The Test Phase

 84 Applying Social Psychology

The Test Phase:

Developing and Testing

the Process Model

INTRODUCTION

Once a set of explanations has been identified and selected by a social psychologist, he or she then develops a process model. This model serves as a template for developing interventions. In this chapter, we discuss the third step of the PATH model, the Test phase. In this stage we make suggestions on how to develop the process model and how to test the empirical validity of the model.

FORMULATING A PROCESS MODEL

The explanations selected in Chapter 3 form the core of the process model. A raw ver­sion of a process model may have been developed in the previous stage already, for example, in the 'why' interview (Chapter 2). A process model is a pictorial representa­tion of the explanatory variables, and their relationships with each other and with the focal problem (see Figure 4.1, p. 86). Each variable is represented as a box. The boxes (the variables) in the model are connected via arrows. The valence of an arrow indicates whether there is a positive (+) or negative (—) relationship between two variables. Formulating a process model helps social psychologists to develop a structured account of the problem and its underlying causes. It should also give clues as to where interven­tions must be targeted in the model.

There are a number of general guidelines for developing a process model. We illus­trate these with two cases:

1.          improving the cooperation among divisions in a pharmaceutical company;

2.          reducing the risk of infection with a sexually transmitted disease (STD).

In both examples, the process model is formulated using relevant social psychological theories that are generated via the topical, conceptual, and general theory strategies that we discussed in the previous chapter.

The Test Phase 85

Case Study: Improving Cooperation in a Pharmaceutical Company

The Problem Definition

A large pharmaceutical firm in the UK spends around £5 million (8 million euros) a year on research and development. The board of directors (for whom is it a problem?) believes the returns are insufficient. The director of Research and Development (R& D) has been told to increase the number of useful ideas and products, otherwise the divi­sion will face budget cuts and job losses (why is it a problem?). The director (for whom is it a problem?) believes the problem may be due to a lack of teamwork among the sci­entists in the division's laboratories. She feels that they prefer to work on their own hobbies rather than collaborate to create commercially successful products for the com­pany (what is the problem?). In addition, she feels that there is poor communication between the scientists and the people in marketing who must judge the saleability of their products (what is the problem?). The director wonders how she could facilitate cooperation among the scientists such that the productivity of the division is raised, and they can avoid downsizing. She knows from other pharmaceutical companies that research divisions that have a more cooperative work structure tend to be more success­ful. She asks a team of applied psychologists to examine the problem and offer sugges­tions to improve team cooperation.

The Analysis

Based on interviews with the director and the scientists at the firm, the applied psycholo­gists realize that the success of the R&D division is dependent upon scientists working closely together on developing a narrow range of new products and ideas. They are also aware that this is not necessarily in each scientist's interest. Each of them would presum­ably rather work on developing their own products rather than helping to develop some­one else's products, even though the latter might be better from the company's point of view. From this analysis, the psychologists conclude that they have to find ways to resolve this conflict of interests in order to facilitate cooperation. They turn to the social psycho­logical literature for answers and find a paradigm that might be useful: the social dilemma paradigm. A social dilemma is a situation in which individuals with competing interests damage the collective interest (see Chapter 3). If many individuals pursue their narrow, selfish interests, the collective suffers and in the end everyone will be worse off. The social dilemma literature identifies many different factors that facilitate group coopera­tion (Kerr & Tindale, 2004; Komorita & Parks, 1994; Van Vugt, 1998).

The psychologists believe that the problem in R&D is essentially a social dilemma and examine the literature for clues. They find that cooperation in social dilemmas is promoted through offering selective (financial) incentives for cooperation (Van Vugt, 1998). They also find that better communication facilitates teamwork and people work better together if they identify more strongly with their team (Dawes, Van de

86 Applying Social Psychology

Nature of the task			Cooperation

Figure 4.1 A preliminary process model: What determines cooperation in this company?

Kragt & Orbell, 1988). Together, these factors promote the commitment that people feel toward their group, which, in turn, increases team cooperation and productivity (Kerr, 1989; Dovidio et al., 2006).

Developing a Process Model

How to develop a process model on the basis of these explanations? First, we must be clear about the outcome variable. What is it that we wish to influence or improve? Ideally, this has already been done in the previous steps of the PATH model, but per­haps there is a concern that the outcome variable is still not concrete enough. For exam­ple, the generic term 'cooperation' could be turned into something more specific like `willingness to help your colleagues'.

