Audit of 2006 Census Related to Selected Security, Administrative and Quality Practices

Archived information

Archived information is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

 

Final Report
Approved by the Internal Audit Committee on October 10, 2007

Internal Audit Division
June 2007

Auditor’s statement

We have completed the Audit of 2006 Census Related to Selected Security, Administrative and Quality Practices. The objectives were: to ensure that pay is processed in an accurate and timely way, in accordance with established legislation, policies and procedures; to ensure that personnel screening for new census Statistics Act employees is carried out according to the Government Security Policy (2002) and Personnel Security Standard (1994); that they swear the oath of office1 under the Statistics Act and have a reliability status before accessing a computing device connected to any Statistics Canada network; and to ensure effective management controls related to monitoring of Census Help Line telephone operators either responding to the Canadian public or contacting them for information as part of failed edit follow-up.

Key activities focussed on visiting ten 2006 Census field sites, examining a sample of records, interviewing officials at every site, reviewing documents, and in the cases of the three Census Help-line sites, administering a questionnaire to supervisors.

Pay was processed in an accurate way, mainly in accordance with established legislation, policies and procedures. However, payments to employees were not always timely as streamlining and adjustment of some controls is necessary to show evidence of due diligence.

Personnel screening for new census Statistics Act employees met most of the requirements of the Government Security Policy (2002) and Personnel Security Standard (1994). Reliability checks were done before employees began work, although improvements to some tools and procedures are required for 2011 Census field operations. Employees swear the oath of office under the Statistics Act and have a reliability status before accessing a computing device connected to any Statistics Canada network.

Overall, management controls were in place related to monitoring of Census Help Line telephone operators either responding to the Canadian public or contacting them for information as part of failed edit follow-up, although their effectiveness should be improved.

These conclusions are based on the assessment of findings against pre-established criteria and agreed to by the Internal Audit Committee on April 5, 2006 and reflect audit work conducted principally between June and July 2006.

In my opinion, sufficient and appropriate audit work has been performed and evidence gathered to support the conclusions contained in this audit report.

Francine Hardy, Chief Audit Executive

Introduction

Gen 1. Canada’s census is the largest statistical undertaking in the country. Every five years, the census obtains information from the entire population of Canada. This information is used to determine representation in the House of Commons, calculate fiscal transfer payments, and develop and monitor social and other programs of the federal, provincial and municipal governments in this country2. May 16, 2006 was Census Day and required a huge collection and processing effort to handle questionnaires covering over 13 million dwellings. Approximately 25,000 temporary employees were needed, along with physical space in which to work or through which administrative support was provided.

Background

Gen 2. Engaging large numbers of temporary employees to work for a very short time, many of whom were new to Statistics Canada, meant challenges to make sure that: everyone followed security practices; employees were accurately and promptly paid, and the work produced is of high quality. Statistics Canada maintains its reputation and the confidence of Canadians when: security standards are respected, there is due diligence in administrative practices, and high-quality service is provided to respondents.

Gen 3. Statistics Canada introduced significant changes to its census collection and processing activities since 2001, including the use of contractors to provide key systems.  The agency has taken unprecedented measures to ensure that systems built by contractors are secure. For example, it established an independent task force to oversee three independent verifications. Part of this work included physical site inspection at the Toronto, Edmonton and Moncton census help line sites. Statistics Canada has conducted internal security assessments and contracted for a security inspection of the Data Processing Centre. Therefore, this audit focused on another security element—personnel screening. Canadians expect that anyone collecting or processing their census information have the reliability status expected of all government employees. While the probability that an incident would occur with an unscreened employee is low, the potential impact on Statistics Canada’s reputation could be high.

Gen 4. While the census has offered a help line for respondents in the past, for 2006 its services were expanded. Not only were there between 250 and 350 operators at each of the three sites providing general assistance to callers, they contacted respondents to do failed edit follow-up3.  They are a significant part of the public face and professional services involving respondents are important—whether related to calls from the public or those that the operators make to respondents. Canadians deserve and expect high-quality exchanges with Statistics Canada, and in turn, this has a payback for the Agency’s reputation and success.  Monitoring this activity was a planned supervisory function requiring effort and diligence especially given tight time frames and temporary employees in an operation as large as the census. 

Gen 5. The pay process must result in accurate and prompt payments—a task made more difficult because of large volumes being hired within a short space of time to work for a matter of weeks only in most cases. The additional risk that this creates is greater because the employees are not public servants and are therefore not paid through existing channels. For this census, Statistics Canada has built a new pay system for temporary employees.

Gen 6. Therefore, the purpose of this audit is to provide Statistics Canada management with an assessment of accuracy and timeliness of pay processes, security-related personnel screening and management controls for quality of service delivered by census help-line operators.

