Chapter V

The Shorter PROMIS Questionnaire (SPQ) in an Addicted treatment population

Index

5.1. Introduction

5.2. Method

5.3. Results

5.3.1. Component Analysis (PCA)

5.3.ii. Factor Analysis

5.3.iii. Cluster Analysis

5.4 Conclusions

5.1. Introduction

The review in chapter three indicated that the comprehension and treatment of all forms of addictive behaviour is an enormous challenge, because they are persistent, resilient and often reoccurring problem behaviours. Indeed, understanding the complexities underlying each unitary disorder has created independent fields of research with many associated specialised journals. It could be argued that with this approach, addiction professionals “hijacked” the term addiction by permitting only a narrow range of areas to be encompassed by the definition. Now, however, with factors such as the incidence of cross addiction and the noting of common dynamics between certain behaviours (e.g. Marks, 1990) an ever increasing number of behaviours seem to be gathering under the addictive umbrella. When considering such a range of potentially addictive behaviours and the cross-overs and relationships between them, addiction may be seen to be an even more serious problem, and the search for a better understanding of the processes involved in and between these relationships seems ever more pertinent.

It was seen in the review in chapter three that the identification of patterns of multiple use in a wide variety of addictive behaviours is in its infancy. To date it is only Stephenson et al’s (1995) research which has attempted to do this. This study found two main factors, the first factor, Hedonism, comprising the following areas: Recreational drug use, Prescription drug use, Alcohol use, Sex exploitation, Relationship exploitation, Gambling and Nicotine use. The second factor Nurturance, was comprised of Food bingeing, Food starving, Caffeine, Shopping/spending, Compulsive helping, Exercise and Work. The results suggest that addictive behaviours are used in systematic ways in men and women.

These findings are important for a number of reasons. First, when it comes to treatment, quite often only one addictive problem is addressed by treatment agencies rather than investigating what may well be the generic addictive process and its “variety” of addictive manifestations. This has important implications for relapse prevention, as different behaviours may be substituted for the original addiction.

Second, even though work has been conducted investigating the links between the several addictive outlets, systematic consideration needs to be given to the greater variety of addictive behaviours. This is to ensure that the process, or processes, which underlie or drive certain characteristic orientations, and which lead to specific combinations of behaviours being used, are more fully understood. Far from the study of addiction being devalued by the increasing range of behaviours which are being termed as addictive, these additional behaviours may in fact illuminate the broader process of mood alteration. By studying patterns of co-variation between addictive behaviours their more general psychological significance may be grasped more securely. This in turn may lead to improved treatment design and consequent improved outcomes for sufferers.

Although there are likely to be some distinctive features in different addictions, and continued research into the separate behaviours prove beneficial, there are undoubtedly common features which require systematic study. Concerns over the pursuit of commonality between syndromes cannot be overlooked as it is possible that this emphasis may ignore important etiological differences in the onset of different addictive behaviours. Notwithstanding this objection, in chapter three it was noted that there are a number of levels of analysis that have been used to “explain” and support the idea of strong commonalities between different addictive behaviours, such as the neuro-chemical (e.g. Koob & Le Moal, 1997). This type of research provides a further level of justification for the pursuit of a more comprehensive understanding of the common processes which may underlie this range of behaviours. Revealing systematic patterns, as this study aims to do, may at least guide further research aimed at clarifying the underlying processes of cross-addiction (cf. Kosten, Rounsaville, Babor, & Spitzer, 1987)

Examples of theoretical ideas which may shed light on the differing patterns of “addictiveness”
It may be argued that there is truth in the idea that the urge to avoid pain and seek pleasure is part of our biological make up, and that both the excessive involvement in and normal use of the wide variety of potentially addictive behaviours mentioned thus far is, at least in part, a consequence of this. However this simplistic notion doesn’t explain why certain behaviours emerge as candidates for addiction over and above others. Individual differences based in biology and psychological motivation play a role, but are unlikely to provide a complete answer. Evidence for addictive orientations may point the way to a more satisfying general theory.

