THE FOUNDATION OF 'PARTNERSHIP'? UNION EFFECTS ON EMPLOYEE TRUST IN MANAGEMENT.

Bryson, Alex

Alex Bryson [*]

This article addresses the question: what impact do trade unions have on employee trust in management? Using nationally representative data on employees in the British Social Attitudes Survey (BSAS) 1998, we assess associations between measures of unionisation and employee perceptions that managers usually keep their promises to employees. The article provides broad support for the three hypotheses explored here. First, employee trust in managers is higher where there is a balance of power between unions and management at the workplace. Secondly, employees' trust in managers is higher where management supports union membership, and is lowest where management actively discourages membership. This is equally true among employees in unionised and non-unionised workplaces. Thirdly, employees' perceptions of union effectiveness are positively associated with higher trust in management.

The positive association between managerial support for union membership and employee trust in management can be interpreted in various ways. But the finding suggests that management can influence employee perceptions of them for better or for worse through their engagement with unions.

Management can also foster a high trust relationship with employees by ensuring that unions have sufficient power to make a positive contribution to the running of the workplace, since unions with the 'right amount of power' are associated with higher trust in management. Weak unions are particularly bad for employee perceptions of management, raising the question why employers often recognise unions for pay bargaining, but preside over a decline in union strength at the workplace.

Unions are best able to create a climate in which employees trust management where they are perceived to be doing their job well, where they have regard to union members' problems and complaints, and where they contribute to the smooth

running of the workplace. In these circumstances, trust in management is generally no different from that found among employees in non-unionised workplaces. However, where unions are viewed as ineffective, trust in management is lower than in comparable non-unionised workplaces.

Introduction

This article addresses the question: what impact do trade unions have on employees' trust in their managers? Using nationally representative data on employees in the British Social Attitudes Survey (BSAS) 1998, we assess associations between measures of unionisation and employee perceptions of whether managers keep their promises to employees.

Until recently, most analyses of attitudes towards workplace governance in Britain were based on managerial perceptions (Fernie, Metcalf and Woodland, 1994; Fernie and Metcalf, 1995). This began to change in the late 1990s, with analyses of BSAS (Bryson and McKay, 1997) and the Workplace Employee Relations Survey 1998 (WERS 98) (Gully et al., 1999; Bryson, 2000). However, relatively little is known about the impact of unions on employee perceptions of management.

The second motivation for the article is to provide a context in which to appraise the impact of the new statutory procedure for union recognition contained in the Employment Relations Act 1999, which came into effect in June 2000. Making the case for the procedure, the government argued that harmonious employee relations based on partnership between workers and their employer improve both the working lives of individuals and the performance of organisations (DTI, 1998). It also argued that these benefits could be jeopardised if, against the wishes of their employees, employers refuse to recognise a union for bargaining. As our data were obtained at a time when employers were still at liberty to decide whether or not to recognise unions, it is not possible to infer directly from our results what impact the new statutory provisions may have on employee trust in management. Even if we were to find that the presence of recognised trade unions was associated with a high-trust relationship, we would have to acknow ledge that forcing employers to recognise unions against their wishes could well sour employee relations rather than improve them, as some have argued (Confederation of British Industry, 1998). However, our analysis can shed light on two issues that are fundamental to any consideration of the links between unions and workplace governance, irrespective of the statutory environment. The first is whether the presence of recognised unions has a beneficial effect on workplace governance and, if so, under what circumstances. Evidence has emerged recently that the influence of unions is diminishing even where, at least nominally, they continue to be granted recognition rights. There are theoretical grounds for suspecting that this development will adversely affect employee trust in management. The second issue we can illuminate is whether the impact of unions depends on how management reacts to them. We contend that a high-trust relationship can be fashioned only with the support of management and employees: it is n ot simply a gift to be bestowed by one side or the other, no matter how willing they may be.

The rest of the article is set out as follows. The following section discusses trends in unionisation and employer and employee orientations to unions that may have a bearing on how employees perceive management. It then outlines theoretical links between unions and trust in management and the three hypotheses tested in the article. This is followed by an introduction of the data used in our analyses. The fourth section discusses our analytical approach. Results are presented in the fifth section and the sixth section concludes.

Theoretical links between aspects of unionisation and employee trust in management

Background

Since the early 1980s, trade union representation has shrunk owing to falls in union membership and a rapid drop in the number of employers recognising unions for collective bargaining (Millward et al., 2000). But the issue at the heart of this article is the influence unions have where they retain a foothold in the workplace. We focus on three aspects of unionisation which have a bearing on union effects on trust in management. These are union strength, employee assessments of managerial support for unions, and the effectiveness of unions as perceived by employees.

Union strength: Unions' influence in the workplace derives from their bargaining power, stemming from their ability to disrupt the supply of labour in pursuance of their members' interests. But it also comes from the union's role as the representative 'voice' of employees in the resolution of workplace grievances and disputes (Freeman and Medoff, 1984). Both sources of influence depend on the credibility of the union in claiming to represent the workforce. This has diminished since the 1980s because, even where unions continue to be recognised for bargaining purposes, there has been a decline in the proportion of employees whose terms and conditions are set by collective bargaining and the proportion who are union members (Millward et al., 2000). Certainly, employees perceive union power at the workplace to have declined (Table 1). This may explain the absence of a general union wage mark-up (Forth and Millward, 2000) and, by the late 1990s, the disappearance of negative union effects on workplace financial p erformance usually attributed to the monopoly power of unions (Addison and Belfield, 2000; Bryson and Wilkinson, 2001).

However, the decline in these 'average' union effects for some workplace outcomes is by no means the whole story. First, some average union effects remain powerful. For instance, unionised workplaces had slower growth rates than non-unionised workplaces in the 1990s, ceteris paribus, suggesting that union effects are not benign (Bryson, 2001a). Secondly, many workplaces still have 'strong' unions with high membership, high bargaining coverage and on-site lay representation (Millward et al., 2000, pp. 179-83). One might expect the 'returns' to well-organised, or strong, unions to be relatively greater now that average union strength has declined. Forth and Millward's (2000) finding that the union wage premium in the late 1990s was confined to workplaces with high bargaining coverage is consistent with this proposition. Thirdly, weak unions may still have appreciable effects on employee perceptions of management (see below).

What has caused the general decline in the 'take-up' of unions among employees and employers lies beyond the scope of this article. [1] However, employer support for unions and employee perceptions of the job done by unions may be important in explaining the effects of unions on employee trust in management. We discuss these in turn.