Objectives

Related to pay

  1. Gen 7. To ensure that pay is processed in an accurate and timely way in accordance with established legislation, policies and procedures.

Related to security

  1. Gen 8. To ensure that personnel screening for new census Statistics Act employees is carried out according to the Government Security Policy (2002) and Personnel Security Standard (1994); that they swear the oath of office4 under the Statistics Act and have a reliability status before accessing a computing device connected to any Statistics Canada network. (Source: Statistics Canada Security Practices Manual, chapter 5.)

Related to Census Help Line offices

  1. Gen 9. To ensure effective management controls related to monitoring of Census Help Line telephone operators either responding to the Canadian public or contacting them for information as part of failed edit follow-up.

Scope

Gen 10. For objectives related to personnel screening and pay, the scope applies to temporary census employees hired for the field operations5 task under the Statistics Act.

Gen 11. Only some aspects of the Personnel Security Standard related to security screening when employees are hired will be assessed, not other sections; for example, those related to updates, disposal of records, and termination.

Criteria

Pay
Gen 12. We expected to find:

  • payment of salaries and wages was accurate and timely
  • roles and responsibilities clearly defined
  • appropriate controls in place

Security
Gen 13. For the personnel screening objective we expected to find:

  • a security screening program in place that complies with the Government Security Policy (2002) and the Personnel Screening Standard (1994) when hiring census Statistics Act employees
  • required checks are authorized, consented to and completed before individuals occupy positions
  • a signed oath or affirmation of office and secrecy under the Statistics Act
  • a user ID was appropriately obtained after a reliability status was granted

Monitoring CHL operators
Gen 14. We expected to find:

  • operators were adequately trained
  • supervisors conducted monitoring as planned
  • evaluations were properly stored on personnel files (according to Statistics Canada’s Security Practices Manual)
  • managers had effective controls to assess monitoring results
  • appropriate action was taken when results were unacceptable
  • management practices in place for public complaints concerning operators

Methodology

Gen 15. Ten sites were visited from early June to mid-July 2006. Eight were local census offices: Dartmouth, Québec, Trois-Rivières, Montreal, Whitby, Sudbury, Hamilton and Edmonton, and all three census help-line sites located in Moncton, Toronto and Edmonton. Edmonton was treated as one site with two components. For each site, a random sample of in-scope employees was selected based on personnel screening information from the ‘Field Operations Project (FOP) System’ in May 2006 and to which a randomly selected pay transaction was matched. Pay batches in which these transactions were grouped for transmission to PWGSC were also selected. Site samples were augmented with some auditor-selected cases.

Gen 16. For personnel screening, 1375 cases were examined and 1315 cases for pay . There were 320 Census Help Line (CHL) operator cases examined.

Gen 17. Using the samples, associated paperwork was examined: for personnel screening—consent and authorization forms, (Form 26B); for pay—pay forms pertaining to the selected transactions and the pay batches in which they were grouped for further processing; and for CHL monitoring—monitoring and observation feedback forms, hereafter called ‘Form 87’.

Gen 18. In the case of pay, an analysis of Common Departmental Financial System (CDFS) and census pay system claims and payment dates was also done. This analysis was used mainly to study the timeliness aspect of the pay process. In total, 311,000 claim payments were part of the study.

Gen 19. Moncton CHL site was the first site visited in early June and it acted as a pilot site. Since it was early with respect to the CHL objective, a follow-up interview was done with the site manager in August 2006. Even with a pilot, methodological adaptations were necessary in the field in response to local circumstances, particularly for the CHL objective. For example, in Edmonton Forms 87 were not kept in personnel files6 as expected, but were organized in cabinets by first name—all the ‘A’ to ‘D’ were together and unsorted. In Toronto, the file folders were appropriately in the possession of the supervisor, some of whom were not available when we were on site. Therefore, the sample size was reduced and ability to judge the frequency of monitoring was more limited than anticipated, particularly in Edmonton, where recording of monitoring events per operator was limited to maximum of four.

Gen 20. The examination of paperwork was complemented by documents review, interviews and, in the case of the CHL sites, a questionnaire. This was administered to supervisors of CHL operators asking them to agree or disagree with statements focussing on monitoring tools being used.

Findings and recommendations

Gen 21. Chart 1 on the next page provides an overview of the audit results for the three objectives. Following the chart, the detailed findings are presented in three sections, with the recommendations pertaining to each objective at the end of its section.

