Canadian public service employee satisfaction and its main drivers.

Hickey, Alexandra ; Bennett, Scott Edward

Introduction

This paper will analyse the employee satisfaction of federal civil servants and the variables that drive it. The policy importance of this inquiry arises from the possibility that understanding the factors important to civil servant satisfaction could lead to improvements in job performance, employee retention, employee recruitment and, ultimately, provision of services to the public. The complete set of linkages involved in addressing this larger question would obviously require additional analysis and data.

What is presented here, however, is a central part of the factors that influence the civil service and its effectiveness. Furthermore, the negative internal and external perceptions of the civil service (Lynch 2009) underline the importance of analysing these issues. Using the results of the 2005 Public Service Employee Survey (PSES) this paper hopes to uncover which factors have the greatest effect on employee satisfaction in the Canadian public service. It will build on the models created by Heintzman and Marson and use research undertaken by Duxbury, Zussman and Kernaghan to further develop the model for analysing the 2005 PSES data.

The six hypotheses that will be examined in this paper come from the articles discussed in the literature review. Substantial research has been conducted on what steps are necessary to improve the public service, but quantitative analysis appears to be under-represented in previous work though not entirely absent. This paper aims to expand current research by making greater use of statistical techniques, including cross tabulations and regression analysis. Lastly, this paper will offer some reflections on the implications of our findings for renewing the Canadian public service and making it an employer of choice.

Hypotheses and Canadian Uterature review

The 1891-92 Royal Commission on the Civil Service of Canada recommended a merit-based public service in the belief that, "the service will soon become attractive to many persons who now seek other avenues of employment and in general the title of public servant will be an honour to be coveted" (Kernaghan 2001: 2). More than a century later, governments are still taking steps to make public service a more desirable career.

Despite the positive intentions behind strategies to improve the public service as a career, there is a considerable amount of research suggesting that the general population does not have a positive view of the public service (Zussman 1982: 73; Phillips, Little and Goodine 1997; Kernaghan 2001: 4; Malloy 2004: 287; Baird and Cote 2007: 11). Some of this research indicates that public servants accept the fact that the public thinks of them in negative terms.

Many scholars including Kernaghan (2001), Malloy (2004) and Zussman (1982) argue that such negative perceptions are largely unfounded and have been the product of self-interest by journalists, politicians and the general public. Kernaghan quotes Gilbert Scott, a former federal public-service commissioner in saying, "negative perceptions of public servants are essentially a function of social mythology, not of who we are or what we really do" (Kernaghan 2001: 6). Furthermore, Bourgault and Gusella (2001) quote Savoie in saying,

the most advanced economies in the world have or have had a strong public service. Less developed countries have at least one thing in common--they lack a non-partisan, professional public service. One of Canada's biggest assets--and a largely unrecognized asset--is the matchless integrity, the exceptionally high quality and the extraordinary commitment of Canada's public services to the success of the nation (Bourgault and Gusella 2001: 30).

Kernaghan believes that efforts to enhance pride in the public service have been based on the belief in a virtuous cycle whereby an increase in pride will lead to improved performance and then, "improved performance will lead to greater public recognition of the public service; and greater public recognition will increase public servants' pride" (Kernaghan 2001: 7). Employees who are proud of their jobs have also been shown to be more committed to the organization's success and are more likely to desire to remain with the organization. A 1999 study of federal knowledge workers in Canada notes that "employees who are highly committed to the organization work harder, are absent less often and are less likely to leave the organization" (Kernaghan 2001: 8).

Our analysis will concentrate on the perceptions of civil servants and the drivers of those perceptions. The literature discussed so far suggests that the first hypothesis to be explored is

H1: Employees who are proud or very proud of their organization will be less likely to desire leaving the public service within five years.

Lee, Duxbury and Higgins (1994) describe the importance of controlling for gender in relationships concerning the workforce. In their research, three hundred working mothers in the federal public service were surveyed for a study concerning employed mothers. One of the key findings was that "almost sixty percent of the sample had considered quitting their jobs" (1994: 2). The researchers also found that, in addition to formal benefits, "a number of employed mothers mentioned that having supervisors, who understand their work and family demands, is critical. Without such understanding, these women noted that they would feel uncomfortable in making use of the benefits to which they were formally entitled to" (1994: 3). We will use this insight from the general literature on job perceptions and satisfaction as a basis for our second hypothesis:

H2: There will be a positive relationship between perceptions of recognition by immediate supervisors and career satisfaction. This relationship may also be more pronounced among women when the relationship is controlled for gender.