Managerial attitudes to union membership: Where employers are at liberty to choose whether they recognise trade unions, unions are heavily reliant on the support, or at least acquiescence, of management to conduct their business in representing members. This was the case in 1998. [2] However, BSAS evidence supports the finding from other surveys (Gallie et al., 1998, pp. 107; Millward et al., 2000, pp. 145-9) that employer endorsement of union membership among its employees has fallen since the 1980s (Table 2).

Interestingly, Table 2 shows that the decline in managerial support for union membership was confined to unionised workplaces. This is consistent with research showing that a clear majority of managers in workplaces recognising unions express a preference for consulting directly with employees rather than with unions (Bryson, 2001b), and that recognised unions are often by-passed in managerial decision-making (Cully et al., 1999, p. 105). These considerations may lead us to suspect that managers are less constrained than they were in the 1980s in pursuing corporate goals, sometimes at the expense of employees. Consequently, other things being equal, unions may be less effective as a constraint on management than they were in the past.

Employee perceptions of union effectiveness: The declining allegiance of employees to unions is amply illustrated by the decline in membership since the early 1980s. Could it be that membership levels have fallen as employees perceive unions as increasingly ineffectual? This might not be surprising given the evidence presented above on managerial attitudes to unions in the 1990s. In fact, the evidence is that employees do not regard unions as increasingly ineffectual. As Table 3 indicates, among employees in unionised workplaces, there has been no decline in perceptions of union effectiveness as measured by whether employees think the union is doing its job well or not.

Hypotheses

Union effects on employee trust in management are uncertain, a priori, because they may have offsetting influences arising from their function as a 'check' on management, on the one hand, and their ability to heighten workers' awareness of management limitations, on the other.

Unions may positively influence employees' trust in management to the extent that they operate as an effective check on management, making it more accountable to its employees for the actions it takes. Aware that employees will seek redress through the union against arbitrary or unfair managerial behaviour, employers are more likely to adopt formal procedures (Millward et al., 2000, pp. 156-7) and to abide by them. As a consequence, employees may view their management as more responsive and less likely to take advantage of them where unions are present. However, unionised workers express greater dissatisfaction with management than non-unionised workers (Freeman and Medoff, 1984; Gallie et a!., 1998; Bryson 2000). Freeman and Medoff offer an explanation for this in the greater politicisation of unionised workers. They suggest that unionised workers are more prone to express their voice 'loudly' to ensure that it is heard, resulting in 'voice-induced complaining' (1984, p. 142) which they distinguish from 'tru e' dissatisfaction. They also suggest that 'some of the critical attitude of union workers is due to their greater awareness of problems and willingness to speak out' (1984, p. 142). As Gallie et al. (1998, pp. 113-14) point out: 'unionism as an oppositional form of representation may highlight organisational inefficiencies and colour perceptions of management competence'. In addition, as Freeman and Medoff note (1984, p. 141), ceteris paribus, the stock of dissatisfied workers will be greater in Unionised workplaces because dissatisfied workers are less likely to quit unionised workplaces than they are to quit non-unionised workplaces (Bryson and McKay, 1997).

This discussion suggests that the average effect of unionisation on employee trust in management is indeterminate, a priori. However, the direction of the effect is likely to depend, in part, on union strength, management attitudes to union membership, and union effectiveness. Below we outline three hypotheses linking these three dimensions of unionisation to employee trust in management.

First hypothesis: Employees' trust in management is higher where there is a balance of power between unions and management at the workplace.

Unions will only be able to operate as an effective check on management where they have sufficient authority in the workplace to act as a legitimate representative of workers. Unions need to be 'strong enough' to influence employer perceptions of the union's legitimacy in representing workers' interests, predisposing them to take greater account of what the union is saying. This may explain why employers are less likely to listen to the union if only a minority of employees back it than if the union represents a majority voice -- even where the employer has chosen to recognise the union (Cully et al., 1999, pp. 105-6).

A high trust relationship seems unlikely if employees have no access to independent sources of power to represent them and protect their interests. Marshall (1992) argues that co-operative relations cannot be maintained where there is a substantial power imbalance between management and unions because the stronger party will opt for unilateral control over co-operation. Where unions are weak, and employers are tempted to exercise unilateral control, employees may be less trusting in management. Where the union is particularly strong, and an oppositional form of representation is in place, trust in management may also be low. In short, we might anticipate that trust in management will be highest where there is a balance of power between unions and management.

Second hypothesis: Employees' trust in management is higher where management actively supports union membership and is lowest where management actively discourages membership.

There is little reason to believe that unions can deliver high-trust employee relations alone. What management says and does is likely to matter just as much. The acts or omissions of one party may be able to undermine trust, but no matter how constructive a union wishes to be, or how strong it may be organisationally, a high-trust relationship is likely to require that management works together with the union, and vice versa. Thus, a co-operative environment is likely to require that management engage constructively with the union, unless it can devise non-union employee involvement strategies that reduce the perceived need for a union. Managerial support for a union, strong or otherwise, may signal employer interest in the concerns of workers, a signal which may lead to more positive attitudes to management.

Consequently, we hypothesise that employees' trust in management is best where management actively supports union membership, and is worst where management actively discourages membership.

Third hypothesis: Employees' perceptions of union effectiveness are positively associated with employee trust in management.

The perception that a union is effectively protecting and advancing its members' interests can result in positive perceptions of management (Deery et al., 1995). This may occur where union instrumentality [3] engenders greater employee allegiance to both the union and the employing organisation ('dual commitment'), thus resulting in more co-operative and harmonious management--employee relations. It may also occur where perceptions of union effectiveness are associated with perceptions of a fairer, more challenging and satisfying work environment. This, in turn, can positively influence perceptions of management (Deery et al., 1999, p. 546).

Gallie et al. (1998, pp. 72-86) find that employees perceive supervision to be tighter, and technical and bureaucratic methods of management control to be more evident where unions are perceived as having greater influence. The authors suggest that 'a reasonable inference, then, is that intensive control systems were preferred by organisations where managerial power was contested' (Gallie et al., 1998, p. 85). It may be that, where unions contest 'the terrain' with management, employee perceptions of the working environment actually deteriorate, in which case trust in management may also deteriorate. This line of reasoning cautions against a simple assumption that effective unionism will translate into a high trust relationship with management. Nevertheless, we hypothesise that employees' perceptions of union effectiveness are associated with higher trust in management.