Chart 1: Overview of audit results (G=green for OK; Y=yellow for caution)
Criteria Result Rating Residual Risk Assessment7
Probability Impact
On pay
Proper controls in place Fully compliant G L L
Roles & respons-ibilities defined Mainly compliant G L L
Wage payment accurate & timely Partially compliant Y M M
On personnel screening
Statistics Act oath is taken This was consistently done across all sites. G L L
Program is in place Procedures exist, including when adverse results arise; information is held securely, & employees are informed of security. Improvements needed to implement a briefing form; improve procedures related to consent & authorization form, with emphasis on who is authorized to grant the status. The high impact assessment recognizes a serious risk to reputation in a worst case scenario if an employee wasn't properly screened and an incident occurred. Y L H
Consent given, checks done, authorized Mostly OK; inconsistency in authorization at some sites is a consequence of procedures, already noted above G8 L L
User ID given after screening completed Reliability status checks were done in a timely manner, so Statistics Act employees requiring access to a Statistics Canada network (mainly those working at CHL sites) were ready to have a user ID. Appropriate procedures were in place. G L L
On CHL monitoring
Operators trained Operators were adequately trained. G L L
Supervisors monitoring Opportunities to improve monitoring; Form 87 monitoring benchmark of 'once-a-week' was not met Y M L
Effective controls Managers had controls but there is room to improve them, e.g., quantify information on extent of monitoring Y M L
Appropriate action taken Supervisors and managers were dealing with unacceptable results when these were identified G L L
Complaints handled Management practices are in place to handle these G L L
Evaluations securely stored Form 87 monitoring evaluations were stored in cabinets that do not meet Statistics Canada's standards Y L L

Pay

  1. Controls and processes adequate to ensure accuracy
    Pay 6. In general, controls and processes were found to be adequate to ensure accuracy of pay transactions. For example, we noted that processes were in place to capture overpayments. Overpayments were rare and were recuperated on an employee’s future paycheque or directly from the person if no longer employed.  This is what we expected to find.

  2. Roles and responsibilities clearly defined in most cases
    Pay 7 In most cases roles and responsibilities were clearly defined. For example, sections 33 and 34, and emergency checks were handled appropriately. However, a pay form (Form 1c), introduced during the pay process did not have all the controls found with other pay forms. According to the Census Management Office, the use of this form will be discontinued. There was also an issue with pay clerks doing their own verification but with other controls in place, such as reconciliation, the risk is mitigated.

  3. Signature cards not used
    Pay 8. Signature cards of supervisors and above were not being used by the pay personnel.  These cards were either filed in employee personnel files, or sent to the regional office, or were simply not filled out so that no signature specimen was available for the pay staff to confirm proper signature. A few local census office managers indicated that the pay staff became familiar with the signatures of appropriate authorities, but this does not constitute a good control. Signature cards were not used mainly because of a lack of understanding of their use, but also for operational considerations such as number of cards to retrieve and analyse, and time constraints. A potential impact of not using signature cards is the risk of fraud.  This risk, however, is greatly mitigated by other controls in place such as reconciliation, pay clerks’ verification and recognition of signatures.  If electronic approval of forms is not introduced for the 2011 Census, a complete set of signatures should be provided to the pay staff.

  4. Some signatures missing but may be redundant
    Pay 9. The majority of pay claims forms—about three quarters—were missing at least one signature date, for example dates accompanying enumerator signature, supervisor signature or field operation supervisor signature (when required) and pay clerk verification. In cases where dates were present at the various steps of the processes, the process flow was good. Roughly one third of Form 33s9 were not signed by supervisors. Some of these forms were not signed by supervisors due to the distance between enumerators and supervisors and the reliance on the daily pay summary form (Form 32) which was signed by the supervisors. The impact of supervisors not signing pay claim forms is the potential risk of fraud. Since Form 32s were signed by the supervisors, however, the risk of fraud is certainly reduced significantly. The need for the supervisors to sign both the pay claim form and the pay summary form (Form 32) may be redundant, or at least perceived to be redundant which may have contributed to the missing signatures on Form 33. This question of redundancy should be reviewed in order to ensure better control related to pay claims.

  5. Timeliness can be better controlled
    Pay 10. When we analyzed timeliness, we looked at 311,000 claim payments. We found that a target of payment within 2 weeks after work period was set in the pay manual.  However, employees, from their terms and conditions of employment, had expectations of receiving their paycheque within 3 weeks after the work period and that is the target that we looked at. This target was met 74% of the time. There is room for improvement. Here are a few more observations on timeliness:

    • Pay 11. Decentralized offices were more efficient than centralized ones.  Decentralized offices met the expectations 78% of the time while centralized offices met the target 69% of the time. The relationship holds true at the other end of the spectrum as the number of late payments and the delay in the payments were greater in centralized offices.
    • Pay 12. There was a slight improvement with time from the earlier to later rush period. From early June to end of August the average number of days after the reference period to get paid went from 21.5 days to 17.5 days.  This could be explained by improvements introduced during this period (e.g. relaxation of reconciliation which was in part planned) and the reorganization of the work that occurred in various local offices.
    • Pay 13. A regression analysis between volume of transactions in an office and timeliness showed little correlation (r = .45).
    • Pay 14. The time it took to complete all pay transactions indicates that there was a long tail.  This is evidenced by the fact that 10,000 claims, representing 3% of the total claims, took 9 weeks or more after the work period to get finalized and 1,000 of these claims (0.3%) took 17 weeks or more. A report that listed late claims was sent to the regions on a regular basis. We were told by the Census Management Office that the delays were caused mainly by enumerators not submitting their claims in a timely fashion or because of the poor quality of the claims. Nevertheless, we can conclude from this that a lack of control existed in terms of the number of weeks to process claims.  We could not find documentation explaining the causes of this long tail.