The study also pointed out another important independent variable to consider in this analysis. Not only are there differences between men and women, but there are also differences between the occupational groups to which they belong. Lee, Duxbury and Higgins say that "employees in 'career positions' (those in professions and in managerial positions) may cope with work-family related conflict in a different manner than those in 'earner' positions (clerical or administrative)" (1994: 3). Research also considers the impact of a compressed work week and working from home on employee satisfaction. How employees feel about work/life balance can greatly influence how they perceive their job satisfaction. The researchers found that "the only benefit which distinguished among mothers who had considered quitting and mothers who had not was whether the mother had the opportunity to work at home" (1994: 20). This leads us to our third hypothesis pertaining to public servant views of their jobs:

H3: It is expected that there will be a positive relationship between variables measuring life balance and overall career satisfaction. It is expected that the results may differ slightly when the relationship is controlled for gender, with the relationship being stronger for women.

Phillips, Little and Goodine explain that early stereotypes in literature on women in management portrayed women as "less committed to and as deriving less meaning from work than men" (Phillips, Little and Goodine 1997: 565). The authors disprove this stereotype with many examples. Hence, our fourth hypothesis:

H4: It is expected that the PSES 2005 results will show that women are just as committed to work organization success as their male counterparts.

Phillips, Little and Goodine describe how the results of the 1990 Task Force on Barriers to Women in the Public Service showed that "the most critical impediment to the advancement of women is a suffocating, even hostile, organizational climate that is characterized by gender stereotyping and negative male co-worker attitudes that undervalue women's contributions" (1997: 571). The authors conducted a survey of 112 managers in the public service and found interesting results. Of the people surveyed, "78.6 percent of the women were married and only 64 percent had children, while 95 percent of men were married and 89.3 percent had children" (1997: 577).

Duxbury, Lyons and Higgins discuss that a preference for public service employment is affected by age, educational level and gender. In their study, respondents' levels of organizational commitment were measured using the nine-item version of Porter et al.'s Organizational Commitment Questionnaire (Duxbury, Lyons and Higgins 2006: 611). Respondents were asked to rate their agreement using a five-point Likert scale ranging from strongly agree to strongly disagree. The reliability coefficient (Cronbach's Alpha) was 0.90. This indicates that the following hypothesis concerning public servants' views of their jobs should be examined:

H5: There will be a positive relationship between commitment to organizational success and pride in the work of the public service unit.

While all the hypotheses presented thus far are based on relations between two variables, our sixth hypothesis involves the influence of a set of variables on an index. It arises from the literature on "chains of values" in organizations. Heintzman and Marson (2005) researched the possibility of a "public service value chain." The authors explain that the existence of a "service profit chain" in the private sector is well studied and accepted by many major companies. The general idea is that there is a

link between employee satisfaction and customer satisfaction and between customer satisfaction and the bottom line. The combination of these two relationships yields a causal chain in which an improvement in customer attitudes and behaviour leads to an increase in growth and profit (2005: 551).

In its simplest form, this concept maintains that there are direct and strong relationships between profit, growth; customer loyalty; customer satisfaction; the value of goods and services delivered to customers; and employee capability, satisfaction, loyalty and productivity.

According to Heintzman and Marson (2005), the idea behind a "public sector value" chain model involves two sets of linkages: "one between the satisfaction and commitment of public employees and the satisfaction of citizen clients with the quality of government service delivery; the other, between citizen satisfaction and citizen trust and confidence in public institutions" (Heintzman and Marson 2005: 554). The model also outlines drivers or independent variables that can be predictors of performance outcomes. For the purposes of this paper, the first of these linkages--the drivers of employee satisfaction--will be reviewed.

An initial attempt to identify drivers of employee satisfaction in the public sector was carried out for the Government of Manitoba in 2002. Analysis of the 2001 Manitoba Employee Survey isolated ten questions that best identified the drivers of job satisfaction within the Manitoba public sector. The study showed that five areas consistently had a large impact on job satisfaction: "authority in the job, adequate training, appreciated as an employee, clarity of departmental goals, recognition for good work" (Heintzman and Marson 2005: 560).