Next, we must draw up a diagram of the preliminary process model (see Figure 4.1). We can put the outcome variable in a box to the right of the figure. Subsequently, we will select one of the possible explanations for the problem. The social dilemma litera­ture suggests that the primary cause of competition among the scientists lies in the nature of the task — competition for new products and ideas — and that this inhibits coop­eration (Van Lange & De Dreu, 2001). Thus, a first sketch of the process model might look something like this figure.

This looks like a plausible first model, but there are several problems with it. First, the explanatory factor (`nature of the task') is not described as a continuous and quan­titative variable (see Chapter 3). It is unclear how this variable could vary from less to more, and therefore how exactly it is related to the outcome variable (`cooperation'). Is there a positive or negative relationship between 'nature of task' and 'cooperation'? One could draw a distinction between different kinds of tasks, for example, cooperative versus competitive tasks (Steiner, 1972), but they are still not continuous.

A second problem is that some of the variables are still too general. It is unclear how the 'nature of the task' might affect cooperation among scientists in R&D. Furthermore, as we have said earlier, the term 'cooperation' could be made more specific. A more concrete description would be 'willingness to help colleagues' because this will ulti­mately be the aim of an intervention programme.

An improved model would use more concrete terms to describe the variables. For example, the nature of the task would be translated into several specific and continuous task elements. One of them relates to the fact that the scientists compete with each other to develop the best and most useful products. This variable could be described as 'com­petition between scientists'. Furthermore, as they all have their own labs there is little coordination between the scientists, hence 'degree of coordination of task activities'

The Test Phase 87

Figure 4.2 An improved process model: What determines cooperation in this company?

may also be included in the process model. Finally, once a product idea is endorsed by the company, it is developed further by one of the labs. Thus, there is little evidence of a shared team product (`the degree of shared group product'). These are all related to the end variable, which we could describe more concretely as 'willingness to help col­leagues'. The improved process model is depicted in Figure 4.2.

The arrows depict the direction of the relationships and the valence of the relation­ships in the model could be described as well. Because there is likely to be a nega­tive relationship between 'competition' and 'willingness to help' each other, we put a minus (—) sign there. Conversely, the 'extent of shared group product' correlates posi­tively with 'willingness to help', hence the positive (+) sign. Finally, the 'degree of task coordination' should correlate positively with 'willingness to help' (+). The valence of these relationships can normally be estimated through common sense, but sometimes consulting the literature can be helpful too.

This second model is a substantial improvement yet it is still lacking in details about the process leading to the manifestation of the problem. We have now identified three task-related characteristics that might cause the problem, but we do not know whether they have a direct or indirect influence on the problem. In other words, does A cause B directly or, does A influence B via its effect upon C? For example, poor coordination of task activities presumably leads to poor communication between scientists, and this in turn affects the absence of a helping culture in R&D. The social dilemma research sug­gests that communication promotes cooperation (Van Vugt, 1998), so this is clearly an important intermediate factor. Similarly, the absence of a shared group product may lead the scientists to not feel very responsible for the activities of the research team as a whole. Thus, a 'shared group product' might influence willingness to help via an

88 Applying Social Psychology

 Figure 4.3 Improved process model 2: What factors determine cooperation within this company?

increased 'responsibility for the group product' that scientists feel. In turn, all these fac­tors may undermine team cohesion within the division, which could be the proximate cause of an unwillingness to help each other.

In addition, the variables introduced in Figure 4.2 might in turn be caused by other factors. For example, salaries tend to depend upon each scientist's personal achieve­ments (how many product ideas they have developed in their labs), which probably cre­ates a climate of excessive competition between individuals. In turn, this might affect the cohesion of the team and their willingness to help each other. Furthermore, the fact that each scientist has their own laboratory may make it difficult for them to coordinate their activities and appreciate that there is ultimately a shared group product. This leads to a much improved model, depicted in Figure 4.3.

If we work our way backwards then this model suggests that the willingness to help each other in the R&D division is undermined by poor team cohesion. This, in turn, is affected by the absence of sufficient communication between scientists and the absence of feelings of personal responsibility for the group product. These factors in return are caused, ultimately, by an excessive competition between the scientists (perhaps due to the salary system) and by the absence of task coordination and of a shared group prod­uct (perhaps due to each scientist only working in their own lab).

This is one possible way to construct a process model, but it is certainly not the only way. In general, it takes a lot of time and adjustment to come up with a process model that describes the relationships between the variables as precisely and completely as possible. It is worth it because a good process model should give a clear recommenda­tion regarding the nature of interventions necessary to tackle the problem. For example, as a result of their process analysis, the psychologists might recommend to the director that laboratories should be shared, where possible, so that scientists communicate more with each other, thereby learning to work together on each other's ideas.