    Timeliness was affected by:

    • Pay 15. Centralization of offices. As noted above decentralized offices were timelier for the two sets of standards as well as for all payments.
    • Pay 16. Complexity and variety of pay forms.  With the amount of work required in reconciliation of the various pay forms, it is clear that these multiple forms were not well understood by some enumerators. We were told by just about all the local managers that this was the case.
    • Pay 17. Poor control in the follow up and documentation of late claims. The evidence is the long tail and problem cases that resulted in newspaper articles.
    • Pay 18. Stringent process as timeliness improved when rules were relaxed.  Strong controls are necessary in this area so this aspect cannot be relaxed too much. Need to assess right balance between enough controls and too many controls.
    • Pay 19. Dual purpose of the forms. According to the Census Management Office this is a U.S.  based concept that didn’t work here. This contributed to make the forms more complicated than required for pay purposes only. The Census Management Office said that the function split was done before the end of the census pay processing. The experience from the 2006 Census indicated that forms should be designed for either pay or production purposes separately.

    Pay 20. Not being timely with payment of salaries and wages to employees can be very damaging to employee morale and the reputation of the organization.  Employees not paid in a timely fashion can be demoralized.  In the case of the 2006 Census, delays in paying some of our employees brought newspaper articles which resulted in some negative press.

    Pay 21. We cannot pretend to know precisely what would be the ideal standard for people to get paid on a regular basis (first paycheque followed by regular pays). It is important however to have a control measuring the period to process claims with a mechanism to deal with cases that have exceeded a certain target or, in other words, that have not been completed within a pre-established time frame.  This process should prevent late payments from escalating and becoming an embarrassment to the agency or to cause hardship to the employees. Another aspect of documenting late payments is that the census managers would know with more assurance the cause of the problem and take measures to rectify the situation appropriately.

Pay-related recommendations

Pay 22. Recommendation 1: Reduce number and complexity of pay forms

  • Such measures are needed to reduce or avoid the long delays observed. Separating the pay and production functions on forms of their own could help. The pay process should be simplified and the potential for further automation should be explored.

Pay 23. Recommendation 2: Streamline control related to supervisor signature

  • In cases where the distance between enumerator and supervisor is significant, signatures are difficult to obtain, causing some delay in the pay process.  Electronic approvals of forms could be an option.  On the other hand, this control (supervisor’s signature on Form 33) could be dropped if the signature of the supervisor on the daily summary (Form 32) is judged to be sufficient.

Pay 24. Recommendation 3: Improve follow-up of late claims

  • A control should be added to the pay process to prevent late pay forms from escalating.  After a certain amount of time following a reference period, a follow up process should be activated.  This process must be able to detect the problem, analyze it and find a solution. It is very important to document all the steps taken, as it will become evidence to show due diligence.

Personnel Screening

  1. Statistics Act oath taken
    PS1. Before taking up duties under the Statistics Act, all employees, in accordance with section 6 of that act must take and subscribe an oath or solemn affirmation that they will conform with the requirements of the Act in carrying out their duties and that they will not, without due authority, disclose or make known information that come to their knowledge by reason of their employment.10

    PS2. Statistics Act census employees are signing the oath and administrators are authorizing it. The timing is such that this is done on or before the day the employee begins work or within a day or two after. Procedures were in place to have employees sign this oath and following them lead to a positive result.

  2. Improve delegation & documentation related to screening program
    PS3. A security screening program makes sure that diligence is exercised when hiring employees to avoid those with track records suggesting they cannot meet Statistics Canada’s security requirements. Such a program has several components: procedures covering security screening checks and briefing employees on their security obligations; an authorization process to grant a reliability status, including a mechanism to handle adverse results; and means to hold security screening documentation securely. 

    PS4. 2006 Census field operations had a program to carry out personnel security screening. However, two key changes are needed to ensure consistency with Treasury Board directives when 2011 Census collection field operations begin: Identify who is delegated as responsible to grant a reliability status, under what circumstances, how this will be accomplished, and reflecting the results in modified procedures and tools; and include a formal security briefing, documented through the completion of the required Treasury Board Security Screening Certificate and Briefing Form (TBS/SCT 330-47(rev.2006/02).

