Canada’s Public Service Employee Survey: using advanced data analytics to focus workplace culture change
2021/11/05 2 Comments
Good to have more people like Philip Lillies looking at the Survey and probing the meaning of the findings, whether by organization or group, along with combining findings with the Staffing and Non-Partisanship Survey (SNPS). More complex analysis than I can do!:
Since 2005, summaries of overall Public Service Employee Survey (PSES) results have been posted on the Government of Canada open portal. The summaries of overall results have facilitated the analysis of shortcomings in the culture of the workplace by human resource personnel, internal auditors, and researchers. Notably, two researchers, Andrew Griffith and Jake Cole have recently published in The Hill Times analyses that have complemented the summaries of overall results posted on the Government of Canada portal.
However, summaries of overall results have a glaring deficiency: they may indicate that corrective action is necessary, but they provide insufficient guidance as to what action might be most effective or which employee groups are most in need of this action. Comparison of variations across departments and across employee groups can make up for this insufficiency. These comparisons can be used to derive associations between responses; these associations often indicate the potential causes and consequences of the variations across groups. And causes and consequences are an important guide to action.
In 2020, I retired from my position as a senior internal auditor in the public service. During my first year of retirement, I have endeavoured to make up for shortcomings in the usual analysis tools by writing a Python program that had the capacity to use response variations to find associations between responses. It then attributed these associations to causes and consequences for particular departments and employee groups. In what follows, I build on the work of Griffith and Cole by presenting some examples of what I have found using my Python program.
Measuring and improving happiness
Cole states that the pandemic has been good for public service employees. According to him, “Whatever the reason, they are a happier bunch.” There are no questions about happiness in the PSES, but many experts, including Cole suggest that employee engagement is a good indicator of happiness. Under the theme of “engagement,” the PSES has seven questions. Across the entire public service, there are, nonetheless, variations in the level of engagement. By focusing corrective action on those groups that show the lowest scores to engagement questions and to associated questions, we can improve the efficiency of the corrective action.
So, from among these seven engagement-themed questions, here are the four that show the most variation across the public service:
- Q11: Feeling valued at work.
- Q50: Recommendation that my department is a great place to work.
- Q51: Satisfied with my department or agency.
- Q52: Prefer my workplace over others in the federal public service.
But to take focused corrective action we need to know which employees in which departments are suffering from lack of engagement. It turns out that there are eight departments that show below average scores in responses to these four questions. Questions associated with these four questions will be the questions from which we can derive causes and consequences and those groups with below average responses to these four questions will be the groups to which corrective action needs to be applied.
To take a concrete example, it turns out that border services employees are one of the most disengaged groups within the Canadian Border Services Agency and the potential causes of their disengagement can be found in the below-average scores of their responses to career-related questions, such as:
- Q41: my department or agency does a good job of supporting employee career development.
Corrective action can be applied accordingly.
Combining results from two surveys
Another important government survey is the Staffing and Non-Partisanship Survey (SNPS), which is also directed at employees, and publishes its results on the government’s open portal in a separate cycle to the PSES. Using Python to combine results from the two surveys is both trivial and insightful.
Table 1 lists the ethical questions that show a high variation in response scores when the SNPS is combined with the PSES. Associated with all of these questions from the PSES, except one, are two questions from the SNPS:
QALL_05D: The process of selecting a person for a position is done fairly. QALL_05B: I believe that we hire people who can do the job. Not only does the association of these questions with so many of the PSES ethical questions highlight the importance of the work of the Public Service Commission, which is responsible for staffing practices, but one would also be inclined to draw the conclusion that these SNPS questions are two important ethical questions that should be included in the PSES rather than the SNPS.
Table 1: Questions from the PSES with High Variation when SNPS is combined with PSES
Ethical workplace Q19: Satisfactory resolution of interpersonal issues. Q38: Know where to go for help on ethical issues. Q39: Promotion of values and ethics. Q40: No fear of reprisal. Leadership: senior management Q31: Leadership by ethical example. Q32: Confidence in senior management. Q33: Effectiveness and timeliness of decisions. Q34: Effectiveness of essential information flows. Harassment Q60: Satisfactory harassment resolution. Q61: Satisfactory harassment prevention program. Discrimination Q63-B: Discrimination from individuals with authority over me. Q67: Satisfactory discrimination resolution. Q68: Satisfactory discrimination prevention program. Empowerment of Black employees
Griffith, in his November 2019 article, reaches the conclusion that Black employees are among the least empowered. His conclusion is based on the overall scores of Black employee responses to organizational culture indicators in the PSES 2019 survey. Interestingly, my Python program indicates that there are nine departments that show not only below average scores in responses to these empowerment questions, but also below average scores in their responses to questions associated with these questions. What is surprising is that among these nine departments are the Public Service Commission of Canada, the Military Police Complaints Commission of Canada, the Courts Administration Service, Canadian Human Rights Commission, the Public Prosecution Service of Canada, and the Social Sciences and Humanities Research Council. These are ethically oriented regulatory and research bodies that should be the first to understand the mechanisms and implications of discrimination; hence, should already be taking the necessary corrective actions. Perhaps, these results indicate that understanding is only a first step in overcoming discrimination, in which case discovering what corrective actions are required to go beyond understanding points to the need for further investigation.
Conclusion
I can only agree with Cole that the PSES provides a rich source of information that, if properly assessed and acted on, could result in positive changes for the employees and subsequently for the Canadians they are there to serve. However, assessment cannot be limited to discussion and comparison of overall results. As I hope the examples provided show, this rich source of information can only be fully exploited by making use of computerized data analytics techniques that highlight associations between responses and pinpoint employee groups where follow-up is needed. Nonetheless, associations should not be confused with definitive results; rather they should be taken as guidance for further assessment and investigation. Speaking from my own professional experience, I would say that the need for this informed cultural analysis provides an exciting opportunity for the next generation of internal auditors if they can rise to the challenge.