Social learning theory has been utilised as a useful model for explaining excessive drinking (Abrams & Niaura, 1987) and can readily be adapted for patterns of addictive use where a number of behaviours are used. People use an addictive behaviour, according to social learning theory, for at least three reasons. First that the effects of the behaviours bring pleasure; second, a person may decide earlier that the behaviours are consistent with personal standards (cognitive mediation) and third, the person may learn to use the behaviours through the observation of others (modelling) (op. cit.). Any one of these factors or a combination is sufficient to initiate and guide use of an addictive behaviour. So here the possibility is that “orientation” may be a question of consistent exposure or learning in people whose values are similar and who are exposed to similar models and reinforcements.

A different explanation involves the idea that there is a biological substrate which links certain types of behaviours with certain physiological types. Jacobs (1989) defines addiction as, “A dependent state acquired over time by a predisposed person in an attempt to relieve a chronic stress condition” (P.35). He posits that two sets of predisposing factors much be present for an individual to develop an addiction. The first is an abnormal physiological resting state that is chronically either excessively excited or depressed and the other being of a psychological nature, characterised by feelings of inferiority, rejection, inadequacy and or guilt stemming from childhood. It is thought that the physiological condition of either being chronically excited (highly aroused) or depressed is stress inducing.

Individuals suffering from either of these extreme arousal levels are therefore thought to be motivated to seek activities or substances that correct this resting state, with the goal of obtaining a more comfortable homeostatic position. So, a person with a hypotensive physiological arousal level may find relief in a stimulating and exciting activity such as gambling. This temporarily eliminates feelings of boredom, or a more general feeling of emptiness or flatness. A person with a hypertensive arousal state may find relief in substances or behaviours that have a depressing or calming effect, such as alcohol or tranquillisers. This line of reasoning suggests that certain sets of behaviours are used flexibly, some sets being stimulating and some calming.

In sum, research has revealed that addictive behaviours co-vary in systematic ways. In addition, some theoretical ideas illuminate why such orientations may occur. However, until we can say with more certainty that such orientations as indicated in Stephenson et al’s 1995 study are indeed present, searching for explanations of co- variation would be premature.

Aims
Bearing the above comments in mind, the aim of the present study is to investigate patterns of addictive co-variation in reported usage in a large variety of addictive behaviours using the Shorter PROMIS questionnaire (SPQ), to see whether the orientation revealed by Stephenson et al’s 1995 study are replicable and are capable of further refinement. It is hoped that a clarification of those earlier findings may be achieved, and a more in-depth exploration of the nature, significance and possible utility of the orientations be made possible.

5.2. Method

The data set
When clients enter PROMIS a full medical history is taken by nursing staff and this, in conjunction with psychiatrists’ reports and other referral information, provides the basis of the primary diagnosis. At a later stage, usually within the first 7 days of treatment, Questionnaires are administered to patients and are used as part of the full assessment process. The questionnaires include the SPQ, and others such as the Eating Disorder Inventory (EDI; Garner, Olmstead & Polivy, 1983) and the Brief Symptom Inventory (BSI; Derogatis, 1993). The results of the SPQ are routinely fed back to patients. The questionnaires are administered by a research psychologist and the usual conventions of psychometric research obeyed. Where a patient was on a chemical detoxification programme, which may have impaired their concentration, questionnaires were administered when the patient was fully detoxified.

Patients were informed of the ongoing research programme at PROMIS and written permission was asked for their responses to be used for research purposes. Before completing the questionnaire they were assured of complete confidentiality and given the opportunity not to participate.

Data were obtained for 543 consecutive admissions to the PROMIS Recovery Centre between 1995 and 1999; 285 were male and 258 were female. Their ages varied from a minimum of 14 to a maximum of 79 with a mean of 35 and a standard deviation of 12.8.

Table 5.1. Summary table for treatment population by diagnostic category

Statistical analysis

Data were analysed using SPSS version 8 statistical software. Distributions were plotted and were seen to be variously skewed, with a number of the scales exhibiting ceiling effects. Patient’s scores according to diagnosis can be found in Appendix 2.

5.3. Results

When dealing with multivariate data there is often evidence of interdependence between the variables. This is easily detected by forming the correlation matrix for the data and studying the off-diagonal elements, which show the interdependence between the variables. These entries take values between 0 and 1, with 0 showing no dependence and 1 showing perfect correlation. The correlation matrix for the SPQ Treatment data is given in table 5.3 below. It can be seen that there are many large and statistically significant (i.e. >0.09) off-diagonal elements, this suggesting that the data can be simplified, or explained by a lesser number of variables.