Another study conducted by Erin Research found a "correlation of 0.57 between job satisfaction and commitment" (Heintzman and Marson 2005: 560). One of the most important things to emerge from Erin Research's Peel study is that job satisfaction was found as a driver of commitment, but commitment is not a driver of satisfaction. This study found eight drivers for satisfaction and four for commitment. Drivers of satisfaction included

a career path that offers opportunities for advancement, fair pay and benefits, the perception of the company providing good value to customers, a satisfactory work environment defined by: a reasonable workload, good relations with immediate supervisor, smoothly functioning organizational dynamics, good relationships with colleagues and effective internal communications. Drivers of commitment were found to be: job satisfaction, a career path that offers opportunities for advancement, a positive perception of senior management and the perception of the company providing good value to customers (Heintzman and Marson 2005: 561).

Each of these drivers has a significant impact on job satisfaction or commitment, or both. The conclusion was that a public sector organization that takes steps to improve performance on these drivers can expect to see improvements in higher job satisfaction or commitment and thus overall employee engagement. This leads us to our next hypothesis:

H6: The types of drivers outlined in the 2001 Manitoba government study and the Erin Research study will have a significant effect on employee satisfaction in the 2005 PSES data.

The literature of interest here is based on employee surveys. Harwood (1998) gives many good reasons why it is advantageous for both private-sector companies and public-sector organizations to conduct employee surveys. He explains that "employee surveys have been a standard practice in the most successful and adaptive organizations for some time. Private sector enterprises especially--IBM is an example--regularly ask their staff for views on a wide range of issues. They have adopted the employee survey, in one form or another, as a best practice, a management tool with proven value" (Harwood 1998: 2). Employee surveys can give the pulse of an organization and improve organizational effectiveness when directed efforts are geared towards improvements. Harwood upholds that employee surveys are valuable management tools and should be implemented more fully in the Canadian public service. Pursuant to this, we will now consider the Public Service Employee Survey data set for analysis.

Public Service Employee Survey: The source of our data

The PSES for the Canadian public service is the world's largest survey of civil servants. In 1997, the clerk of the Privy Council introduced the idea of a voluntary survey of all federal public servants. The first survey was conducted in 1999; subsequent surveys have been conducted in 2002, 2005 and most recently 2008 and have increasing response rates.

The survey is a census style with a cross-sectional design. There was no change in instrument design from 2002 to 2005 so the results can easily be compared. Since the 2008 and 2011 public micro-data files were not yet released at the time of analysis, this paper studied the 2005 survey results.

The target population for 2005 consisted of 180,824 individuals. The overall response rate for the 2005 Public Service Employee Survey was fifty-nine per cent (Statistics Canada 2006). This means that 106,456 public servants responded to the 2005 PSES. This is a large sample size and will give a fairly good indication of how public servants as a group feel about their jobs and which factors contribute to improved employee satisfaction.

One of the limitations of studying the PSES data is the fact that much of the data has been re-coded to protect the identities of individual public servants. According to Statistics Canada, "it is prohibited by law to release any data which would divulge information obtained under the Statistics Act that relates to any identifiable person, business or organization without the prior knowledge or the consent in writing of that person, business or organization" (2006). Because of the requirements of the Statistics Act, this paper includes very little analysis using socio-economic or demographic variables.

In addition, approximately three per cent of the records were treated by local suppression when there were fewer than five responses in any cell of a table when all possible combinations of all the demographic variables were cross tabulated. One or more of the demographic variables were treated by randomly assigning a "not stated" code (Statistics Canada 2006).

In the case of some non-demographic variables, it is possible to create more variation by combining several variables into a complex index. This is in fact is what was done with some of the regression-oriented analysis presented later.

Very little academic research has been undertaken specifically on the PSES since its inception in 1999. The Treasury Board Secretariat (TBS) publishes findings about the survey, along with a management guide and response. The TBS findings are at the univariate level and only present the frequencies of the results. More complex statistical techniques including bivariate and multivariate analyses seem to be absent from the analysis.

Given that this research will use the 2005 PSES data, it is important to discuss the context under which the 2005 PSES was conducted. Prior to the launch of the PSES, Justice Gomery released the report of the Commission of Inquiry into the Sponsorship Program and Advertising Activities (Canada, Commission of Inquiry into the Sponsorship Program and Advertising Activities 2005). In addition, during the time that the survey was open for completion, a federal election campaign was underway. The effects of these two events cannot be accounted for without further comparative analysis of the latest wave of PSES data from 2008, and Statistics Canada had not made that available at the time of our analysis. A brief description of the main findings of the 2005 survey will be provided.