    PS5. The remainder of this section elaborates on the security screening program—what was planned. In section 3 starting on p. 13 , a related but distinct criterion is addressed— information about what happened.

    Screening program to identify who grants the reliability status
    PS6. A key control, demonstrating due diligence in properly security screening those being hired, is a well-completed consent and authorization form—Form 26B, called Census Reliability Assessment by 2006 Census. It documents key program steps: getting candidates’ consent to do security screening; checking identification, conducting a reference check; and doing a criminal record check through the RCMP.  Other activities support this, such as procedures for when fingerprints were required; and responsibilities for decision-making when there were adverse results or situations where a screening delay was unavoidable. However, an important aspect was not included on the form—identifying who is delegated to grant the reliability status.

    PS7. An old version of Treasury Board’s consent and authorization form was used and consequently did not contain changes since before 2002, for example the latest wording changes concerning the section of the form covering the Privacy Act. While the risk from using an old form is low since the relevant checks have not changed, good house-keeping requires that the 2011 census form reflect the Treasury Board consent and authorization form. Whatever version is in force when the Census 2011 schedule requires would be appropriate. Titling the form appropriately—consent and authorization form—would clarify its purpose. Nor did the 2006 census form contain space for authorization of the reliability status, separate from certification that the checks have been carried out. 2006 Census officials were not informed of changes in the Treasury Board form when they provided their form to Data Access and Control Services Division for vetting in December 2004.

    PS8. Procedures should describe who is authorized to certify that all steps have been carried out and who is authorized to give the reliability status. Normally, reliability status approval is given by an authorized organizational security official only. Clearly, when many thousands of people are hired in a short space of time, Statistics Canada needs to delegate this responsibility to the appropriate Statistics Canada official in line with operational constraints.

    PS9. The RCMP conducts criminal-record name checks and processes fingerprints as necessary to determine if a candidate has a criminal record or not. Statistics Canada uses these results along with the results of other checks it has done to decide whether to grant a reliability status. The RCMP does not make this decision. However, some standard procedures for field operations employees leave this impression which should be addressed. On the other hand, documentation on the role and responsibilities of regional security committees appropriately describes decision-making as belonging to Statistics Canada.

    Regional security committees a positive feature
    PS10. The 2006 Census established regional security committees to handle two types of cases where hiring was desirable: where criminal record name check results were inconclusive and fingerprint results were not yet available or where the candidate had a criminal record.  Having the regional security committee authorize the decision to grant a reliability status on the consent and authorization form under these circumstances would improve control.

    PS11. Using regional security committees is a Statistics Canada initiative and is not required by Treasury Board.  It was a good strategy for the 2006 Census and improving it for 2011 would be beneficial. Committees were to report their results on a monthly basis although this was spotty, particularly when collection was at its peak due to more pressing priorities. Reports were required so that they could be forwarded to the departmental security officer in DACS for documentation purposes. These reports could be seen as a control to assess how the regional committees were working, but they have not been used for any reason up to early May 2007. Their contents vary by region and coverage appears spotty in the absence of nil reports. Reports provide no differentiation between these two types of case.  If regional security review committees are required in 2011, their reports have the potential to provide a consolidated, corporate picture for the census once documentation requirements are reviewed and improved so that they are standardized and designed to meet the identified need. The need for regional committees may evaporate if software capable of providing results on fingerprints in a few days is in use by the RCMP and available to Statistics Canada when 2011 Census fieldwork begins.

    PS12. Data Access and Control Services Division officials had exchanges with Treasury Board about the acceptability of regional security committees granting interim reliability statuses in the context of a draft Treasury Board personnel security screening standard. However, results were not documented at the time, though in effect, Statistics Canada policy was being created. The draft standard of November 2006 does not explicitly cover the Statistics Canada census circumstances although there is one reasonable interpretation that it implicitly covers them. This should be confirmed and documented in preparation for the 2011 Census along with any other clarifications that might be required.

    Employees informed about confidentiality, but no formal briefing
    PS13. Employees were informed about the importance of security and the importance of protecting respondent confidentiality through written procedures. Employees took an oath under the Statistics Act to protect respondent confidentiality. Terms and conditions of employment contained similar information.

    PS14. However, no formal briefing was done as is required by the Treasury Board directive on Personnel Security Screening (1994) and as would be evidenced by the completion of the required Treasury Board security screening certificate and briefing form. Giving a verbal briefing to complement signing the form is best practice. Taking an oath and being briefed on government security requirements are two different, although complementary events, in Statistics Canada’s case. However, one cannot substitute for the other. Further, the security screening certificate and briefing form communicates to the employee what his or her authorized level is.