Data

Our analysis is based on the British Social Attitudes Survey 1998. The survey yields a representative sample of adults aged 18 or over living in private households in Great Britain (see Jowell et al., 1999 for details). Analysis is restricted to employees working at least ten hours per week, a cut-off used to filter respondents on questions relevant to employees. With weighting to account for complex survey design, survey results can be generalised with confidence to the population of employees in Britain working at least ten hours per week.

Measuring employee trust in management

Our indicator of employee trust in management is based on a question asked of employees completing a self-completion questionnaire in the 1998 BSAS. It is based on responses to the statement: 'Managers at my workplace usually keep their promises to the employees'. Almost half of employees agreed with this statement, but they were less likely to do so if they worked in a unionised workplace (Table 4).

Measures of trade unionism

Union strength: Employees were asked whether there was a union or staff association recognised 'for negotiating pay and conditions of employment' at their workplace. We can also identify instances where unions were present but not recognised, since those who said there was no recognised union present were asked:

Are there any trade unions or staff associations that are active at your workplace?

We would expect recognised unions to be stronger than non-recognised unions because recognition is a clear indication that the union has a formal role in representing workers in pay and conditions negotiations, a role that has been legitimised by the employer. In addition, there is a very strong association between union recognition and union density, itself an important indicator of union strength and influence (Cully et al., 1999).

Another indicator of union strength is the presence of a worker representative at the workplace. Those in workplaces with unions or staff associations were asked:

Do you know who the (union/staff association) representative is at your workplace or is there not one based where you work?

Similarly, those in non-unionised workplaces were asked whether there were any other worker or staff representatives at their workplace and, if so, whether they knew whom the representative was. The presence or otherwise of a representative is a good proxy for the organisational strength of a union on the ground. There is an association between the presence of an on-site representative and higher union density. Moreover, the presence of a representative indicates at least some organisational capacity on the ground. We can therefore assume that, on average, union strength is greater in workplaces with an on-site representative than it is in those without an on-site representative. Furthermore, WERS 98 provides evidence that worker representatives are attaching increasing importance to 'dealing with problems raised by the treatment of employees by management, and to resolving disputes' (Cully et al., 1999, p. 201), as opposed to the more 'traditional' activities of maintaining or enhancing wages and benefits. I f they are effective in this role, we might expect the presence of representatives on-site to contribute to higher trust in management.

Our third indicator of union strength is the subjective measure presented in Table 1. For analysis purposes, the categories 'far too much power' and 'too much power' are collapsed into one, as are the categories 'too little' and 'far too little power'.

Management support for union membership: Our measure of management support for union membership is the one featured in Table 2.

Analysts have frequently used management support for union membership as a measure of union strength, but union strength and management support for unions are conceptually different. A union may be strong without management support. Where it is strong in the face of management opposition, trust in management may be low. Where it is strong and has management support, one would expect this to be conducive to a high trust relationship.

Union effectiveness: We use three measures of union effectiveness based on questions asked of employees working in workplaces with unions or staff associations present. Our main measure is employee perceptions as to whether the union is doing its job well, presented in Table 3 above. The second measure is whether respondents agreed or disagreed that 'union(s)/staff association(s) at your workplace take notice of members' problems and complaints'. Our third measure of union effectiveness is the extent to which workers agree or disagree with the statement that their union or staff association 'makes things run more smoothly at work'. Rather than a measure of the extent to which unions deliver for individual workers, this question is an indication of another mechanism by which unions might effect a higher trust relationship, namely by simply contributing to the better running of the workplace. Two-fifths of workers agree or strongly agree that unions make things run more smoothly, with union members more likely to say so than non-members (Bryson, 2001b).

Individual union membership: It is important to control for individual union membership status since we wish to distinguish perceptions of management associated with individual union membership from 'workplace effects' associated with the unionisation of the workplace. We identify union members by responses to the question: 'Are you now a member of a trade union or staff association?'

Control variables

Our analyses control for a wide range of individual, job and workplace-related characteristics to minimise estimation bias arising from omitted variables. We introduce these controls below, with the exception of the union variables discussed above.

Demographic characteristics of respondents: our analyses incorporate gender, age and ethnicity, all of which have been associated with employee perceptions of management (Bryson and McKay, 1997; Gallie, et al., 1998; Bryson and Wilkinson, 2001). We include qualifications because more highly educated workers often have higher expectations of involvement, and may therefore be particularly critical of management where participation is denied.

Job-related characteristics: we control for three aspects of individuals' jobs: their occupational class (based on the Hope--Goldthorpe schema); whether or not they were working part-time; and years working continuously with the current employer to capture individuals' attachment to their workplace. [4]

Workplace-related characteristics: One difficulty in relying wholly on employee data gathered as part of a general attitude survey is that information on workplace characteristics is necessarily limited. Nevertheless, we are able to control for industrial sector, ownership and workplace size, all of which have been significantly associated with employee perceptions of workplace governance in recent studies (Bryson and Wilkinson, 2001; Bryson, 2000).

Attitudes: Some have argued that 'attitudes to work and to trade unions can be formed from experiences both inside and outside the workplace' (Deery and Walsh, 1999, p. 263) and that patterns in employees' attitudes may be attributed, not only to aspects of their employment relationship, but to 'differences in personal underlying values' (Cully et al., 1999, p. 165). Some of these experiences may be captured, in part, by demographic characteristics. However, BSAS also contains direct measures of the value set that employees bring with them to work. We test the sensitivity of our results to the incorporation of this value set using the BSAS left--right index. The index, based on general attitudes to distributive justice, helps control for differences across individuals which are not directly related to their work experiences, but which may affect their orientation to work. [5]

Research has established strong associations between individuals' attitudes towards aspects of their own job -- pay, job security, their say at work -- and perceptions of governance at their workplace (Bryson and McKay, 1997; Deery et a!., 1995; Dastmalchian et al., 1989). Those viewing their own position more favourably are likely to view their working environment more favourably. However, although analysts frequently include individuals' perceptions of their own jobs in models estimating perceptions of workplace governance, it is not possible to discern the direction of causation in cross-sectional analyses. So it is not really possible to 'explain' trust in management by reference to employees' satisfaction with their own situation (although the latter may explain some of the variance in the former). We therefore omit employee attitudes towards their own work situation from our analyses. [6] However, it seems reasonable to link employee perceptions of general conditions at their workplace with their percep tions of management. In particular, there is a well-developed literature linking perceptions of distributive justice at the workplace with perceptions of the employee relations. BSAS contains a useful question on distributive justice at the workplace: 'Thinking of the highest and lowest paid people at your place of work, how would you describe the gap between their pay, as far as you know?' [7] We test the sensitivity of our results to the inclusion of this variable.