Table 5.3: Treatment population. Correlation table for the sixteen scales of the SPQ

 

Alcohol (AL), Shopping (SH), Food Bingeing (FB), Compulsive Helping Submissive (CHS), Nicotine (NI), Gambling (GA), Food Starving (FS), Compulsive Helping Dominant (CHD), Drugs (DR), Sex (S), Work (W), Relationships Dominant (RD), Caffeine (CA), Prescription Drugs (PD), Exercise (EX), Relationships Submissive (RS).

It can be seen that the correlations are predominately positive, though there are a very few notable negative exceptions, in particular with respect to correlations with food bingeing. Overall the matrix indicates that factor analysis will yield a solution which reflects the generally positive pattern of co-variation.

There are two main approaches to reducing the dimensions of this data, namely Principal Component Analysis (PCA) and Factor Analysis (FA), and it was thought that both of these approaches would be useful in attempting to reduce the dimensions.

5.3.1. Results

PCA works by sequentially searching for the dimensions that explain the greatest amount of variation in the data, such that each dimension is mutually orthogonal. These dimensions are defined by the eigenvectors of the correlation matrix (or of the co-variance matrix when all variables are measured on the same scale), and their relative importance is defined by the size of their corresponding eigenvalues. Because this new co-ordinate system is not invariant to scale, a spectral decomposition was performed to find these dimensions on the correlation matrix.

Hence, the Principal Components (PCs) are linear combinations of the original variables with weights given by the standardised eigenvectors and relative importance given by the eigenvalues. When PCA is performed on the correlation matrix the proportion of variance explained is given by the ratio of eigenvalue to number of variables. So for these 16 variables, PCs are of interest with eigenvalues greater than 0.8 (5%) but preferably greater than 1.6 (10%).

Application of these criteria to the analysis of the SPQ data, it is likely that there are probably between 3 and 6 underlying dimensions. However, only the first two of these PCs are particularly interpretable with the first being a rough average of all 16 variables and the second being a contrast of “hedonistic” type behaviour (drugs, alcohol, sex, etc) and “nurturant” type behaviour (shopping, food bingeing, etc). A further mathematical analysis that allows greater manipulation to produce more meaningful variables, is Factor Analysis (FA).

To perform Factor Analysis a number of assumptions need to be made about the number of underlying factors needed in the model. There are a number of ways of choosing the number of factors to be included, based on the spectral decomposition and the variance explained by each dimension. By using the Kaiser Criterion the eigenvalues of the correlation matrix greater than one may be considered to be significant. In addition to this, simulation studies by Zwick and Velicer (1986) found that scree plot and MAP (minimum average partial correlation) were the best performing rules for deciding on model size. In this study model size was based on the scree plot which is an intuitive and reliable method, and the results from the PCA. The scree plot is a plot of the eigenvalues associated with a factor versus the number of the factor and the point where the smooth decrease of the eigenvalues appears to level off to the right of the plot is looked for.

Figure 5.1: Treatment population. Scree plot of the 16 scales from the SPQ

 

 

Table 5.4: Treatment population. Results from the Principal Components Analysis

 

Principal components	Eigenvalue	Proportion	Cumulative proportion
PRIN 1	4.66755	0.291722	0.29172
PRIN 2	2.36730	0.147956	0.43968
PRIN 3	1.61876	0.101172	0.54085
PRIN 4	1.16459	0.072787	0.61364
PRIN 5	0.91446	0.057154	0.67079
PRIN 6	0.82208	0.051380	0.72217

Looking at figure 5.1, clearly this plot, in conjunction with the results from the Principal Components Analysis (table 5.4) suggests that it is models with between 2 and 4 dimensions that need to be investigated, as further dimensions do not have a noticeable effect in reducing the variance. A Factor Analysis was conducted to identify 4 factors and this was compared with a 3 and a 2 Factor Analysis.

5.3.ii. Factor Analysis

Factor Analysis works by assuming a mathematical model for the data that consists of a number of underlying factors that are not necessarily measurable as such (e.g. intelligence). It assumes that these factors can explain all the co-variance between the measurable variables, and most of the variance. Estimation of this model can be done in a number of ways, the most common being either Maximum Likelihood or Principal Axes Factoring. For this analysis we chose Principal Axes Factoring (PFA) as it involves fewer distributional assumptions, and converges quite quickly. (In this application typically after 15 - 20 iterations).