    PS15. Statistics Canada is not using the Treasury Board security screening certificate and briefing form for any employee holding a reliability status, so it is not surprising that 2006 Census officials preparing procedures for steps related to personnel screening were not aware of its existence. They did not get advice from departmental security officials with respect to the consent and authorization form when they asked for assistance concerning personnel screening procedures simply because it has not been used in Statistics Canada. Complying with this requirement through the delivery of a verbal briefing acknowledged by each employee in completing the briefing form would describe the importance of security at Statistics Canada and reinforce the fundamental value the organization places on respecting confidentiality.

    PS 16. Incorporating the briefing form into the 2011 Census operations will require careful consideration to carry out this task for thousands of employees within a short time-period. Census officials and the Agency’s security officials can draw upon advice from Treasury Board in interpreting the Personnel Security Standard as it stands when 2011 Census operational decisions must be taken. The exigencies of a national census impose operational constraints that may necessitate a different process than what one might put in place under other circumstances.

    Cabinets are locked but keys are not secure
    PS17. The screening information that was gathered and generated was held securely11 in that cabinets were locked and perimeter security was good in general but weak points exist—one local census office did not have sufficient cabinets and so employee files were held in boxes in unlocked offices. LCOs need to improve key management so that keys to cabinets are not ‘hidden’ in easily accessible places. While associated risks are low, in an organization that prides itself on its ability to protect confidentiality, practising good key management contributes to reinforcing the Agency’s core value of protecting confidentiality.

  3. Checks were consented to and completed in a timely way
    PS18. In the previous section we noted ways to improve the screening program; for example, clear delegation of authorities is needed. This section shows what happened in practice.

    PS19. Employees consented to personnel screening checks; the checks were done12; and, for the most part completed by the time employees started work. Regional security committees made decisions about criminal-record cases and cases requiring an interim reliability status, in keeping with departmental procedures. Reliability statuses were not authorized properly at several sites, however.

    Consent given for required checks
    PS20. Personnel screening checks were conducted with employees’ consent. Completion of this portion of the consent and authorization form related to personnel screening (Form 26B) was built into the hiring process.

    PS21. Not every case was checked, however, since these forms had not yet been sent in from the field. Form 26B was not found in 50 of about 1380 cases, and almost half (23) from Québec LCO; 11 from Edmonton and 9 from Sudbury. A few field staff did not respond to direction to forward it, according to interview results in Sudbury and Québec. These three sites covered vast recruitment areas, where field staff was far from local census offices, making control more difficult.

    Checks were completed
    PS22. There are four checks to be completed, pertaining to: personal identification; employment history; references; and criminal record. The initials of the hiring officer doing the check are to be recorded on Form 26B. This evidence facilitates the work of the person with responsibility to certify that the checks are done and authorize the reliability status. When they are missing, additional confirming work must be conducted by the authorizer, duplicating effort.

    PS23. Personal identification was well-checked everywhere. Precise instructions existed, the form was handy to the hiring official and the check was viewed as important. In Montreal, Whitby, and Moncton the employment history, reference, and criminal record checks were indicated as complete, using the consent and authorization form (Form 26B) as the evidence. Their success is attributed to good control over the process, with recruitment staff close to hand.

    PS24. However, these three checks were less well-completed at other sites—again, using the consent and authorization form as the basis. This tool is not completely fulfilling its role as a key management control. It was seen as redundant with information in the census automated recruitment system (CARS), procedures were unclear—they did not all map to the form (Form 26B), and control was more difficult when areas were vast. However, managers maintained that checks were done in these cases too.

    PS25. That the reference and criminal record checks were done is substantiated by information in CARS, the census automated recruitment system. Records contained the date by which references were checked and the date the criminal record name check (CRNC) was done. 

    Authorization not consistently done
    PS26. Certification was not properly authorized in about half the cases13. Assistant Managers of Recruitment (AMR) were to sign the consent and authorization form, although procedures did not make this clear. Even when AMR were signing this form—and presumably certifying that all checks had been done—the signing date was frequently the same day as the candidate gave consent. This is impossible since it takes 24 hours at least to do a criminal record name check. Others involved in the hiring process signed in the authorization space; sometimes these were field staff—crew leaders.

    PS27. At some sites authorization was properly done by the Assistant Manager Recruitment (AMR)—in Montreal, Moncton, Dartmouth, and to a lesser extent in Trois-Rivières. In most other sites, another employee was usually signing. This was acceptable to AMRs because they maintained that they were doing the checks even if Form 26B paperwork did not reflect this. Hiring officials were not well-served by all aspects of the procedures, substantially contributing to difficulties. Despite this, officials recognized the importance of due diligence in this process and their role in it.

    PS28. Authorization is the step that grants a reliability status. As such, it is a key control; when it is missing, it raises questions about managerial diligence and transparent decision-making. Statistics Canada not only wants to exercise due diligence, but to be seen to do so through sound documentation.