Region: We group data from the ten standard regions into seven regions, separately identifying Greater London. These identify where the employee lives, rather than where the workplace is located.

Analysis

The sample

Our sample are employees working ten hours or more taken from BSAS 1998. Others have confined their analyses of employees' perceptions of workplace governance to non-managerial employees, perhaps because managers are overwhelmingly positive in their assessment of governance as they lie on one side of the management--employee line, while non-managerial employees lie on the other (Gully et al., 1999, pp. 276-83). We adopt an alternative approach, analysing the perceptions of all employees with non-missing data. After all, most managers experience 'being managed' or supervised. Our models include occupational controls to account for more positive perceptions further up the occupational hierarchy.

Modelling procedures

Our outcome variable is a categorical indicator defined in terms of ordered responses. We use an ordered probit estimator to model the relationship between employee trust in management and sets of independent variables. In ordered probits, an underlying unobservable score is estimated as a linear function of the independent variables and a set of unknown 'threshold' parameters, or cut points. The probability of observing outcome i corresponds to the probability that the estimated linear function plus random error is within the range of the cut points estimated for the outcome.

All models are run on data weighted by the inverse of the respondent's sampling probability. As well as allowing the results to be generalised to the population from which the sample is drawn, the use of probability weights also guards against estimation bias which can arise through differential sample selection probabilities (Skinner, 1997). We employ the Huber--White robust variance estimator that produces consistent standard errors in the presence of heteroscedasticity.

Results

It is plausible that, by holding employers accountable to their employees for the actions they take, unions may increase the likelihood of employees agreeing that 'managers at my workplace usually keep their promises to the employees'. In fact, descriptive analyses show employees in workplaces recognising unions for pay bargaining were less likely to agree with this statement than employees in non-unionised workplaces (Table 4 above). Of course, there may be many reasons for this association. It may simply reflect value sets of unionised workers, who may be more critical of management per se, regardless of their own workplace circumstances. It may arise through the politicisation of unionised workers resulting in 'voice-induced complaining', or a heightened awareness of managerial shortcomings due to the information available to unionised workers. It may arise because longer workplace tenure, associated with greater dissatisfaction at work, is associated with unionisation. All of these possible causes were di scussed earlier, and we seek to account for them in our multivariate analyses. It is also possible that the association arises because unions find a foothold among poorer employers where bad employment practices provide unions with a basis for organisation. We control for a range of employer characteristics, but we cannot ascertain whether poor management came before unionisation or not. We can only point to the relative durability of unionisation and assume that our union measures pre-date employee perceptions of workplace governance at the time of interview.

To establish whether there is a truly independent association between unionisation and perceptions that management keep their promises to employees, we ran ordered probits on the measure described in Table 4. Our dependent variable runs from 0 ('strongly disagree') to 4 ('strongly agree') so that negative coefficients indicate stronger disagreement with the statement that management 'usually keep their promises to employees'.

Table 5 presents six models for the whole economy. Without controls, union recognition was associated with an increased perception that managers do not keep their promises (Model (1)). Some of the effect is associated with the presence of on-site lay representatives (Model (2)). The effect is confined to recognised unions, as opposed to recognised staff associations (Model (3)). However, the effect is not significant with the inclusion of control variables (Model (4)). As anticipated, union members were less likely to trust managers than nonmembers (Model (5)). Model (6) distinguishes between union members in workplaces with and without union recognition. Compared with the reference category, non-members in non-unionised workplaces, members in recognised workplaces were less likely to think managers kept their promises, but the effect was only weakly significant. However, there was a clear membership effect among employees working in unionised workplaces: union members were less likely to think that managers kept their promises. [8]

When added to Model (5) in Table 5, the left-right index was positive and significant (0.27, t = 4.10), indicating that those who were more 'rightward' leaning were more inclined to trust managers. Furthermore, its introduction reduced the union membership coefficient to insignificance. When interacted with union membership, the dummy variable identifying low scorers on the index was negative and significant (-0.25, t = 2.76), whereas the union main effect and interaction were not significant. This is strong evidence that employees' trust in management was accounted for in part by their broader perspectives on distributive justice, and not purely through their work experiences. Nevertheless, perceptions of distributive justice at the workplace also mattered, since those viewing the gap between the lowest and highest paid as 'too big' or 'much too big' were less inclined to trust management to keep their promises. [9]

Running separate analyses for employees in the private and public sectors shows membership and workplace effects differed across employees in the two sectors. In the private sector, perceptions that management kept their promises were not associated with features of worker representation. But union members were less likely than non-members to think management kept their promises, whether they worked in a workplace with a recognised union or not. In the public sector, membership effects were absent, but trust in management was lower in the presence of a recognised union, and particularly low where there was union recognition but no on-site representative. [10]

We can conclude from this analysis that perceptions that managers keep their promises were strongly influenced by employees' value sets, as measured by the left-right scale, and that this goes some way to explaining the membership effect identified. Workplace unionisation was associated with a more critical attitude towards management, but only in public sector workplaces where employees had no on-site worker representative.

Union power and employee trust in management

If unions are able to hold managers accountable to employees for their actions, we might expect trust in managers to be lower where the union is weak, and higher where employees believe that there is at least a balance of power between union and employer. This hypothesis, outlined above, is confirmed by the results presented in Table 6.

In the economy as a whole, employees in unionised workplaces who thought the union had 'too little' or 'far too little' power were significantly less likely to think that managers at their workplace usually kept their promises to employees than employees in non-unionised workplaces (Model (1)). Where unionised workers thought the balance of power was 'about right', they were significantly more likely than non-unionised workers to think that managers kept their promises. [11]

Where unions were thought to have 'too much' or 'far too much' power, trust in management was not significantly different from trust in non-unionised workplaces.

When introduced to this model, the left-right scale was positive and significant (0.23, t = 3.50) but the union power effects were unaffected. This suggests that perceptions of union power were not simply proxying the value set employees brought with them into the workplace. The power effect was strong in the unionised sector, for members and non-members alike (Models (3) and (4) respectively). It was also apparent for private sector employees in general (Model (5)). In the public sector, trust in management was not significantly different across employees in unionised and non-unionised workplaces, regardless of perceptions of union power (Model (6)). However, among employees in unionised public sector workplaces, weak unions were associated with lower trust than unions with the 'right' amount of power (Model (6), -0.61, t = 4.50).