There were several noteworthy findings published by the Treasury Board in its 2005 report. First, many employees signalled that the quality of their work suffered as a result of one or many factors. These include having to do the same or more work with fewer resources (43% agree), lack of stability in the organization (41% agree), constantly changing priorities (40% agree), too many approval stages (40% agree) and unreasonable deadlines (30% agree). These findings are averaged across all federal departments which participated in the 2005 PSES. The TBS report makes the following remarks in response to this finding: "At the time of the 2005 survey, several large departments had recently experienced extensive reorganization and major shifts in their mandates. This may have contributed to the considerable increase in concerns in 2005 compared to 2002 with respect to organizational instability, the number of approval stages and shifting priorities" (Treasury Board Secretariat 2005). Second, almost one-third of respondents (31%) reported having more than three supervisors in three years. Third, a considerable number of public servants stated that they have difficulty managing their work/life balance (30% stated that they could only sometimes or rarely/never do so).

Variables and methods

Here, we describe in greater detail the variables and methods used to test our six hypotheses. Most of these hypotheses can be tested using cross tabulation and related techniques. Multiple linear regression will also be useful in testing one of them.

The first step to testing the arguments put forth in the literature was to find similar variables within the PSES 2005 data. Since the PSES asked many dimensions of questions, the hypotheses put forth by other authors were easy to replicate for this analysis.

The first five hypotheses: Table-oriented analysis The first hypothesis was put forth by Kernaghan (2001) and involved the relationship between pride and desire to leave their jobs. The following questions were used from the 2005 PSES to test this relationship:

--Q033- I am proud of the work carried out in my work unit

--Q099- Are you planning to leave the public service within the next five years?

Pride was the independent variable; desire to leave the public service was the dependent variable. Since both variables were coded as ordinal, gamma and chi-square statistics were used for this analysis.

The second hypothesis concerned support by supervisors and job satisfaction. It was advocated by Lee, Duxbury and Higgins (1994) who also thought that this relationship would be stronger for women than for men. The following PSES 2005 variables were used to test this hypothesis:

--Q021- I get adequate recognition from my immediate supervisor when I do a good job

--Q089- I am satisfied with my career in the public service.

Career satisfaction was used as the dependent variable; support by supervisor was the independent variable. Both variables were ordinal variables so, again, Gamma and chi-square were selected as the appropriate techniques. This hypothesis contended that this relationship would be more pronounced in women. To test this, the original bivariate relationship was controlled with gender.

The third hypothesis analysed concerned the relationship between work/life balance and overall job satisfaction. This hypothesis, also advocated by Lee, Duxbury and Higgins (1994), purported that this relationship might be stronger for women than men. The variables used for this analysis were as follows:

--Q014- I can balance my personal, family and work needs in my current job

--Q089- I am satisfied with my career in the public service.

Work/life balance was input as the independent variable and career satisfaction was used as the dependant variable. The relationship was then controlled for with a gender variable.

The fourth hypothesis that will be examined explores the relationship between gender and organizational commitment. For this analysis, cross tabulations will be used to compare the percentage of respondents in each category. The variables used to test this hypothesis were:

--Q109- Gender

--Q086- I am strongly committed to making my organization successful.

Gender was the independent variable and organizational commitment was the dependent variable. Cramer's V and chi-square were selected as the statistics for this analysis, as gender is treated as a nominal variable.

The fifth hypothesis looked at the relationship between pride and organizational commitment. Kernaghan (2001) hypothesized that people who are proud of their jobs will also be more committed to making their organization successful. To test this theory, the following variables from the 2005 PSES data were selected:

--Q033- I am proud of the work carried out in my unit

--Q086- I am strongly committed to making my organization successful.

Organizational commitment was the dependent variable and pride was the independent variable. Chi-square and Gamma were selected as the statistics that would be used for this analysis to show the strength, direction and significance of the relationship.

Hypothesis six: Regression-based analysis

The sixth hypothesis was tested by multiple linear regression analysis and explained which variables contribute to employee satisfaction[degrees] The dependent variable--employee satisfaction--was computed by combining several variables into an index. The independent variables used in the regression were suggested by the Peel Study as discussed by Heintzman and Marson (2005), and will be reviewed in greater detail below.