The Test Phase 89

Box 4.1 Interview with Professor Michael West of the Aston Business School (UK) 'When I finished my PhD in psychology, I worked underground as a labourer for a year in a coal mine, in order to earn money to pay off student debts. Once I got over my fear of working in such a dangerous environment, I became fascinated by the teamworking and camaraderie that are necessary for safety and effectiveness in such a hostile environment. That experience shaped much of my subsequent interest in applied social and organizational psychology. Research too is about teamworking and vigorous collaboration across boundaries. 'Since my work in the coal mine, I have worked at col­leges all over the world: among others, at the University of Kent (UK), the Bermuda College (Bermuda), the University of Queensland (Australia) and the Technical University of Eindhoven (The Netherlands). Since joining Aston Business School in 1999, I have bridged many more boundaries (disciplinary and international) and feel privileged to work in the domains and with the wide variety of colleagues that academic life offers. 'In general, my work in social/organizational psychology has been applied with a particular focus on influencing organizational and public policy. I am currently running an annual survey for the English National Health Service on staff attitudes and experiences and linking the findings to outcomes such as patient mortality, patient satisfaction, errors and staff behaviours. It is very rewarding to see applied social and organizational psychological research being translated into government policies and thereby positively affecting patient care. 'For the future I foresee an important role for social psychology in the field of ageing. The ageing of the population is a major change in society and social psy­chologists must urgently engage with this currently neglected topic. In addition, globalization, international travel and migration make diversity a fundamentally important topic. And who better than social psychologists to explore and explain how we can make a strength and virtue of these changes? We must discover how we can successfully overcome the inherent pathology of inter-group prejudice which threatens our survival as a species. This requires understanding positive social processes associated with successful diversity and promulgating those far and wide.' Interested in Professor West's work? Then read, for instance,: West, M.A., Borrill, C.S. & Dawson, J.F. (2003). Leadership clarity and team inno­vation in health care. Leadership Quarterly, 14(4-5), 393-410. West, M.A. (2004) The secrets of successful team management: Now to lead a team to innovation, creativity and success. London: Duncan Baird Publishers.

90 Applying Social Psychology

Intention to use a condom		Condom use

Figure 4.4 A preliminary process model: What determines condom use?

Case Study: Reducing the Risk of STD Infection The Problem Definition

According to an estimate by UNAIDS in 2005 more than 26 million people world­wide are suffering from HIV/AIDS (see also Chapter I). In Britain alone around 80,000 people are believed to be infected with the HIV virus. There is currently no cure or remedy available for HIV/AIDS. The only feasible strategy is to engage in behaviours that reduce one's risk of contracting HIV/AIDS. The main preventative behaviour is condom use. Condom use also protects against the transmission of other, less deadly, types of sexually transmitted diseases (STDs) such as chlamydia and gonorrhoea.

One particular health authority in a coastal area of Britain (for whom is it a prob­lem?) has documented a rise in STDs (what is the problem?) among the young adult population. This is attributed to an increase in late night entertainment facilities in the city that attract a lot of youngsters to the area, particularly during the summer months (causes of the problem). It has been decided that there will be a campaign to promote safe sex practices among youngsters in the area, and the authorities decide to focus on condom use in particular. A team of health professionals and social psy­chologists is hired to set up a campaign to foster condom use and they begin by reviewing the relevant behavioural and psychological literature (see, for example, Adih & Alexander, 1999; Buunk et al., 1998; Helweg-Larsen & Collins, 1994; Schaalma, Kok & Peters, 1993; Sheeran, Abraham & Orbell, 1999; Sheeran & Taylor, 1999). From this literature, several potential causes appear for the failure to use condoms:

·         A lack of awareness about STDs, and the relation between STDs and unprotected sex.

·          People don't believe they are at risk of contracting an STD, possibly because they use other contraceptives, such as the pill.

·          People think that sex without condoms somehow 'feels better'.

·          People believe their partner is not bothered about using a condom.

·          People do not bring up condom use, possibly because they are too embarrassed or they think their partner will reject them.

·          People are too drunk to consider using a condom.

·          People do not bring condoms when they go out.

The Test Phase 91

Figure 4.5 An improved process model: What determines condom use?

·          People lack practice in putting on a condom.