Did management attitudes to union membership influence employee trust in management in 1998?

Table 7 shows that, in the economy as a whole, employees were more likely to believe that managers usually kept their promises to employees where they also thought management was supportive of union membership (Model (1)). Trust was least evident where employees thought managers at their workplace discouraged membership. These findings held within unionised workplaces (Model (2)) among both members and nonmembers (Models (3) and (4) respectively), and across the private and public sectors (Models (5) and (6) respectively). They were also robust to the introduction of the 'left-right' scale. The effects were quite large. In the whole sample model, the marginal effect of managers discouraging membership, relative to viewing union membership as 'not an issue', was to reduce the chances of agreeing that managers kept promises by 10.5 per cent, whereas the marginal effect of managers encouraging membership was to increase those chances by 14.9 per cent, ceteris paribus.

Since the question is also asked of employees in non-unionised workplaces, we ran the model on these employees separately. Once again, trust in management was greater where management supported union membership, but support for membership was very rare (only six unweighted cases). Among employees in non-unionised workplaces there was no significant difference in trust for management between employees who thought their managers discouraged membership and those who simply saw it as 'not an issue'. But where managers discouraged membership, employees were less likely to agree that managers kept their promises than where managers were accepting of union membership (0.45, t = 2.19).

It is not immediately obvious why employees were more likely to trust managers at their workplace where the employer actively encouraged union membership, and were less trusting of those who discouraged membership. One possible explanation is that management support for worker representation signals managerial openness with employees.

Did perceptions of union effectiveness influence employee trust in management in 1998?

Using models with identical controls to those in Table 6 above, we found that employees were less likely to believe that managers usually kept their promises to employees where the union was thought not to be doing a good job. [12] Trust in management in unionised workplaces was similar to that in non-unionised workplaces where unions were thought to be doing their job well. These findings held in models for the whole sample, members and non-members in unionised workplaces, and for the public sector. In the private sector, trust did not differ significantly between employees in non-unionised workplaces and those in workplaces with ineffective unions. However, trust in management was higher where unions were thought to be doing their jobs well, though the effect was only weakly significant. [13]

Our second measure of union effectiveness, unions' responsiveness to members' problems and complaints, was also associated with trust in managers. Trust in managers did not differ significantly between those working in a workplace without a union or staff association, and those working in a workplace with a union that was viewed as responsive to members' needs. However, where employees thought the union took little notice of members' problems and complaints, employees were less inclined to think that managers usually kept their promises to employees. These findings held in our separate analyses for different groups of employees, except for union non-members and employees in the public sector, where union responsiveness was not significantly associated with trust in management.

We ran identical analyses for our third measure of union effectiveness, that is, the contribution of unions to the smooth running of the workplace. Where unions were perceived to be contributing to the smooth running of the workplace, employees' trust in management was higher than where there were no unions. Where employees disagreed with the statement that unions were contributing to the smooth running of the workplace, trust in management was lower than where there were no unions, ceteris paribus. [14]

Conclusions

This article has shown that, where unions were present, trust in managers was significantly higher when they had sufficient power to challenge employers, where they represented employees effectively, and where they were supported by management. In these circumstances, trust in managers was either higher than, or similar to, trust among employees in non-unionised workplaces. This suggests that unions can hold managers accountable for their behaviour, increasing employees' belief that managers keep their promises. However, where unions were weak, ineffective or faced management opposition, employees were less trusting of management than where there was no union present. The results are summarised in Table 8.

The positive association between managerial support for union membership and employee trust in management can be interpreted in a number of ways. But the finding suggests that management can influence employee perceptions of it for better or for worse through its engagement with unions.

Management can also foster a high trust relationship with employees by ensuring that unions have sufficient power to make a positive contribution to the running of the workplace, since unions with the 'right amount of power' are associated with higher trust in management. Since weak unions are particularly bad for employee perceptions of management, the finding presents a puzzle: why is it that employers so often recognise unions for pay bargaining, but do so little to enhance union influence at the workplace?

Perceptions of management are better where unions are perceived to be doing their job well, where they have regard to their members' problems and complaints, and where they contribute to the smooth running of the workplace. In these circumstances, trust in management is generally no different from that found among employees in non-unionised workplaces. However where unions are viewed as ineffective, trust in management is lower than in comparable non-unionised workplaces. Future research will address the question of what distinguishes effective unions from ineffective ones, and what conditions are conducive to effective unionism.

The central question for government raised by this article is: will the new statutory framework for union recognition support union effectiveness, or will it bring into being unions that do not receive the support of management, or are otherwise weak or ineffective? The answer to this question will go some way towards explaining the likely impact of the statutory recognition procedure on workplace governance.

(*.) Policy Studies Institute. I wish to thank the Economic and Social Research Council for funding this research (grant no. R00222898) and the National Centre for Social Research for access to the data.

NOTES

(1.) For discussion of this issue see Millward et al., 2000.

(2.) Although statutory rights to recognition diminish reliance on employers for formal recognition, in practice it is likely that unions will remain reliant on employer support if they are to make effective representations on behalf of their members.

(3.) In this context, union instrumentality means 'the degree to which the union achieves the valued goals of employees' (Deery et al., 1995, p. 9).

(4.) This variable may be endogenous with respect to perceptions of workplace governance since those least satisfied with their situation are likely to leave. However, we retain it in our analyses to control for the union effect in raising average tenure. Thus our results are net of union effects in increasing the stock of dissatisfied workers arising from the fact that unions raise tenure.

(5.) See Bryson (2001b) for details.

(6.) Union effects may be mediated through satisfaction with the job and terms and conditions. However, if it is assumed that organisational practices such as union recognition influence these intermediate variables, but not vice versa, then the intermediate variables can be omitted from the model without biasing estimates of the total effects of unionisation on trust.

(7.) If employees are unable to distinguish between their own situation and the workplace situation, this measure would not be particularly useful. In fact, although there was a correlation between lower wages and a belief that the workplace pay gap was too large, it was not particularly strong (one-tailed Spearman's Rank coefficient 0.147, p [less than] 0.000).

(8.) Using the SVYLC option available in STATA to compute significant differences for linear combinations of coefficients, we find that the coefficient is -0.20 with at-statistic of 2.07. This finding was confirmed when we ran separate models for employees in unionised and non-unionised workplaces. The union membership effect is only significant among employees in workplaces recognising unions (-0.26, t = 2.38).

(9.) When introduced to Model (5) in Table 5, the pay gap dummy was significant (-0.38, t = 5.25) but the union membership effect remained significant (-0.20, t = 1.95). When interacted, both main effects were significant and the interaction was not significant.

