When considering the choice of a theory to use when investigating the role of control beliefs in the relationship between SES and health, several points need to be considered. Firstly, the theory should include an understanding of 'perceived control' and not be limited to a 'personal control' understanding. One reason for this is that each of the major
'personal control' theories were developed to explain and predict behaviour (e.g. Abramson et al., 1978; Bandura, 1977). In that context it makes sense that belief in
'personal control' is beneficial to the health of the person because a sense of 'personal control' increases the person' s motivation to perform beneficial behaviours, while belief that control is held outside oneself does not encourage the performance of these
behaviours (Wallston, 1992). However, when the aim is to explain and predict health in general, a broader understanding of control is needed because there are factors other than behaviours which affect health. For example, stress has been shown to influence health. To investigate the role of control in the relationship between SES and health, an
understanding of control which is not limited to the affects of health behaviours is needed. A second reason it is important to include a 'perceived control' understanding of control, is that some argue that attributing the relationship between SES and health to personal control beliefs is like blaming the victim (Brownell, 199 1 ). Franzblau and Moore (2001) argue that the emphasis in western society on 'personal control' blames the victim because it attributes to them responsibility for their situation, rather than recognising the influence of the individual's social context. In effect it suggests that the person's health is under their control and therefore their responsibility (Carroll et al., 1993). One major benefit of a
'perceived control' approach is that it recognises the impact of the social context on one's 'sense of control' . 'Perceived control' suggests that a person' s 'sense of control' comes from a variety of experiences including interactions with friends and family and people in positions of authority and therefore much of the danger of victim b1aming is removed because the emphasis is not solely on the person' s experiences of success or fail ure.
'Perceived control' is dependent on the interaction between beliefs about ability and beliefs about causation (which is entailed in a 'sense of control'). For example, the belief that control is held by a 'powerful other' can produce a sense of control if it is also believed that the powerful other is benevolent. Therefore it is important that the theory adopted concerns beliefs about a 'sense of control' .
It is, also important to include as much information as possible on control beliefs. Bailis, Segall, Mahon, Chipperfield, & Dunn (2001 ), in discussing their investigation of the role of control beliefs in the relationship between SES and health, suggested that a more detailed approach to control beliefs is needed to better understand its role because the measure they used was Pearlin and Schooler' s (1978) Mastery Scale, which gives just one single score. Many measures of control beliefs are like the Mastery Scale and therefore limit the amount of information they gather. In particular Bailis et al. suggest that the conceptualisation of control outlined by Skinner ( 1996) provided potential for future gains in understanding because of the variety of information it involves. Therefore it would be
ideal for the theory to consider different aspects of control beliefs including a 'sense of control' .
Few theories are able to fulfil these criteria. A number of theories include only one aspect of control beliefs. An example of this is Pearlin and Schooler's (1978) Mastery scale which only concerns beliefs about personal contingency. Levenson' s (1972)
multidimensional locus of control does consider beliefs about multiple 'means' of control but does not consider 'sense of control' because it does not include the access to those
'means' . While the theory outlined by Shapiro and Astin (1998) considers 'sense of control' and recognises that a 'sense of control' can be gained through belief in control held by others, the Shapiro Control Inventory (Shapiro, 1994) only measures control from two sources, 'self and 'others' . Other theories such as self-efficacy theory (Bandura,
1977) and the theory of Weisz and Stipek (1982) include a distinction between beliefs about ability and causation and their interaction to produce 'sense of control' beliefs, but they come from a 'personal control' perspective only.
One theory that includes all the aspects of control reviewed in this chapter, and, in
particular, those which are relevant to an investigation of the role of control beliefs in the relationship between SES and health, is the conceptualisation of control outlined by Skinner (1995, 1996). It incorporates beliefs about ability and causation and theorises that
it is the interaction of these beliefs which forms a 'sense of control' . It also recognises that
a 'sense of control' can be gained through control being held by a variety of 'means' and that a 'sense of control' is not dependent on belief in 'personal control' . In this way it could be seen as a 'multidimensional sense of control' concept. This aspect builds on the
interaction of 'capacity beliefs' and 'strategy beliefs ' . It is argued that when there is belief
that control is held by others, a 'sense of control' can still be experienced if it is also believed that there is access to that 'means' . Measures can also be developed to measure outcomes at any level of specificity.