The dependent variable that emerged from this analysis was overall employee satisfaction with their jobs. This occurred after combining the following variables into an index: proud of work carried out in my unit (Q033), commitment to organizational success (Q086), organization is a good place to work (Q088) and satisfied with my career (Q089). All of the variables were measured using a 4-point Likert Scale with responses ranging from strongly agree to strongly disagree. Neutral responses as well as those who responded don't know were removed from this analysis as missing values. The resulting index was tested for measurement reliability and dimensionality, and results indicated that the index met conventional measurement standards. More details on this are reported in an appendix. The specific wording of the questions in the index is provided in Table 1.

Independent variables that could potentially be predictors of employee satisfaction were obtained from Heintzman and Marson (2005). The literature discussed eight areas that were found in other studies to contribute to employee satisfaction: opportunities for advancement, fair pay and benefits, good customer service, reasonable work load, good relations with supervisor, good relations with colleagues, effective internal communications and positive perceptions of senior management. See Table 2 for a list of variables that were selected from the PSES 2005 data for this regression analysis.

Variables that were as close as possible to the questions asked in the Peel Study were selected from the PSES. Some of the questions were worded slightly differently but encompassed the same theme. For example, the Peel Study indicated that fair pay and benefits were important. The variable selected from the 2005 PSES was "I am classified fairly (my current group and level) compared with others doing similar work in my organization or elsewhere in the Public Service." In the public service, the level of classification directly relates to the pay scale associated with it. Another example is how the Peel Study suggested perception of senior management was an important predictor of employee engagement. The 2005 PSES asked very specific questions about employee perceptions of senior management. For this analysis the question "I believe senior management will try to resolve concerns raised in this survey" was selected because it showed how confident people were in the abilities of senior management.

All of the variables were ordinal rather than interval due to the requirements of the Statistics Act mentioned above. No neutral option was given to respondents.

Findings

The results of each of the first five hypotheses are provided below. They are then summarized in Table 3. The first hypothesis tested was the relationship between pride and likeliness to leave the public service in the next five years. This hypothesis stemmed from the research conducted by Kernaghan (2001).

H1: Employees who are proud or very proud of their organization will be less likely to desire leaving the public service within five years.

A statistically significant but weak negative relationship was found. This finding makes sense given the way the variable was coded. People who strongly disagree that they are proud of their work unit are more likely to want to leave the public service in the next five years. People who are proud of their jobs are less likely to want to leave the public service. This confirms the finding of Kernaghan (2001).

The second hypothesis tested was the relationship between support of a supervisor and job satisfaction. This hypothesis was based on the research by Lee, Duxbury and Higgins (1994), who also noted that this relationship would likely be more important to women.

H2: There will be a positive relationship between perceptions of recognition by immediate supervisors and career satisfaction. This relationship may also be more pronounced among women when the relationship is controlled for gender.

A statistically significant, strong, positive relationship was found. People who get adequate recognition from their supervisors are more likely to be satisfied with their job. Lee, Duxbury and Higgins (1994) suggested that this relationship might be stronger among women so to test this hypothesis the relationship was controlled with gender. The Lee, Duxbury and Higgins hypothesis was not confirmed. The relationship was very similar for both men and women and was actually slightly stronger in the male group.

The third hypothesis was also advocated by Lee, Duxbury and Higgins (1994). They argued that there would be a positive relationship between work/life balance and overall job satisfaction. They also noted that this relationship might be more important to women than men.

H3: It is expected that there will be a positive relationship between variables measuring life balance and overall career satisfaction. It is expected that the results may differ slightly when the relationship is controlled for gender, with these values being of more importance to women than men.

The relationship was significant, positive and moderate in strength. This confirms the findings of Lee, Duxbury and Higgins (1994). When the relationship was controlled for gender, there was little change. It remained significant, positive and moderate for both males and females. This does not support the Lee, Duxbury and Higgins hypothesis that this relationship would be stronger among women than men. In fact, the relationship was slightly stronger in men than women.

The fourth hypothesis emerging from the literature was that women are just as committed to their jobs as men. This hypothesis was put forward by Phillips, Little and Goodine (1997), who advocated that this would be true despite initial research which suggested that men were more committed to work than women, who were more likely to be committed to their home lives. While this was the logic of the argument in early literature, it is highly likely that commitments to home, work and other aspects of life are determined in a complex manner. The views that there may or may not be gender differences in commitment are, however, part of the literature and thus worthy of testing.

H4: It is expected that the PSES 2005 results will show that women are just as committed to their work organization success as their male counterparts.