·         There is no positive inclination to use a condom.

hi working towards an intervention programme to foster condom use, the psychologists decide to concentrate on two well-known attitude-behaviour models: the theory of rea­soned action (TRA; Fishbein & Ajzen, 1975) and the theory of planned behaviour (TPB; Ajzen, 1987). They specify condom use as the outcome variable in their process model. Next they select a potential cause. The main determinant of condom use, according to these theories, is a positive intention to use a condom. Hence, the prelim­inary causal model that they draw up (see Figure 4.4) contains two factors.

Of course, the main problem with this model is that it does not specify the underlying causes for not engaging in safe sex practices such as condom use. According to TRA and TPB, individuals' intentions are determined by their attitudes. The psychologists therefore first concentrate on the attitudes of youngsters towards condom use. According to TRA and TPB, the attitude is formed first by beliefs regarding the costs and benefits of not using a condom. For example, a potential cost would be susceptibility to contracting an STD, whereas a potential benefit would be that-the sex is more pleasant. The second component of the attitude is the evaluation of these beliefs, for example, how much do people care about avoiding an STD or having a perfect sexual experience? This results in a more detailed process model such as the one sketched in Figure 4.5.

This looks like a better model, but there are still several problems with it. First, it ignores the fact that to engage in sexual intercourse requires the cooperation of another person (Kashima, Gallois & McCamish, 1993). Hence, it is probably also important what the sexual partner thinks about condom use. Furthermore, the decision to use a condom is probably also influenced by the wider social environment, for example, what someone's friends think about condom use, and perhaps what someone's parents might

92 Applying Social Psychology

Figure 4.6 Improved process model 2: What determines condom use?

think. Thus, a more elaborate model would take these subjective (social) norms into account. In TRA and TPB the social norms consist of two components:

1.          the attitude of relevant others (like parents or friends) toward condom use (so-called norma­tive beliefs);

2.          a motivation to comply with the attitudes of relevant others.

Thus, a more detailed model is depicted in Figure 4.6.

In many ways, this model looks like it offers many promising directions for setting up an intervention programme. However, it still ignores the fact that although young­sters may have positive attitudes towards condom use and the social norms might be conducive, there might still be important obstacles for the desired behaviour (Schaalma, Kok & Peters, 1993; Sheeran, Abraham & Orbell, 1999). An important obstacle may be that a person feels too incapable to put on a condom, for example, because of limited experience or perhaps because of drunkenness. TPB refers to these obstacles in terms of control beliefs (also called perceived behavioural control or `self-efficacy'), namely

The Test Phase 93

Positive evaluation of beliefs (e.g., 'I am concemed about STDs')

Intention
to use a
condom

Condom use

Figure 4.7 Improved process model 3: What determines condom use?

the belief that you can do something if you want to. Thus, we should add a box to the model which specifies the sense of control that teenagers feel over their condom use. This results in a model like Figure 4.7.

We have now analysed several different factors which presumably underlie condom use. Using TRA and TPB models, we have developed a process model in which the relevant behaviour (condom use) is shaped by control beliefs and intentions, with the latter being further influenced by attitudes and social norms. Each of these factors con­tains both belief and evaluation components. We could go further by going backward

94 Applying Social Psychology

Figure 4.8 An improved process model 4: What determines condom use?

in the model and hypothesizing about factors influencing these attitudes and social norms. For example, it is likely that partners in long-term relationships will be less affected by their attitudes about unprotected sex, because for them the risk of contract­ing an STD is smaller. Furthermore, education level could increase the importance of social norms promoting condom use (Sheeran, Abraham & Orbell, 1999). So, we could add length of the relationship and educational level as distant variables in the model, interacting with more proximate psychological factors like attitudes and social norms. This is depicted by an arrow running from the reinforcing variable (for example, edu­cation level) to the arrow between two existing variables (see Figure 4.8). In addition,

The Test Phase 95

it could be argued that control beliefs exercise both an indirect effect on condom use, via shaping the intention, as well as a direct effect. For example, a positive intention towards using a condom might fail to materialize in the desired behaviour, because a person might simply have forgotten to bring a condom (Gage, 1998).

Thus, this example serves as a nice illustration of the development of a process model, because it contains four different effects:

·          Direct effects variations in one variable directly affect variations in another variable, for example, intentions affect condom use directly.

·          Indirect effects: one variable affects another variable via a third variable. For example, the belief that condom use may prevent STD affects the behaviour by adding to the intention to use a condom.