What is of particular interest is the loading of each factor for the particular variables, as this permits the interpretation of the factors. Large loadings indicate agreement between the variable and the factor, and hence allows an interpretation of what the underlying factor may be measuring. For instance, if factor 1 has large loadings for nicotine, gambling, and alcohol, it could be said that the underlying factor is measuring legal forms of mood altering.

Because of the statistical assumptions of the factors, zero mean and unit variance, calculating the variance explained by the model for a particular variable amounts to calculating the sum of squares for the loadings, known as the communality. So the communality of a variable is the proportion of the variance of that variable explained, and the total communality is the total proportion of variance explained by the model.

One positive aspect of this type of analysis is that any consistent factor model will remain a consistent factor model after any orthogonal transformation. Hence the initial factor model can be rotated to produce a model that is easier to interpret. The most usual rotation to use is the “Varimax” rotation, which is the method used in this analysis. Varimax rotation is an orthogonal rotation which works by maximising a variables loading with respect to one and only one factor, hence resulting in a distinct construct. It can be opposed to an oblique rotation which is more complex to analyse. After 15 iterations the PFA process converges to produce the following results for the rotated factor loadings and their communalities estimates.

The graph of the communality values shows that this model explains a large part of the variability for most of the variables. There is, nevertheless, a wide range of values, and Alcohol and Gambling, in particular, fit less convincingly into the factor structure than do others. Some reasons for this will be discussed later, especially in relation to the results for alcohol.

The proportion of variance explained by each factor is given by the ratio of the sum of the squared loadings and the total variance. Hence the proportion of variation explained by each factor is:

FACTOR 1 16.9%
FACTOR 2 14.8%
FACTOR 3 11.4%
FACTOR 4 11%

The following four graphs concern the loadings of the factors. The loadings indicate the proportion each variable contributes to a particular factor and so by looking at them an interpretation of what each factor is measuring can be made. For this analysis a Varimax rotation was used which works by maximising a variable’s loading with respect to just one factor. So each factor’s main constituents should be different from each other factor’s main constituents.

The first graph shows high loadings (.5 and above) for Food disorders, Shopping and Caffeine, which could be seen as behaviours which are more characteristic of women’s behaviour (Chrisler, 1989). Using Stephenson et al’s (1995) descriptors, these Nurturant behaviours if engaged in to excess can be seen to be to do with the consumption or use of behaviours concerned very much with legal and acceptable ways of soothing or coping with the self. And so this factor has been labelled, “Self orientated Nurturance”. This factor can be seen to be undoubtedly more socially acceptable than behaviours such as drug taking and gambling, and can be viewed in a different light again when compared with working, and helping others (factor 2).

The second graph clearly shows the high correlation of the factor with variables such as the Compulsive Helping dimensions and Work. These behaviours could be aptly termed “Other orientated Nurturance”, as the focus here seems to be on providing comfort or service for others. In this factor Nurturant behaviours can be characterised as behaviours which when normally used can be seen to be essentially pro-social habits: determined engagement with work, and helping others, which may be seen as acceptable even commendable activities that others may aspire to. However, used in an addictive fashion, the addict’s mood may well be altered though the utilisation of such activities.

In the third graph we can see that the variables that contribute most to the factor are Drugs, Prescription Drugs, Nicotine and, to a lesser extent, Alcohol. These may be considered as the ‘physical addictions’. Here we are looking at the “aggressive”, obvious and traditional addictive forms of mood altering. Factor 3 has a flavour of recklessness and a sense of active and demonstrative self indulgence. Here we have mood altering in a deliberate less socially desirable way. This factor may be termed, “Sensation seeking Hedonism”.

Finally in the fourth graph it can be seen that the main constituents are the Relationship dimensions, Sex and, to a lesser extent, Gambling. This factor seems to be related to behavioural mood altering, which again may be seen as active, deliberate and related to sensation seeking. Recall that the Relationships scales reflect the exploitation of other people, whether through the exercise of power directly, or through the use of more subtle psychological manipulation. A similar desire to control is evident in Gambling and more particularly in Sex. These behaviours, although not substance based, when engaged in excessively are nonetheless largely frowned upon. This dimension will be termed “Power related Hedonism”.