The results from the PSES 2005 show that there is a statistically significant, weak relationship between gender and commitment to organizational commitment success. In the context of this large a data set, this should be taken to mean that there is virtually no gender effect on job commitment, as even a very weak relationship would appear to have inferential significance with such a huge sample. Thus, the hypothesis was substantively, if not inferentially, confirmed that women are just as committed as men to their organizations. In fact, men and women were almost equally committed to organizational success in the public service. Fifty-five per cent of men were strongly committed to their organization's success, while fifty-five point four per cent of women were equally committed. In the mostly agree category, forty-one point eight per cent of men and forty-two point four per cent of women were strongly committed to organizational success. These percentages are virtually equal and show that neither gender is more likely to be committed to their job than the other. They are almost equally committed to making their organizations a success.

The fifth hypothesis was the relationship between organizational commitment and pride in the work unit. This hypothesis was put forth by Duxbury, Lyons and Higgins (2006).

H5: There will be a positive relationship between commitment to organizational success and pride in the work of the public service unit.

There is a significant, strong, positive relation between organizational commitment and pride in the public service. This means that people who are proud of their work unit are more likely to be committed to making the organization a success. This confirms the hypothesis by Duxbury, Lyons and Higgins.

The sixth hypothesis was examined using multiple regression analysis on multiple variables obtained from the literature to identify what factors have the greatest impact on employee satisfaction.

H6: The types of drivers outlined in the 2001 Manitoba government study and the Erin Research study will have a significant effect on employee satisfaction in 2005 PSES data.

As mentioned, the dependent variable was employee satisfaction. It was computed by creating an index of multiple variables after asking employees their overall opinion of their jobs. Details on the characteristics of this index are contained in Appendix 1. In reporting the results of this multiple linear regression analysis, this paper provides general, non-technical descriptions of the results, in so far as possible. At the same time, it provides the more specific statistical results that will be of interest to some readers. These results are found in Table 4. Note that the regression was examined for any of the major problems that might typically occur in such analysis, and it was found to be free of any obvious problems. For example, multicolinearity (unusually strong relations between or among independent variables) did not present a problem.

All of the variables included in this regression analysis were significant predictors of job satisfaction. This is not surprising as all variables came from the literature. In addition, all independent variables had positive effects on employee satisfaction except for "I can complete my assigned workload during my regular working hours," which had a negative effect.

The variable that had the largest Beta and thus was the greatest relative predictor was whether or not a work unit was perceived to function as a team. The second largest predictor was the confidence the employee had in senior management to respond to concerns revealed in the PSES data. It is surprising that these two variables were the largest predictors of employee satisfaction as measured by this index. One would think that pay/ classification and life balance would be more important. The theme supporting our findings appears to be trust in the structure of work and the transparency of the uses of the PSES. The overall model explains fifty-five per cent of the variation in employee satisfaction. This model is a strong predictor of overall employee satisfaction and this is particularly true when one considers that it is based on survey generated micro data.

Conclusion

This analysis has helped to explain the factors that contribute to employee satisfaction in the Canadian public service. This is important knowledge given the current challenges with recruitment and retention, as well as the probable impact of civil service employee satisfaction on external perceptions. The PSES data help the employer to understand the pulse of its workforce. The data contain robust information for analysis and merit greater attention by academics in the future despite the limitations created by the Statistics Act.

This paper was founded on six hypotheses advocated by other researchers and has largely confirmed their findings. Important predictors of employee satisfaction are belief in opportunities for promotion; fair classification; adherence to client service standards; life balance; recognition by immediate supervisors; team relationships with colleagues; how well information is shared; and finally, belief in the abilities of senior management to make use of feedback. This study confirmed the findings discussed in Heintzman and Marson (2005). The results of the first five hypotheses, explored through cross tabulations and advocated by various researchers, were also largely confirmed with only minor differences noted in the control variables. In those first five hypotheses, the importance of commitment to organizational success and pride in an organization's work were particularly noteworthy. It will be interesting to see how these relationships might change when the 2008 PSES results are released.

Continued research into this area is needed. Public service reform is a complex and ever-changing process. There is no one solution that will solve all of the problems of a diverse and changing public service today. Small, incremental improvements can work to slowly make a difference. A recent Globe and Mail opinion piece by Karim Bardeesy's echoed this point with an article aptly titled, "We can't fix the public sector in one budget." Bardeesy wrote, "nevertheless, cash-strapped governments feel the need to lead by example and reduce program spending, of which salaries make up a significant part ... Each measure, of course, involves more pain than gain" (Bardeesy 2010). Cutting funding to the public sector in the name of efficiency is not the solution. The public service needs to change the way it markets itself to new recruits and emphasize the opportunity to contribute and make a difference. Initiatives that promote recognition, collaboration and increased communication will help to improve employee satisfaction.