·          Reinforcing effects: one variable strengthens the impact of another variable on the behaviour. For example, education level might interact with subjective norms to produce a stronger effect on the intention to use condoms.

·          Undermining effects: one variable mitigates the relationship between two other variables. For example, being in a long-term relationship might weaken the influence of positive attitudes on the intention to use a condom.

Box 4.2 A Case Study: Losing Weight

Being overweight is a problem of great concern. Currently, about 70 per cent of adults in the USA and 45 per cent of the adults in European countries, such as the Netherlands, are overweight (see digital infobase World Health Organization, www.who.int). Individuals who are overweight are at risk of developing serious medical conditions, such as hypertension and diabetes. In addition, obesity cre­ates a major economic burden through loss of productivity and income and con­sumes about 10 per cent of overall health care budgets.

For many overweight individuals, efforts to lose weight meet with very limited suc­cess. Nonetheless, it can be done. Social psychologists Debora Schifter and Icek Ajzen from the University of Massachusetts wanted to uncover the differences between overweight individuals who succeed in losing weight and overweight indi­viduals who do not*. They based their study on the theory of planned behaviour (TRB) and asked female participants who considered themselves to be overweight to fill in two questionnaires, one initially (Time 1) and another six weeks later (Time 2). Participants answered questions about their weight and rated items about attitudes towards losing weight (for example, 'For me to reduce weight during the next six weeks is desirable/undesirable'►, a subjective norm (for example, To what extent do most people who are important to you think you should reduce weight over the next six weeks?'), intentions (for example, 'I intend to reduce weight over the next six weeks'), and control beliefs (for example, 'How likely is it that if you try you will man­age to reduce weight over the next six weeks?').

(Continued)

96 Applying Social Psychology

After six weeks, the participants filled in another questionnaire that asked about their experiences during the preceding six weeks and their weight The social psychologists found that no less than 58 per cent of the participants man­aged to lose weight during the six weeks of the study. Surprisingly, they also found that weight loss had little to do with the strength of women's intentions to lose weight (or, in other words, their motivation). A much better predictor of weight loss was the degree to which women believed they had control over their body weight. According to the social psychologists, individuals with perceived high control are more likely to try losing weight than are individuals with perceived low control, even if their perceptions are unrealistic. Those women who strongly intended to lose weight and also believed that they were capable of doing so were most likely to succeed.

Their study strongly suggests that intervention programmes for overweight individuals should not primarily focus on individuals' motivation to lose weight (for example, by pointing out the disadvantages of being overweight and the advan­tages of being slim) but instead, should try to strengthen individuals' beliefs that they can control their body weight.

Schifter, D.E. & Ajzen, I. (1985). Intention, perceived control, and weight loss: An application of the theory of planned behaviour. Journal of Personality and Social Psychology, 49, 843-851.

HEURISTICS FOR DEVELOPING A PROCESS MODEL

The above examples should help you to develop a process model, regardless of the par­ticular behaviour being studied or explanation offered. These are examples of a way of thinking that is important for applying social psychology, which involves a consider­able amount of logic and interpretation. Although there is not one particular blueprint for developing a process model, there are various heuristics that could be helpful in

organizing a process model. The following 11 rules of thumb may help develop a process model:

1.         Make a list of the possible explanations and variables involved.

2.         Ensure that all these variables are social psychological, specific (rather than general), concrete (rather than abstract) and continuous rather than binary.

3.         Ensure that these variables affect behaviours, attitudes, or motivations.

4.         Draw up the outcome variable on the right-hand side of the process model.

5.         Move from the right to the left-hand side of the model by asking yourself which variable influences the outcome variable.

6.         Draw arrows between these variables to depict the direction of the relationship (+ = positive, — = negative, ? = don't know yet).

The Test Phase 97

7.         Make sure that the relationship between the variables is not too remote. Otherwise, consider putting in a mediating variable (for instance, in the condom use example, the atti­tude is likely to affect the behaviour via the intention).

8.         When coming up with new variables think about whether they have direct effects, indirect effects, reinforcing or undermining effects on the other variables in the model.

9.         When there are several variables influencing the outcome variable, work each of them out in detail and then consider whether they are related to each other. Draw an arrow between them if you think they are.

10.       In practice, a process model should contain no more than about 10 variables (excluding the outcome variable) to make it workable. One should make sure the model is parsimonious, while at the same time providing sufficient detail about each of the explanations.

11.       To develop a practical model there should not be too many steps between the outcome vari­able and the most distal variables. Aim for about four different steps to make it manageable.