(10.) Using the same controls as those used in Table 5 (with the obvious exception of the sector dummies), trust in management was not significantly different among employees in non-unionised workplaces and those working in workplaces with recognised unions and an on-site representative. However, where employees worked in a unionised workplace without on-site representation, they were significantly less likely to say that management kept their promises (-0.49, t = 2.07).

(11.) The marginal effect of having a union with too little power, relative to having no union at all, was to increase the chances of disagreeing that managers kept promises by 6 per cent, whereas the marginal effect of having a union with the right amount of power was to increase those chances by 6 per cent, ceteris paribus.

(12.) In the whole sample model, the marginal effect of having a union that was not doing its job well relative to having no union at all, was to increase the chances of disagreeing that managers kept promises by 5.4 per cent, ceteris paribus.

(13.) Relative to employees in non-unionised workplaces the coefficient for a union doing its lob well was 0.23, t = 1.82.

(14.) Among employees as a whole, the marginal effect of having a union which employees thought did not contribute to the smooth running of the workplace increased the probability of disagreeing that managers keep their promises by 6 per cent, relative to having no union. The marginal effect of having a union which employees thought did contribute to the smooth running of the workplace was to reduce the probability of disagreeing that management kept promises by 4.4 per cent.

REFERENCES

Addison, J.T. and Belfield, C. (2000), 'The impact of financial participation and employee involvement on financial performance: a re-estimation using the 1998 WERS', Scottish Journal of Political Economy, 47, 5, pp. 571-83.

Bryson, A. (1999), 'Are unions good for industrial relations?', in Jowell, R., Curtice, J., Park, A. and Thomson, K. (eds), British Social Attitudes: the 16th Report -- Who shares New Labour values?, Aldershot, Ashgate.

---(2000), 'Have British workers lost their voice, or have they gained a new one? An analysis of matched employer--employee data', London, Policy Studies Institute Discussion Paper No.2.

---(2001a), 'Employee voice, workplace closure and employment growth: a panel analysis', London, Policy Studies Institute Discussion Paper No. 6.

---(2001b), The Impact of Unions on Workplace Governance, 1983-1998, London, Policy Studies Institute, forthcoming.

Bryson, A. and McKay, S. (1997), 'What about the workers?', in Jowell, R., Curtice, J., Park, A. and Thomson, K. (eds), British Social Attitudes: the 14th Report, Aldershot, Ashgate.

Bryson, A. and Wilkinson, D. (2001), 'Collective bargaining and workplace performance: an investigation using the Workplace Employee Relations Survey 1998', Department of Trade and Industry Research Paper (forthcoming).

Confederation of British Industry (1998), Human Resources Report No. 16, London, Confederation of British Industry.

Cully, M., Woodland, S., O'Reilley, A., Dix, G. (1999), Britain at Work: as depicted by the 1998 Workplace Employee Relations Survey, London, Routledge.

Dasmalchian, A., Blyton, P. and Adamson, R. (1989), 'Industrial relations climate: testing a construct', Journal of Occupational Psychology, 62, pp. 21-32.

Deery, S.J., Erwin, P.J. and Iverson, R.D. (1995), 'Union management, co-operation, dual allegiance and organisational performance', Department of Management and Industrial Relations, University of Melbourne, Working Paper No. 89.

---(1999), 'Industrial relations climate, attendance behaviour and the role of trade unions', British Journal of Industrial Relations, 37, 4, pp. 533-58.

Deery, S. and Walsh, J. (1999), 'The decline of collectivism? A comparative study of white-collar employees in Britain and Australia', British Journal of Industrial Relations, 37, 2, pp. 245-69.

Department of Trade and Industry (1998), Fairness at Work, Cmd 3968.

Fernie, S. and Metcalf, D. (1995), 'Participation, contingent pay, representation and workplace performance: Evidence from Great Britain', British Journal of Industrial Relations, 33, 3, pp. 379-415.

Fernie, S., Metcalf, D. and Woodland, S. (1994), 'Does HRM boost employee-management relations?', Centre for Economic Performance Working Paper No. 548.

Forth, J. and Millward, N. (2000), 'The determinants of pay levels and fringe benefit provision in Britain', National Institute of Economic and Social Research Discussion Paper No. 171.

Freeman, R. and Medoff, J. (1984), What do unions do?, New York, Basic Books.

Gallie, D., White, M., Cheng, Y. and Tomlinson, M. (1998), Restructuring the Employment Relationship, Oxford, Oxford University Press.

Jowell, R., Curtice, J., Park, A., and Thomson, K. (1999), British Social Attitudes: the 16th Report. Who Shares New Labour Values?, Dartmouth, Ashgate.

Marshall, R. (1992), 'The future role of government in industrial relations', Industrial Relations, 31, I, pp. 31-49.

Millward, N., Bryson, A. and Forth, J. (2000), All Change at Work? British Employment Relations, 1980-98, as portrayed by the Workplace Industrial Relations Survey Series, London, Routledge.

Skinner, C. (1997), 'The use of sample survey weights in the regression analysis of the Workplace Industrial Relations Survey Data', University of Southampton, mimeo. Employee perceptions of union power, 1989-1998 1989 1998 Far too much * * Too much 4 2 Right amount 52 45 Too little 32 41 Far too little 6 6 DK/Can't say 6 5 Weighted 822 753 Unweighted 852 714

Source: BSAS. Based on responses to the question: Do you think that the (trade unions/staff association/trade unions or staff association) at your workplace (has/have) too much or too little power? Please use a phrase from this card.' Base is employees working ten or more hours per week in workplaces with a recognised union or staff association. Note: (*.)means under 0.5 per cent. Employee perceptions of management attitudes to union membership at their workplace, 1989-1998 1989 1998 No Recog Total No Recong Total recog. recog. Encourages * 18 10 1 11 6 Accepts 4 52 32 8 54 30 Discourages 18 5 10 17 5 11 Not an issue 75 23 45 68 27 48 Don't know 2 2 2 6 2 4 Not answered 1 * 1 * 0 * Weighted 607 825 1432 791 753 1544 Unweighted 607 855 1462 714 714 1428

