sciences. Cambridge, USA: The MIT Press.
Myers, S. G., & Wells, A. (2005). Obsessive–compulsive symptoms: The contribution of metacognitions and responsibility.
Journal of Anxiety Disorders, 19, 806−817.
Mykletun, A., Stordal, E., & Dahl, A. A. (2001). Hospital anxiety and depression scale: Factor structure, item analyses and
internal consistency. British Journal of Psychiatry, 179, 540−544.
Papageorgiou, C., & Wells, A. (2003). An empirical test of a clinical metacognitive model of rumination and depression.
Cognitive Therapy and Research, 27, 261−273.
Roussis, P., & Wells, A. (2006). Post-traumatic stress symptoms: Tests of relationships with thought control strategies and beliefs
as predicted by the metacognitive model. Personality and Individual Differences, 40, 111−122.
Spada, M. M., Hiou, K., & Nikčević, A. V. (2006). Metacognitions, emotions and procrastination. Journal of Cognitive
Psychotherapy, 20, 319−326.
Spada, M. M., Moneta, G. B., & Wells, A. (2007). The relative contribution of metacognitive beliefs and expectancies to drinking
behaviour. Alcohol and Alcoholism, 42, 567−574.
Spada, M. M., Nikčević, A. V., Moneta, G. B., & Ireson, J. (2006). Metacognition as a mediator of the effect of test anxiety on
surface approach to studying. Educational Psychology, 26, 1−10.
Spada, M. M., Nikčević, A. V., Moneta, G. B., & Wells, A. (2007). Metacognition as a mediator of the relationship between
emotion and smoking dependence. Addictive Behaviors, 32, 2120−2129.
Spada, M. M., & Wells, A. (2005). Metacognitions, emotion and alcohol use. Clinical Psychology and Psychotherapy, 12, 150−155.
Spada, M. M., & Wells, A. (2006). Metacognitions about alcohol use in problem drinkers. Clinical Psychology and
Psychotherapy, 13, 138−143.
Spada, M. M., Zandvoort, M., & Wells, A. (2007). Metacognitions in problem drinkers. Cognitive Therapy and Research, 31(5),
709−716.
Stacy, A. W., Widaman, K. F., & Marlatt, G. A. (1990). Expectancy models of alcohol use. Journal of Personality and Social
Psychology, 58, 918−928.
Wells, A. (2000). Emotional disorders and metacognition: Innovative cognitive therapy. Chichester, UK: Wiley.
Wells, A., & Matthews, G. (1994). Attention and emotion. A clinical perspective. Hove, UK: Erlbaum.
Wells, A., & Matthews, G. (1996). Modelling cognition in emotional disorder: The S-REF model. Behaviour Research and
Therapy, 34, 881−888.
Wells, A., & Papageorgiou, C. (1998). Relationships between worry and obsessive–compulsive symptoms and meta-cognitive
beliefs. Behaviour Research and Therapy, 36, 899−913.
Yussen, S. R. (1985). The role of metacognition in contemporary theories of cognitive development. In D. L. Forrest-Presley, G. E.
MacKinnon, & T. G. Waller (Eds.), Metacognition, cognition, and human performance. New York, USA: Academic Press.
Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety and depression scale. Acta Psychiatrica Scandinavica, 67,
361−370.