    Employees were security screened before starting work
    PS29. Overall, security screening was completed before employees started work or the regional security committee made a decision in keeping with its responsibilities. In a handful of cases in Edmonton, Sudbury and Québec, employees were hired without proper procedures being followed.

    PS30. The 2006 Census hired many thousands of employees all across Canada for short-term work within tight time frames. Our sample resulted in only a handful of problem cases, and one might be tempted to discount them. But Statistics Canada holds confidentiality as a very important value, and proper personnel screening helps maintain that value. Therefore, the impact is high if a case where due diligence was not exercised resulted in harm to a respondent or another employee. STC tolerance must be extremely low. So it is instructive to ask what caused the relatively few cases identified.

    PS31. Several causes are at play. First, authorization procedures were unclear, as described earlier. Second, not all field staff—some at a very long arms length—understood that screening must be completed in advance of work starting. Unlike previous censuses it was no longer acceptable to have screening completed after employees start work. Third, CARS did not have a built-in control to prevent these cases from slipping through. Fourth, managers wanted results in a tight time frame and took a calculated risk and we have some evidence that the regional security committee eventually approved the case. Finally, regional security committee cases were to be documented and a register was to be kept; however specific details were left to the each committee. Sometimes precise details were not always forthcoming about specific cases.

  4. User ID obtained after a reliability status was granted
    PS32. A process with appropriate controls was in place so that Statistics Act employees who required access to a Statistics Canada network did not receive it until after the checks for their reliability status had been completed. In local census offices, there were extremely limited occasions when such employees required access. In CHL sites, employees required access to the census network. It was given after the assistant director for census in each region authorized it and the actual access was given through headquarters—local informatics representatives could not give access.

Personnel screening-related recommendations

PS33. Recommendation 1: In keeping with the 2011 Census timetable, review the procedures related to personnel screening and clearly identify who is delegated as responsible to grant a reliability status, under what circumstances, and how this will be accomplished.  Reflect the results in modified procedures and tools, done with the benefit of consultation, as elaborated below.

PS34. Tools: A ‘delegation of security authorities’ chart for the 2011 Census would be a useful tool to concisely communicate this sort of information. A second tool is an updated Personnel Screening, Consent and Authorization Form (presently TBS/SCT 330-23E (Rev 2006-02)). A third tool is the implementation of the Security Screening Certificate and Briefing Form (TBS/SCT 330-47(rev.2006/02). The latter two tools are required by Treasury Board directive.

PS35. Procedures: These should be more comprehensive (including regional security committees’ procedures and standards of documentation, making obvious why individual tasks are necessary and who is authorized to do them) and cover appropriate forms completion and include a specific role for site managers to inspect and monitor the way in which consent and authorization forms are completed, if the authorizing security official is not the site manager.

PS36. Consultation: In keeping with the 2011 Census timetable, 2011 Census and Data Access and Control Services Division officials should formally seek advice from Treasury Board security officials to determine if Statistics Canada is appropriately interpreting the draft personnel security screening standard (or its final version) as it will be applied in the context of the 2011 Census, document the results and make any changes as needed to the Statistics Canada Security Practices Manual.

PS37. Recommendation 2: Implement an additional control so that employees cannot be assigned work until their security screening is complete or an interim status is granted by the regional security committee. This control will detect any case that slips through earlier preventive control points.

PS38. Recommendation 3: Provide sufficient cabinets and keys and include instructions about appropriate key management.

CHL monitoring

CHL1. All three Census Help Line (CHL) sites were visited—Moncton in early June; Edmonton at the end of June and Toronto in July. Since full production only began in early May, we saw Moncton early on—there was less time to adapt to challenges that any census collection brings.

CHL2.  In response to privacy concerns, to provide respondents with internet access, and to harness efficiencies through technology, the 2006 Census became more centralized and automated than its predecessors. As a result, the 2006 Census was the first census where CHL operators contacted respondents to carry out the failed edit follow-up  In 2001 these activities were done by enumerators. As in past censuses, there was a census help-line. With a much-reduced field staff in 2006, the help-line service took on added importance with Canadians placing 1.8 million calls compared to .8 million calls in 2001.

  1. Operators are adequately trained
    CHL 3. A training program exists and details of the training program are set out in the CHL supervisor’s manual (Form 80B).  Additional training materials exist, for example a training guide (Form 57A), self-study materials (Form 57B), and a work-book (Form 57C).

    CHL 4. Training was taken and it was evaluated positively by operators. They offered suggestions for improvement in some cases —for example, more practical experience during training—and these, along with the CHL managers’ feedback on training documented in their debriefing reports—they offered many suggestions—are control mechanisms for continuous improvement available to guide changes for 2011.