Source: BSAS. Note: based on responses to the question: 'How would you describe the management's attitude to trade unions at the place where you work? Would you say that management encourages trade union membership, accepts it or would accept it. discourages trade union membership, or isn't it really an issue at your workplace?' Based on employees working ten or more hours per week. Note: (*.)means under 0.5 per cent. Percentage of employees agreeing that union is doing its job well, 1983-1998 Cell percentages 1983 1984 1985 1986 1987 1989 1990 1991 1993 All 60 63 57 61 62 57 61 61 58 Members 63 67 61 64 66 63 63 63 60 Non-members 51 52 49 53 52 42 56 56 55 Unweighted base 531 493 530 945 864 848 753 694 628 1994 1996 1998 All 55 59 62 Members 61 64 67 Non-members 42 50 55 Unweighted base 790 763 714

Source: BSAS. Based on responses to the question: 'On the whole, do you think (trade union(s)/staff association(s) at this workplace do(es) its) job well or not?' Percentage of 'yes' responses calculated on base including 'Don't knows'. Base is employees working ten or more hours per week in workplaces with a recognised union or staff association. Perceptions that managers keep promises, 1998, per cent No recognition Union recognised All Agree strongly 8 3 6 Agree 44 36 40 Neither agree/disagree 27 36 31 Disagree 17 22 19 Disagree strongly 4 4 4 Weighted 600 603 1203 Unweighted 539 575 1114 Source: BSA98. Based on responses to the statement 'Managers at my workplace usually keep their promises to the employees' by employees working ten hours or more per week. Perceptions that managers keep promises (1) (2) Union/staff association recognition (ref.: no recognition) -0.262 -0.149 (3.82) [**] (1.70) Nature of representation (ref.: no union/staff assn.) Union recognised Staff association recognised Union/staff association, but not recognised Individual membership (ref.: never member) Union member Staff association member Ex-member of union/staff association Workplace and individual unionisation (ref.: no recognition/non-member) Member, recognition Non-member, recognition Member, no recognition On-site worker representative (ref.: no representative) -0.167 (1.89) cut1:Constant -.1.882 -1.912 (21.47) [**] (21.05) [**] cut2:Constant -0.875 -0.903 (14.21) [**] (14.13) [**] cut3:Constant -0.005 -0.031 (0.09) (0.54) cut4:Constant 1.469 1.446 (19.61) [**] (19.11) [**] Controls? NO NO F-statistic (1,1169) = (2,1126) = 14.56 8.58 Observations 1116 1116 (3) (4) Union/staff association recognition (ref.: no recognition) Nature of representation (ref.: no union/staff assn.) Union recognised -0.191 -0.102 (2.03) [*] (0.93) Staff association recognised 0.015 0.053 (0.08) (0.26) Union/staff association, but not recognised -0.142 -0.079 (0.68) (0.37) Individual membership (ref.: never member) Union member Staff association member Ex-member of union/staff association Workplace and individual unionisation (ref.: no recognition/non-member) Member, recognition Non-member, recognition Member, no recognition On-site worker represetative (ref.: no representative) -0.144 -0.001 (1.56) (0.01) cut1:Constant -1.920 -1.750 (20.91) [**] (7.36) [**] cut2:Constant -0.911 -0.666 (13.94) [**] (2.82) [**] cut3:Constant -0.038 0.267 (0.64) (1.13) cut4:Constant 1.441 1.851 (18.94) [**] (7.60) [**] Controls? NO YES F-statistic (4,1124) = (35,1091) = 4.75 3.99 Obervations 1116 1114 (5) (6) Union/staff association recognition (ref.: no recognition) Nature of representation (ref.: no union/staff assn.) Union recognised -0.015 (0.14) Staff association recognised 0.013 (0.06) Union/staff association, but not recognised -0.057 (0.28) Individual membership (ref.: never member) Union member -0.202 (2.02) [*] Staff association member 0.070 (0.36) Ex-member of union/staff association 0.017 (0.18) Workplace and individual unionisation (ref.: no recognition/non-member) Member, recognition -0.180 (1.62) Non-member, recognition 0.020 (0.20) Member, no recognition -0.067 (0.35) On-site worker represetative (ref.: no representative) 0.006 (0.06) cut1:Constant -1.734 -1.737 (7.28) [**] (7.31) [**] cut2:Constant -0.640 -0.645 (2.69) [**] (2.72) [**] cut3:Constant 0.296 0.290 (1.24) (1.22) cut4:Constant 1.881 1.873 (7.68) [**] (7.65) [**] Controls? YES YES F-statistic (38,1088) = (34,1092) = 3.93 4.29 Obervations 1114 1114 Notes: T-statistics in parentheses. (*.)=significant at 95 per cent confidence level. (**.)=sifnificant at 99 per cent confidence level or above. Controls are: gender, age, ethnicity, qualifications, region, hours worked, sector, workplace size, workplace tenure, occupational class. Perceptions of union power and management keeping promises (1) (2) Union power at workplace: Too much power -0.266 -0.035 (1.11) (0.13) Right amount of power 0.314 0.704 (3.04) [**] (6.76) [**] Too little power -0.301 (2.55) [*] Power data missing 0.054 0.364 (0.37) (1.73) Individual membership: Union member -0.134 -0.071 (1.33) (0.54) SA member 0.055 0.163 (0.32) (0.78) Ex-member of union/SA 0.036 0.248 (0.38) (1.70) cut1: Constant -1.769 -1.381 (7.41) [**] (4.27) [**] cut2: Constant -0.644 -0.108 (2.69) [**] (0.34) cut3: Constant 0.319 1.001 (1.33) (3.06) [**] cut4: Constant 1.926 2.798 (7.77) [**] (7.82) [**] Controls? YES YES F-statistic (38,1088) = (37,1089) = 5.16 3.97 Observations 1114 615 (3) (4) Union power at workplace: Too much power -0.344 0.291 (1.53) (0.70) Right amount of power 0.697 0.759 (5.28) [**] (4.61) [**] Too little power Power data missing 0.486 0.284 (1.40) (1.02) Individual membership: Union member SA member Ex-member of union/SA cut1: Constant -1.572 -1.429 (3.44) [**] (2.51) [*] cut2: Constant -0.284 0.097 (0.64) (0.18) cut3: Constant 0.703 1.572 (1.54) (2.90) [**] cut4: Constant 2.454 3.671 (4.91) [**] (6.31) [**] Controls? YES YES F-statistic (34,1092) = (33,1093) = 3.64 2.68 Observations 363 252 (5) (6) Union power at workplace: Too much power -0.126 -0.635 (0.63) (1.07) Right amount of power 0.388 0.205 (3.13) [**] (0.75) Too little power -0.241 -0.402 (1.52) (1.42) Power data missing 0.007 0.279 (0.04) (0.76) Individual membership: Union member -0.249 -0.081 (1.73) (0.48) SA member 0.225 -0.020 (1.11) (0.08) Ex-member of union/SA -0.017 0.174 (0.16) (0.80) cut1: Constant -2.087 -0.863 (6.54) [**] (1.77) cut2: Constant -0.926 0.237 (2.89) [**] (0.49) cut3: Constant -0.024 1.374 (0.08) (2.83) [**] cut4: Constant 1.608 3.065 (4.90) [**] (6.10) [**] Controls? YES YES F-statistic (36,1132) = (35,1133) = 5.83 7.62 Observations 712 368

Notes: T-statistics in parentheses.