TESTING THE PROCESS MODEL

In choosing a particular theory or set of theories, the applied social psychologist will often have already looked at the empirical literature for evidence. Particularly in the issue-related approach, it is often clear from research what the strength and direction of the relationships are between the variables in the model. Hence, there may already be a solid empirical basis for the model. Meta-analytic tests and review articles are most useful because they summarize the results of various studies on a particular topic. In the condom use example, the health psychologist finds an article that contains a very helpful and recent meta-analytic review of the psychological and behavioural correlates of condom use, entitled: 'Theories of reasoned action and planned behaviour as models of condom use: A meta-analysis.' The article is published in the scientific journal Psychological Bulletin and is written by psychologist Dolores Albarracin and her colleagues. This review suggests the correlations between the different variables shown in the model (see Figure 4.9).

It can be seen from this model that the intention to use a condom has a moderately strong positive influence on condom use (r = .45). Attitudes, subjective norms and con­trol beliefs all have a moderately strong positive influence on the intention to use a con­dom (r = .58, r = .39, and r = .45 respectively). In a similar vein, attitudes are positively influenced by beliefs about condom use (r = .56) and subjective norms by normative beliefs (r = .46). Control beliefs also show a direct positive influence on condom use (r = .25). However, the article also shows that, after controlling for intentions, the direct influence of control beliefs on condom use becomes very small. That is, the influence of control beliefs on condom use can be explained almost entirely through their effect on intentions.

The psychologist may therefore decide to remove the arrow between control beliefs and condom use from the model. He may also search the literature for articles that do report evidence of a significant direct effect of control beliefs on condom use. If the psychologist finds such an article he should take a close look at the background vari­ables of the study, such as the age and educational level of the participants. If these

98 Applying Social Psychology

Figure 4.9 Empirical support for the process model

background variables are comparable with the characteristics of the situation and pop­ulation the social psychologist aims to influence, he might decide to keep the arrow in the model, despite the negative conclusion of the review article. In a similar vein the social psychologist should try to find empirical support in the literature for all of the relationships in his process model.

Thus, it is important to examine the relationships between the model variables by con­sulting the empirical literature. This means that for every relationship in the model one must specify the strength and direction of the relationship. Obviously, the variables with the strongest influence are the most useful ones for an intervention, because a change in that variable is likely to lead to a significant change in the variables it influences.

How does one find relevant research? In seeking to find literature one could use one of the three approaches that we outlined in Chapter 3 to finding explanations: the top­ical, the conceptual, and the general theoretical approach. It is generally advisable to search for research that is directly relevant to the problem definition, thus the issue-related approach. The reason is that the relevance of the findings diminishes the greater the difference between the research and the problem the social psychologist is

The Test Phase 99

interested in. Although there are various literature search programmes available (like PsycINFO or Google scholar), it may not be easy to find all the available evidence. First, not all research is published and this is more likely for null effects, that is, research that supports the null hypothesis of no difference. It is therefore often useful to send a message to an email list of researchers in a particular field (like the SPSP or EAESP mailing lists for social psychologists).

Second, not all the published research is of the same relevance and quality so it is important to carefully read each article and judge their quality and relevance. Third, research articles may contradict each other in their results or conclusions. One's best bet is often a meta-analytic review of a research domain, which summarizes the avail­able evidence and estimates the strengths of a particular effect across a broad domain of studies (as in the condom use example). Such meta-analytic reviews appear in jour­nals like Psychological Bulletin, Personality and Social Psychology Review, Health Psychology, and Health Education Quarterly, and it is worth looking out for them.

Even if there is no research available on the topic of interest then it may still be useful to examine the literature. For example, a social psychologist may be interested in organizational citizenship behaviour, but because there is not yet a lot of research available on it, he may look instead at the broader helping literature for clues (Penner, Dovidio, Pilavin & Schroeder, 2005). But even if a conceptual approach does not work then a general theory approach might be helpful. For example, research on the Theory of Reasoned Action (TRA) or the Theory of Planned Behaviour (TPB) might be help­ful even when the behaviours which are of interest are entirely different. For example, the knowledge that subjective norms are an important influence on smoking decisions, might help a social psychologist in thinking about what factors might influence recy­cling. Similarly, a general evolutionary psychological theory like reciprocal altruism might give an applied psychologist the suggestion that stress and burn-out in the work­place may be caused by feelings of inequity and unfairness (Buunk & Schaufeli, 1999).