Unfortunately for clarity in the present situation, Skinner (1995) uses the term 'perceived control' when referring to the conceptualisation that she outlines. In the present chapter, following the work of Walker (2001), the term 'perceived control' has been used to refer to the understanding that a 'sense of control' can be experienced when it is believed that control is held by a variety of 'means' and has been distinguished from 'personal control' .
Consequently, a new term is needed to refer to the understanding of control which includes both what this chapter has termed a 'sense of control' and what it has termed
'perceived control' . It was decided to use the term 'multidimensional sense of control' . The term 'sense of control' will be retained and the word 'multidimensional' i s added because it recognises that a 'sense of control' can be gained through belief in control by a variety of 'means' .
While the conceptualisation outlined by Skinner (1995, 1996) is relatively new, Bandura (1997) has criticised the theory and it is appropriate to discuss those criticisms here. One of his major arguments is that the theory, which includes three components ('agent',
'means' and 'ends'), should include a fourth part. He suggests that this fourth part is 'performance' and that it lies between the 'means' and the 'ends' . Bandura argues that 'means' produces 'performance' and that 'performance' produces 'ends' . To illustrate this point he states it is not correct to say that effort produces royalties, but rather one should say, effort produces books, which produce royalties. There are two problems with this argument. Firstly, it raises the question of what constitutes an 'end' . Skinner ( 1 996) defines 'ends' as "the desired and undesired outcomes over which control is exerted." (p. 552). In Bandura's example, the outcome is book royalties, but the outcome could quite possibly be 'food on the table' because that is what the book royalties are used for. There are a number of different 'ends' available in every situation which vary in their degree of specificity, and 'performance' can be seen as just one of those 'ends' .
A second problem with Bandura's (1997) argument is that Bandura's own model does not include the 'performance' component which he argues should be included in Skinner's ( 1 995, 1996) model. As discussed previously in this chapter, Bandura (1977, 1986, 1997) makes a distinction between beliefs that is very similar to that made by Skinner. 'Strategy beliefs' as outlined by Skinner is very similar to Bandura's 'outcome expectancies' , and Skinner's (1995) 'capacity beliefs' is very similar to Bandura' 'self-efficacy' beliefs. Bandura (1997) even uses three very similar terms in his explanation of 'self-efficacy' beliefs and 'outcome expectancies ' . He says that 'self-efficacy' beliefs describe the relationship between the 'person' and the 'behaviour' and 'outcome expectancies'
describe the relationship between the 'behaviour' and the 'outcome'. 'Person' , 'behaviour' and 'outcome' are very similar to 'agent' , 'means' and 'ends' . The major difference is the focus of the two theories. Bandura's (1977, 1986, 1997) model focuses around 'personal
action' and so 'behaviour' is specified as the 'means' whereas the conceptualisation outlined by Skinner (1995) does not have that focus and so the 'means' is not specified. A second argument of Bandura (1997) is that the 'means' which have been included in the measures (SPOCQ [Wellbom, Connell & Skinner, 1989; CAMI [Skinner, Chapman & Baltes, 1988]) based on the conceptualisation outlined by Skinner (1995, 1996) are conceptually different. In particular, he argues that 'luck' cannot be a 'means' of control because it cannot be controlled. The problem with this argument is that it is based on a
'personal control' understanding. Bandura may be correct in saying that 'luck' cannot be controlled, but Skinner does not suggest that it can. She merely says that if people believe that 'luck' controls outcomes and that they are a lucky person then this will increase their
'sense of control' . As discussed above, Skinner's model takes the focus off the need for personal action and argues that a 'sense of control' can be felt without 'personal control' . Bandura seems to have missed this point.
Summary
In summary, a multidimensional sense of control approach was chosen a suited to investigation of the role of control in the relationship between SES and health. This approach has three key aspects. One is the differentiation between capacity beliefs and strategy beliefs. The second is the proposal that it is the combination of these two beliefs that indicates whether a person has a sense that things are under control. The third aspect is that it recognises that a 'sense of control' can be gained from a variety of 'means' and so removes the focus from 'personal control' . This approach to control has several benefits for the present investigation.