Interestingly this four factor solution has produced two Hedonistic and two Nurturant factors which are, in effect, subdivisions of the previously found Hedonistic and Nurturant dimensions in the previous study (Stephenson et al, 1995). To help draw a more explicit comparison with that earlier study, I shall now report the results of a Cluster Analysis, which is more suited to displaying what appears to be a quasi-hierarchical structure in the pattern of co-variation.

5.3.iii. Cluster Analysis

Cluster Analysis is a technique used for combining observations into groups or clusters such that each group contains relatively homogenous observations, and that each group has different characteristics from other groups. This analysis is not directly comparable to Factor Analysis, but adds evidence to whether there are in fact different categories for the observations, and whether these categories can be characterised by particular features. Cluster Analysis works by measuring the dissimilarity between each observation and grouping together observations that are most similar.

Pairs of observations are looked for that are closest and then these are grouped together. This process is then repeated, grouping together observations that are closest together. When a cluster is found it has to be decided where to start measuring from. It was decided that the three main different methods would be investigated to see how they related to the Factor Analyses. The three methods of measuring from a cluster to an observation outside the cluster are as follows:

Single Link: measures from outside observation to nearest observation in cluster
Complete Link: measures from outside observation to furthest observation in cluster
Average Link: measures from outside observation to each observation and finds the mean.

Of the three methods, the conventional Complete Linkage technique revealed a coherent hierarchical structure which confirmed expectations of the importance of the two Hedonistic and Nurturant orientations. The dendogram of the cluster analysis using the Complete Linkage method shows the possibility of grouping the variables into the following four groups given above.

1. Shopping, Food disorders, Caffeine;

2. Compulsive helping dimensions, Work, Exercise;

3. Alcohol, Drugs, Nicotine; Prescription drugs;

4. Gambling, Sex, Relationship dimensions.

These clusters closely match the four factors of the Factor Analysis, and provide strong confirmation the significance of the 4 factors grouping, within the broad framework of Hedonistic and Nurturant orientations.

5.4. Conclusions

There are a number of points to be made at this juncture, before I discuss more fully the significance of the individual factors.

1. Positivity in the Addictive Behaviours Correlation Matrix
The review of literature revealed an increasing tendency to assume that common motivations may underlie different addictive behaviours, such that one might expect a broadly positive pattern of co-variation to obtain between different addictive behaviours. Indeed, looking at the correlation matrix (Table 5.3.), it can be seen that this, by and large, is the case. With few exceptions involving food scales and “aggressive” hedonistic scales for the most part, the correlations in the matrix are positive. The evidence, so far as it goes, is supportive of the notion of a general, albeit weak, tendency towards excess across the full range of behaviours assessed by the SPQ.

2. Factorial Structure
The relationships between many variables are insignificant, and explanation of the co-variation benefits hugely from a consideration of the structure of inter-relationships as revealed by factor and cluster analysis. Factor analysis suggests that four factors are necessitated, these being replicated in the cluster analysis.

3. Replication of the Stephenson et al, (1995) study
One of the main aims of the study was to investigate patterns of co-variation in a large variety of addictive behaviours to see whether the orientation revealed by Stephenson et al’s 1995 study was replicable.

On the basis of the scree plot, the principal components analysis, and the factor analysis, a four factor solution was chosen as the best model to explain the data. Additional evidence for the appropriateness of this number of factors was given by the cluster analysis. These four factors represented subdivisions of the two factors found in the previous study. Hence, the results from the current factor analysis indicates that the findings from Stephenson et al’s 1995 study are broadly supported, and further clarified, employing a more valid psychometric instrument.

Hierarchical pattern

The cluster analysis suggests that the four factors are grouped in a way that not only replicates the results of the previous study, but also portrays the further links between the two orientations of Hedonism and Nurturance, suggestive of a hierarchical structure. At the broadest level, the participants in the study can be described in terms of the extent of their overall addictive tendencies. At the next level, they can be seen to be differently orientated towards Hedonism or Nurturance. At the next level, more precisely, they exhibit systematic variation in addictive tendencies, towards self, or other directed, nurturance; and sensation-seeking, or power-related, hedonism. Finally, they can, of course be depicted in terms of specific behavioural addictive characteristics.