DOING YOUR OWN RESEARCH

Unfortunately, not all the relationships in the model will have been documented in the research literature. For example, there simply may not be any existing research on the relationship between alcohol intake and condom use. Thus, it is important that, if nec­essary, social psychologists are able to do their own research to find answers. This book is not the place to discuss in any great detail the research methods needed for doing applied social psychological research. Each person studying for a degree in social psy­chology will have a solid training in research methods in psychology. And there are many good texts available on conducting social psychological research (see, for exam­ple, Breakwell, 2004; Reis & Judd, 2000).

The importance of doing research in answering an applied social psychological problem cannot be overstressed. Often, there is empirical literature on a particular topic but the research may not have been adequately done. Or else the research has tested a particular relationship in a slightly different domain than the one the social psychologist is interested in. For example, a psychologist interested in the determinants

100 Applying Social Psychology

of using contraceptives cannot simply rely on the research on condom use. Nevertheless there may be important clues to be found in this literature. In addition, many field stud­ies contain cross-sectional data (all gathered at the same time) and the resultant corre­lation between two variables may not indicate a causal relationship whereby one variable causes another. One must therefore be cautious in simply applying the results of other studies to the problem under investigation.

Furthermore, many studies in social psychology are laboratory-based, primarily with students as research participants. These studies are high in internal validity (that is, the results are reliable), but they are sometimes lacking in external validity (that is, the results may not be applicable to other situations and/or populations). Hence, one should consider the possibility that there are limitations in generalizing the results to apply to the real world. For example, some people might take task risks in the laboratory, gambling with small sums of money, but does it mean that they also take risks in real-world tasks like drug use, not wearing seatbelts or having unprotected sex? Even if results have been obtained in field research there is often the question of whether data can be generalized from one popula­tion to another. This is particularly important in cross-cultural research. Research suggests that American people differ from Asian people in the way they respond to social psychol­ogy experiments. American students, for example, see themselves as more independent whereas Japanese students see themselves as more interdependent (Kitayama, Markus, Matsumoto & Norasakkunkit, 1997). This difference might be relevant in finding strate­gies to promote, for instance, cooperation in the workplace or at school.

In conducting research to establish the validity of a process model, it can be useful to conduct a quick survey. In survey research, the main variables are operationalized and measured and then through statistical procedures the researcher determines the relationship between the variables. Whenever possible, researchers use existing scales of variables that have been tested in previous research and reported in the psychological literature. An example of a commonly used scale is Rosenberg's self-esteem scale (containing 10 items) which is depicted in Box 4.3. This can be used in a school, sports, or work environment.

Box 4.3 Rosenberg's Self-Esteem Scale (Rosenberg, 1965)

Below is a list of statements dealing with your general feelings about yourself. If you strongly agree, circle SA. If you agree with the statement, circle A. If you dis­agree, circle D. If you strongly disagree, circle SD.

1. On the whole, I am satisfied with myself.	SA	A	D	SD
2.* At times, I think I am no good at all.	SA	A	D	SD
3.              I feel that I have a number of good qualities.	SA	A	D	SD
4.              I am able to do things as well as most other people.	SA	A	D	SD
5.* I feel I do not have much to be proud of.	SA	A	D	SD
6.* I certainly feel useless at times.	SA	A	D	SD
7. I feel that I'm a person of worth, at least on an equal plane with others.	SA	A	D	SD

The Test Phase 101

8.* I wish I could have more respect for myself.	SA	A	0	SD
9.* All in all, I am inclined to feel that I am a failure.	SA	A	D	SD
10. I take a positive attitude toward myself.	SA	A	0	SD

Scoring: SA = 3, A = 2, D = 1, SD = 0. Items with an asterisk are reverse scored, that is, SA = 0, A = 1, D = 2, SD = 3. Sum the scores for the 10 items. The higher the score, the higher the self-esteem.

Another example of a commonly used scale is the PANAS (Watson, Clark & Tellegen, 1988), which contains 20 items and measures one's emotional state. Normally, researchers first determine the reliability of a scale using a Cronbach's alpha index that may run from 0 to 1. In social research a Cronbach's alpha of .70 or higher is consid­ered acceptable. It means that the items on the scale all measure the same underlying construct, and that therefore the scale is reliable. The social psychologist may then use a median split to categorize the sample into those with a low level and those with a high level of the trait under investigation in order to compare the two groups on a dependent variable. For example, when a social psychologist wants to know whether individuals with low self-esteem experience different emotions than individuals with high self-esteem, she could ask individuals to fill in a questionnaire that assesses their self-esteem (Box 4.3) and their emotions on a certain day (Box 4.4). After having estab­lished that the Cronbach alphas for the two scales are satisfactory, she could then divide the group of participants in half: individuals with low self-esteem and individuals with high self-esteem. With a t-test or analysis of variance she could calculate to what extent individuals with low self-esteem experience different emotions than individuals with high self-esteem.