(*.)=significant at 95% confidence level.

(**.)=significant at 99% confidence level or above.

Model (1) = whole sample.

Model (2) = employees in unionised sector.

Model (3) = union members in unionised sector.

Model (4) = union non-members in unionised sector.

Model (5) = employees in private sector.

Model (6) = employees in public sector. Controls are: gender, age, ethnicity, qualifications, region, hours worked, sector (exc. from Models (3) and (4)), workplace size, workplace tenure, occupational class. Reference categories for union power: In models (1), (5) and (6) no recognised union; in (2), (3) and (4), union with too little power. Reference categories for individual membership: in models (1), (2), (5) and (6): never union/staff association member. Perceptions of management attitudes to unions and management keeping promises (1) (2) Management attitudes to membership (ref.: 'not an issue') Encourages membership 0.709 0.569 (3.71) [**] (2.88) [**] Accepts membership 0.117 -0.050 (1.35) (0.47) Discourages membership -0.393 -0.975 (3.07) [**] (4.43) [**] Don't know -0.043 -0.299 (0.23) (1.32) Nature of worker representation: Union recognised -0.156 (1.34) Staff association recognised -0.035 0.201 (0.15) (0.87) Union/staff association, but not -0.122 -0.033 recognised (0.58) (0.15) On-site worker representative 0.002 -0.061 (ref.: no representative) (0.02) (0.51) Individual membership (ref.: never member): Union member -0.266 -0.241 Staff association member 0.014 0.050 (0.07) (0.21) Ex-member of union/staff 0.051 0.196 association (0.54) (1.38) cut1: Constant -1.828 -1.882 (7.49) [**] (5.36) [**] cut2: Constant -0.715 -0.617 (2.96) [**] (1.85) cut3: Constant 0.237 0.471 (0.98) (1.40) cut4: Constant 1.853 2.241 (7.55) [**] (6.35) [**] Controls? YES YES F statistic (42,1084) = (41,1085) = 4.14 3.54 Observations 1114 615 (3) (4) Management attitudes to membership (ref.: 'not an issue') Encourages membership 0.730 0.200 (3.30) [**] (0.57) Accepts membership 0.149 -0.351 (1.06) (1.90) Discourages membership -1.217 -0.919 (3.78) [**] (2.61) [**] Don't know -0.085 -0.228 (0.23) (0.60) Nature of worker representation: Union recognised Staff association recognised 0.354 0.294 (1.53) (0.90) Union/staff association, but not 0.058 -0.396 recognised (0.14) (1.37) On-site worker representative -0.047 0.169 (ref.: no representative) (0.27) (0.80) Individual membership (ref.: never member): Union member Staff association member Ex-member of union/staff association cut1: Constant -1.731 -1.666 (3.32) [**] (2.73) [**] cut2: Constant -0.419 -0.165 (0.84) (0.30) cut3: Constant 0.558 1.283 (1.10) (2.28) [*] cut4: Constant 2.328 3.361 (4.42) [**] (5.60) [**] Controls? YES YES F statistic (38,1088) = (37,1089) 3.88 2.66 Observations 363 252 (5) (6) Management attitudes to membership (ref.: 'not an issue') Encourages membership 1.002 0.560 (2.28) [*] (2.70) [**] Accepts membership 0.169 0.103 (1.45) (0.72) Discourages membership -0.340 -0.694 (2.44) [*] (1.77) Don't know 0.035 -0.266 (0.15) (0.77) Nature of worker representation: Union recognised -0.037 -0.710 (0.26) (2.49) [*] Staff association recognised 0.257 -0.721 (0.82) (1.71) Union/staff association, but not 0.069 -0.640 recognised (0.30) (1.41) On-site worker representative -0.136 0.443 (ref.: no representative) (1.03) (2.84) [**] Individual membership (ref.: never member): Union member -0.392 -0.190 Staff association member 0.200 -0.001 (0.94) (0.00) Ex-member of union/staff -0.007 0.227 association (0.07) (1.09) cut1: Constant -2.105 -1.242 (6.57) [**] (2.50) [*] cut2: Constant -0.950 -0.157 (3.01) [**] (0.33) cut3: Constant -0.057 0.984 (0.18) (2.04) [*] cut4: Constant 1.605 2.667 (5.04) [**] (5.42) [**] Controls? YES YES F statistic (40,1086) = (39,1087) = 3.73 6.74 Observations 712 368

Notes: T-statistics in parentheses.

(*.)= significant at 95% confidence level.

(**.)= significant at 99% confidence level or above Model (1) = whole sample. Model (2) = employees in unionised sector. Model (3) = union members in unionised sector. Model (4) = union non-members in unionised sector. Model (5) = employees in private sector. Model (6) = employees in public sector. Controls are: gender, age, ethnicity, qualifications, region, hours worked, sector (exc. from Models (3) and (4)), workplace size, workplace tenure, occupational class. Summary of results Whole economy Private sector Perceptions of union power at the workplace (relative to: no recognised union) Right amount Positive Positive Too little Negative Not significant Too much Not significant Not significant Perceptions of managerial attitudes to union membership (relative to: membership is 'not an issue') Supportive Positive Positive Opposed Negative Negative Perceptions of union effectiveness (relative to: no recognised union) Effective Not significant Not significant Ineffective Negative Not significant Public sector Perceptions of union power at the workplace (relative to: no recognised union) Right amount Not significant Too little Not significant Too much Not significant Perceptions of managerial attitudes to union membership (relative to: membership is 'not an issue') Supportive Positive Opposed Negative Perceptions of union effectiveness (relative to: no recognised union) Effective Not significant Ineffective Negative

Note: These findings summarise results from multivariate models controlling for a range of other factors such as demographic characteristics, job characteristics, workplace characteristics, and attitudes. All positive and negative effects are significant at a 95 per cent confidence level or higher.