Factors and illusory correlations

From the previous literature it seemed that different addictions did co-vary. Some of these suggestions have been confirmed by our results, for example drugs and nicotine. Given the existence of a general tendency towards “addictiveness” as suggested by the overall correlation matrix, it may be expected that when looking at the correlation between any two behaviours i.e. drugs and nicotine, that there may well be a positive relationship, as it were, by chance. However, the factorial structure indicates which addictive behaviours consistently co-vary, and factorial structure suggests that some “correlations” observed in the past may well have been illusory. For example, in one study high levels of alcohol and drug abuse were observed in 40% of individuals with a diagnosis of anorexia or bulimia (Zerbe, Marsh & Coyne, 1990). The results here do not support the view that these two behaviours consistently co-vary. Moreover, the data suggest that there is a stronger, more consistent association between the food disorders with other behaviours such as work and caffeine. This may prompt future research to examine what might be the common motivational basis for the observed co-variation within such clusters of addictive behaviours, thereby leading to considerable enhancement of theory.

The nature, significance and utility of the four factors.

Factor 1 “Self Orientated Nurturance”

Looking at the first factor this can be seen to be a Nurturant factor with the inclusion of both the food dimensions, shopping, and caffeine. Though exercise fails to meet the .5 threshold this dimension loads more highly on this factor than any other and when looking at the cluster analysis it can be seen to be most linked with this grouping of addictive behaviours. This grouping of behaviours may be seen to be motivated by or related to control of body image.

The high correlation between the two food disorders may appear strange, as intuitively they seem to be at opposite poles of a food problem continuum. But in practice it seems that those who have difficulty in controlling excessive consumption have periods of severely restricting their food intake, and would therefore on this questionnaire score on the food starving dimension. This occurs in the opposite direction as well, in that people who severely restrict their food intake, on occasion go through periods of what they at least would term, excessive consumption (Slade & Duker, 1988). This may or may not actually constitute a binge, but for the sufferer the binge episode, or period of time where their control weakens, feels excessive. Indeed it is not unusual to find those in the food disorder category who go through phases of over-eating and then severely under-eating, to the extent that these different forms of “using “ food have become commonplace (op cit.).

These extremes may reflect a generalised tension and problem with food. In bulimia research there has been debate as to whether this disorder represents one end of a generalised eating disorder continuum or is somehow categorically different from sub-threshold bulimia or an absence of eating disorders. Results have been mixed with the continuity hypothesis being supported for measures of weight concerns and the discontinuity hypothesis being supported for measures of psychopathology (Stice Killen, Hayward, & Taylor,1998).

The other scales in this cluster (shopping and caffeine and, marginally, exercise) seem to be characterised by a concern with consumption, this either being positive or negative, which is directly related to the self. It has been suggested that these behaviours may be linked, in that they may perform some form of compensatory behaviour, possibly to do with masking feelings of lack, or feeling unloved or not good enough. Feelings of unworthiness or emptiness may be alleviated by over-eating or over-spending; and with these behaviours there is a sense of “filling up”. The idea of “retail therapy”, it may be observed, has become commonplace within western culture.

The power in this set of addictive behaviours lies in the dominance of these more feminised activities: body image and concerns over planning eating/non-eating, exercise regimes and shopping.

Factor 2 “Other Orientated Nurturance”

Interestingly with this set of dimensions when looking at the cluster analysis it can be seen that they form a sub division of the Nurturant dimension, and therefore are more associated with Factor 1 “Self orientated nurturance”. Helping others and determined engagement with work can, on the surface, be seen as commendable activities, though their excessive use is now starting to be reported as problematic (Cermak, 1991). This is due to the fact that when engaged in to excess, though the individual may feel better as any internal discomfort or confusion is externalised, there is a reduction in the individual’s ability and desire to look at themselves. The concept of control here is interesting as it is possible that with this factor that engagement in this range of addictive behaviours is also related to the desire to achieve control (as in the previous factor). A further point is that the methods chosen to achieve this are very different, the first factor involving a focusing on the self and this second factor energy being directed to the external world.