Looking to other countries can continue to offer insight and ideas into best practices that the Canadian public service could adopt to improve efficiency, value for tax dollars and employee satisfaction. The goal of greater efficiency cannot be achieved at the expense of decreasing employee morale because that potentially leads to a public service that will not be equipped to face the challenges of twenty-first century government. These two goals must be achieved hand-in-hand in order to produce an efficient, high-performing, public service that becomes an employer of choice and an institution of which Canadians can be proud.

Appendix 1. Some technical clarifications relating to the employee satisfaction index

The measurement reliability of an index consisting of 4 variables was tested using Cronbach's Alpha. Generally, if Alpha is over 0.7, it is thought to be reasonable to combine variables in an index. The particular figure for Alpha in this case was 0.745 and, therefore, the index had reasonable measurement properties.

Cronbach's Alpha is used to determine the reliability of a set of items in measuring an underlying construct. If several variables tend to be highly correlated, they will yield a high reliability estimate based on Cronbach's Alpha. Alpha is commonly, but not necessarily, related to indicators of whether a set of variables relate to the same underlying construct or dimension of meaning, but this is a secondary and imperfect implication of Alpha.

In order to ensure that the employee satisfaction index was primarily capturing one dimension of content and meaning, a principal components analysis was also conducted. This also produced favourable results. The specific values relevant to assessing the quality of the principal components analysis are as follows:

1. Kaiser-Myeyer-Olkin Measure of Sampling Adequacy (KMO) = 0.707 which is considered good but not superb.

2. Bartlett's Test of Sphericity was significant at the 0.000 level indicating that there were suitable correlations among the underlying variables in the index.

3. Only one component had an Eigen Value of greater than 1 and it accounted for 56.67% of the variance in the underlying variables.

This shows that the index is essentially one-dimensional and can be used to accurately describe employee satisfaction.

References

Baird, K., and A. Cote. 2007. "A vital national institution? what a cross-section of canadians think about the Prospects for Canada's public service in the 21st century." Public Policy Forum 1-25.

Bardeesy, Karim. 2010. "We can't fix the public sector in one budget." The Globe and Mail. Retrieved 19 March 2010 from http://www.theglobeandmail.com/news/opinions/wecant-fix-the-public-sector- in-one-budget/article1488640/

Bourgault, Jacques, and Mary Gusella. 2001. "Performance, pride and recognition in the Canadian federal civil service." International Review of Administrative Sciences 67: 29-47.

Canada. Commission of Inquiry into the Sponsorship Program and Advertising Activities. 2005. Who is Responsible? Ottawa: Public Works and Government Services.

Duxbury, L., S. Lyons, and C. Higgins. 2006. "A comparison of the values and commitment of private sector, public sector, and parapublic sector employees." Public Administration Review July: 605-618.

Harwood, P. 1998. "Employee surveys in the public service: Experiences and success factors." Canadian Centre for Management Development 20 (1): 1-64.

Heintzman, R., and B. Marson. 2005. "People, service and trust: is there a public sector service value chain?" International Review of Administrative Sciences 71 (4): 549-575.

Kernaghan, K. 2001. "An honour to be coveted: Pride, recognition, and public service." Canadian Public Administration 44 (1): 67-83.

Lee, C., L. Duxbury, and C. Higgins. 1994. "Employed mothers: Balancing work and family life." Canadian Centre for Management Development 15: 1-37.

Lynch, K. 2009. Perceptions and Realities of Today's Public Service. Berlin, Germany.

Malloy, J. 2004. "The next generation? Recruitment and renewal in the federal public service." In How Ottawa Spends, 2004-2005: Mandate Change and Continuity in the Paul Martin Era, edited by G. Bruce Doern (pp. 277-295). Montreal and Kingston: McGill-Queen's University Press.

Phillips, S., B. Little, and L. Goodine. 1997. "Reconsidering gender and public administration: five steps beyond conventional research." Canadian Public Administration 40 (4): 563-581.

Statistics Canada. 2006. Retrieved 28 September 2009 from Public Service Employee Survey: http://www.statcan.gc.ca/survey.

Zussman, D. 1982. "The image of the public service in Canada." Canadian Public Administration 25 (1): 63-80.