  Box 4.4 The Positive and Negative Affect Schedule (PANAS;
Watson, Clark & Tellegen, 1988)

This scale consists of 20 words that describe different feelings and emotions. Read each item and then mark the appropriate number in the space next to that word. Indicate to what extent you have felt this way today.*

1 = very slightly or not at all; 2 = a little; 3 = moderately; 4 = quite a bit 5 = extremely.

_______  interested

_______  distressed

excited

102 Applying Social Psychology

_________ upset   inspired ________ strong ________________ nervous ___________ guilty   determined scared   attentive ________  hostile   jittery enthusiastic                                      active __________ proud   afraid * The PANAS can also be used with the following time instructions: Indicate to what extent: ·           you feel this way right now, that is at the present moment; ·           you have felt this way during the past few days; ·           you have felt this way during the past week, ·           you have felt this way during the past few weeks; ·           you have felt this way during the past year, ·           you generally feel this way, that is, how you feel on average.

It is possible that there are no scales available that directly measure what the social psychologist is interested in, for example, a scale on attitudes toward terrorism. In this case, psychologists must develop their own scale based on a careful problem analy­sis. Constructing a reliable scale is not as easy as it seems. However, this book is not the place to discuss the way to construct a reliable scale. To learn more about scale construction, you may have to take an (advanced) course in methodology and/or read a good book about scale construction (see, for example, Introduction to test construction in the social and behavioural sciences: A practical guide, by Fishman & Galguera,

2003).

Developing an empirically based process model includes the following steps: Box 4.5 The Test Phase: Developing and Testing a Process Model

1.     Develop a process model and work your way from the outcome variable on the rightto the causal variables on the left; picture the process model in a diagram.

2.     Determine whether a particular variable has a direct, indirect, reinforcing or undermining effect on the outcome variable.

3.     Limit the number of boxes (variables) in the model to about 10.

4.     Don't take more than four steps back in the model.

5.     Determine the direction of the relationship between two variables by using + or — signs.

6.     Use, as much as possible, concrete and continuous variables.

7.     Determine the empirical relationships between variables either by reviewing the empirical literature or by doing your own research.

The Test Phase 103

SUGGESTED FURTHER READING

Albarracin, D., Johnson, B.T. & Fishbein, M. (2001). Theories of reasoned action and planned behaviour as models of condom use: A meta-analysis. Psychological Bulletin, 127( 1), 142-161.

Buunk, B.P. & Schaufeli, W.B. (1999). Reciprocity in interpersonal relationships: An evolution­ary perspective on its importance for health. In W. Stroebe & M. Hewstone (Eds.), European Review of Social Psychology (Vol. 10, pp. 260-291).

Fishman, J.A. & Galguera, T. (2003). Introduction to test construction in the social and behav­ioural sciences: A practical guide. Lanham: Rowman & Littlefield.

Van Vugt, M. (1998). The conflicts of modern society. Psychologist, 11(6), 289-292.

104 Applying Social Psychology

ASSIGNMENT 4

(a) Make a list of the variables you selected in Assignment 3 (Chapter 3) and develop a process model. Remember to:

·                 Work your way from the outcome variable on the right to the causal variables on the left

·                 Ask yourself whether each variable has a direct, indirect, reinforcing or undermining effect on the outcome variable, and determine the direction of the relationship between two variables by using + or — signs.

·                 Ensure that you limit the number of variables to about 10 and don't take more than four steps back in the model.

·                 Ensure that the model does not consist of two or more separate processes.

(b) Read the following articles in detail and examine the empirical validity of the rela­tionships in the process model. If necessary, examine what causes have to be added to the model.

Caughlin, J.P. & Huston, T.L. (2002). A contextual analysis of the association between demand/withdraw and marital satisfaction. Personal Relationships, 9(1), 95-119.

Caughlin, J.P. & Vangelisti, A.L. (2000). An individual difference explanation of why mar­ried couples engage in the demand/withdraw pattern of conflict. Journal of Social and Personal Relationships, 17(4-5), 523-551.