When busy looking after other people’s needs or compulsively working, there is, by default, a reduction in the time available for addressing the self. Perhaps the mechanism also involves an unworthy self and this is compensated for through helping others and through over work, i.e. excelling in other people’s opinions. It is in this sense that this cluster of behaviours could be to do with conformity and approval seeking and control. It is possible that when these dimensions are used in isolation without other forms of addictive use, consequent problems stemming from these behaviours may be more difficult to detect. Here the power of the combination of activities is effective as huge effort is expended in looking after external situations and people.

Factor 3 “ Sensation Seeking Hedonism” (Substances)

With this combination of addictive behaviours comes the more radical, reckless and pharmacological ways of mood altering that are less socially acceptable than the previous two factors. In the same way as in factor one this combination seems very much to do with the self, but in this factor the combination of behaviours share the fact that these methods are more direct and aggressive forms of mood altering. It has been stated before that the issue of control is a useful dimension when looking at addiction, and it is likely with this factor that engagement in this range of addictive behaviours is to do with the desire to lose control.

A further idea is that here the individual’s feelings, whether negative or dull etc, are such that solace can only be found in a fast-acting and perhaps more importantly a dramatic way of mood alteration. Perhaps this degree of mood change or shift in perception is determined by a need for a sharp elevation in mood, a shift which can be perceived immediately. Perhaps the same feelings of inadequacy and feeling not right in some way are, for some reason, directed into these extreme behaviours. Perhaps this orientation is led by a combination of factors (such as the environment in which they are used) that leads the individual past the more acceptable/legal ways of mood altering, into the more aggressive forms and so a cycle of drug taking (whether prescription or illicit or a combination of both) and alcohol use is entered into.

Factor 4 “Power Related Hedonism”

The significance of the grouping of these dimensions seems to relate to behavioural mood altering which is hedonistic and related to power. These dimensions are hedonistic and the distinction between the previous factor and this one is that substances are not involved and it is largely behavioural. The power element seems to be an apt concept to unify both the relationship dimensions and sex, as it is feasible that a degree of power or manipulation over another person is necessary to deliver the response for the addict’s satisfaction. Also with gambling there is a sense that an underlying dimension which may run beneath the thrill of the chase and of a winning rush in gambling may also be related to power.

One difference which is similar to the distinction between factors one and two is that in factor four though Hedonistic these behaviours are to do with others or are external to the self (i.e. gambling). When considering the concept of control with this dimension, it seems that the underlying objective has more to do with the desire to achieve control over others or the external (i.e. gambling) in order to achieve the desired effect.

In sum from these results it seems as though there may be an additional way to view “choice” of addictive behaviour which adds a further dimension of explanation to the field. Instead of viewing these behaviours as completely different and in isolation or as sharing many common dynamics as a number of theorists are moving towards (e.g. Marks, 1990) it may be useful to think in terms of orientation. This permits the differences between these behaviours to be acknowledged while at the same time room for their similarities and co-occurrence is made.

Drawing from the review chapter it can be argued quite convincingly that there is enough evidence which suggests that there are substantial commonalities between the array of addictive behaviours, what these results add is that the four factors may provide a convenient structure with which these commonalities and reported incidence of cross addiction can be understood. For example the high incidence of alcoholism and drug addiction can be seen as a general orientation towards a need to lose control in an aggressive and direct way. Though it must be said that this orientation doesn’t investigate how these disorders develop it may aid in the generation of new ideas or provide support for established theories.

Another area where this orientation may help is when health professionals are assessing patients suffering from addiction, as using this structure may help them to keep an open mind to further problem behaviours, such as exercise being an associated problem with eating disorders. A further utility may be in the area of predicting which behaviours may become problematic if a genuine and deep recovery for the primary addiction isn’t instated during treatment.

In general it may be said that at a fundamental level these behaviours have similarities such as loss of control, tolerance etc. At a level beyond this a splitting seems to occur and this may have something to do with personality, developmental factors or environmental pressure i.e. learning, exposure or availability. If this is indeed the case this casts doubt on the premise that these behaviours are the same, as the origin, function and motivation for there use may stem from quite different sources.

In the next chapter factor analysis was performed on a set of data completed by a group of people not suffering from addiction to investigate whether any form of orientation is present in a population where, for the vast majority of participants, these behaviours are used in a “normal” and non-problematic fashion.