Alexandra Hickey is a graduate of the Bachelor of Public Affairs and Policy Management Program, Carleton University, and currently works for Human Resources and Skills Development Canada. Scott Edward Bennett is associate professor, Department of Political Science, Carleton University. The authors thank the Journal's anonymous reviewers for their helpful comments on an earlier version of this manuscript. This article reflects the analysis and conclusions of the authors, and it is not necessarily representative of the perspectives of their employers.

Table 1. Components of Dependent Variable Employee Satisfaction Index

Question #                   PSES 2005 variables

Q033         I am proud of work carried out in my unit.
Q086         I am committed to making my organization a success.
Q088         My organization is a good place to work.
Q089         I am satisfied with my career.

Cronbach's Alpha equals 0.745. The variables primarily load on one
dimension. See the appendix for details.

Table 2. Independent Regression Variables

Peel Study indicators                PSES 2005 variables

Opportunities for        Q47--I believe I have opportunities for
advancement              promotion within the public service, given
                         my education, skills and experience.

Fair pay and benefits    Q005--I am classified fairly (my current
                         group and level) compared with others doing
                         similar work in my organization or elsewhere
                         in the public service.

Good customer service    Q073--My work unit regularly applies the
                         client service standards.

Reasonable work load     Q013--I can complete my assigned workload
                         during my regular working hours.

Good relations with      Q021--I get adequate recognition from my
supervisor               immediate supervisor when I do a good job.

Good relations with      Q034--In my work unit, we work cooperatively
colleagues               as a team.

Effective internal       Q078--I feel that senior management does a
communications           good job of sharing information.

Positive perception of   Q084--I believe senior management will
senior management        resolve concerns in this survey.

Table 3. Summary of Cross Tabulation Results

                 Chi-square     Coefficient of
Hypothesis #    significance     association     Strength   Direction

      1         0.000           -0.105 (Gamma)   Weak       Negative
Desire to       (n =103,522)
leave the
public
service
within five
years, by
pride in
organization

      2         0.000           0.496 (Gamma)    Strong     Positive
Career          (n =101,017)                     (rounded
satisfaction                                     upward)
by
perceptions
of
recognition
by immediate
supervisors

  2 Control     Men             0.512 (Gamma)    Strong     Positive
Career          0.000
satisfaction    (n = 42,514)
by
perceptions     Women           0.475 (Gamma)    Moderate   Positive
of              0.000
recognition     (n = 57,267)
by immediate
supervisors
controlling
for gender

      3         0.000           0.337 (Gamma)    Moderate   Positive
Overall         (n =103,143)
career
satisfaction
by life
balance

  3 Control     Men             0.354 (Gamma)    Moderate   Positive
Overall         0.000
career          (n = 44,693)
satisfaction
by life         Women           0.321 (Gamma)    Moderate   Positive
balance         0.000
controlling     (n = 57,085)
for gender

      4         0.000           0.031            Weak
Commitment to   (n = 102,738)   (Cramer's V)
organization
success by
gender

      5         0.000           0.524 (Gamma)    Strong     Positive
Commitment to   (n = 102,648)
organization
success by
pride in work
of public
service unit

Significance levels of less than 0.05 are taken to be conventionally
significant. Measures of association values less than 0.3 are taken
as weak, 0.3 up to 0.5 moderate, and greater than 0.5, strong.

Table 4. Regression Results with Employee Satisfaction Index as the
Dependent Variable

                                   Standard
Independent variables       B       error      Beta    Significance

Q47) I believe I have      0.334   0.007       0.141      0.000
opportunities for
promotion within the
PS, given my education,
skills and experience

Q005) I am classified      0.173   0.006       0.084      0.000
fairly (my group and
level) compared with
others doing similar
work in my organization
or elsewhere in the PS

Q073) My work unit         0.363   0.008       0.131      0.000
regularly applies the
client service
standards

Q013) I can complete my   -0.014   0.006      -0.006      0.0027
assigned workload
during my regular
working hours

Q021) I get adequate       0.262   0.007       0.108      0.000
recognition from my
immediate supervisor
when I do a good job

Q034) In my work unit,     0.754   0.008       0.273      0.000
we work cooperatively
as a team

Q078) I feel that          0.315   0.009       0.129      0.000
senior management does
a good job of sharing
information

Q084) I believe that       0.552   0.008       0.240      0.000
Senior Management will
try to resolve the
concerns raised in this
survey

Adjusted R Square = 0.549. Significance level of adjusted R square =
0.000 (n = 68,082)