    CHL 5. Two days of classroom training were planned for operators and this was not designed to get operators to a ‘production-ready’ level. Census collection is a short-term operation, so there was a correspondingly modest expectation that operators would get the job done adequately: they would not become seasoned professionals in a short space of time. Even so, one site manager observed that training needed to be better, since Statistics Canada’s reputation is very important to its success. Operators’ formal training was to be supplemented through on-the-job training. However, the extent to which this occurred was affected by much higher than expected call volumes around and after census day, May 16, 2006. There was too much work to allow supervisors adequate time to perform monitoring and provide feedback. This is discussed fully in the next section which takes on added importance because, while training was adequate, it is just so.

  2. Monitoring can be improved
    CHL 6. Supervisors have different ways to monitor operators’ work: reviewing work samples on a daily basis; using MIS reports; and listening or observing their calls. Results of formal listening or observing of calls were recorded on Form 87 and discussed with the operator. This type of monitoring is referred to as ‘Form 87 monitoring’. These tools are controls to make sure work is being accomplished efficiently and effectively at the supervisory level14.

    Daily monitoring
    CHL 7. Supervisors reported that they were responsible to carry out daily monitoring (91%) indicating that nearly all understood this responsibility. Those who did not agree that this was something they were responsible for were all located in Moncton, where the manager assessed that supervisory roles and responsibilities needed clarification, particularly given Moncton’s use of a Form 87 monitoring team: having this team did not remove supervisors of their daily-monitoring responsibilities.

    CHL 8.  Most supervisors (85%) in Edmonton reported that they were successful in reviewing their employee’s work on a daily basis (17/20). They reported this less often in Toronto and Moncton, 47% (7/15) and 38% (6/16) respectively. However, there are a significant number of supervisors answering neutrally, which may indicate a misunderstanding of the question. However, the option to answer “not applicable” was provided. To minimize this effect, we have looked at the proportion disagreeing with the statement ‘I check the completed work of each employee every day’. No one disagreed with this in Edmonton, while 5 or 31% did in Moncton and 3 or 20% did in Toronto. Daily monitoring was being done, with supervisors in Edmonton most successful in doing so. There appears to be room for some improvement elsewhere.

    CHL 9. In Edmonton, supervisors were made accountable for this activity—they were to hand in their daily duties checklist15 after every shift. We did not observe this type of daily accountability at the other two sites.

    Supervisors didn’t use MIS reports
    CHL 10. For the most part, supervisors do not use the MIS reports described in their supervisor’s Manual (Form 80B). The main exception is in Toronto, where FEFUMIS6 report was used and supervisors were drawing the information from it to create their own team-based reports. With this one exception, managers have not required that supervisors use MIS reports. They were “not very user friendly” in the words of one supervisor—a sentiment echoed by others.

    CHL 11. MIS reports did not meet the daily on-the-spot, ‘in real time’ needs of supervisors. There was too many and not enough time to learn about them. They did not readily provide the information from a supervisor’s perspective—for example, a supervisor wants to know how employees reporting to him or her are doing so information needs to be organized this way. While the MIS system has flexibility to tailor reports, the strategy of leaving this to supervisors to figure out meant that it largely did not happen.

    Form 87 monitoring16
    CHL 12. Form 87 is the monitoring tool that the supervisor or monitor completes to evaluate a call. It evaluates twenty dimensions covering four areas—delivery (6), subject-matter (3), professional behaviour (9) and data processing (2)—either on the basis of listening to a call or observing the operator on a call. Feedback is given, with the operator acknowledging this and undertaking to make any improvements identified. It represents an important control and the benchmark for its completion is weekly, and to avoid over-monitoring—no more than twice in a day.

    CHL 13. When Form 87 monitoring was conducted, it was well-done. Despite its length and need to give feedback with sign-off by the operator, all the steps were followed. Over-monitoring did not occur. Form 87 monitoring did not meet the benchmark of once-a-week frequency.

    CHL 14. Of the three CHL sites, Edmonton operators were monitored using Form 87 more frequently than in Moncton and Toronto. Edmonton results show that it is possible to reach a benchmark of conducting Forms 87 monitoring weekly at least for a sizable portion of the operators. The latter two sites indicated that they did not monitor with a weekly frequency in May 2006 and audit results confirm this. Additional data from Moncton shows that monitoring rates improved over time to approximately once every two weeks.

    CHL 15. The approach to Form 87 monitoring was different at each site. In Toronto, a team-based approach was adopted, with importance placed on the operator-supervisor relationship so that operators would know who they were working for and consistent performance management feedback could be easily delivered. In Edmonton and Moncton, operators had a supervisor assigned for the shift. In Moncton the advantages of a consistent operator-supervisor relationship were recognized, but scheduling considerations made it difficult to keep an operator with the same supervisor beyond a shift. In Edmonton, the supervisors did the Forms 87 monitoring, while in Moncton, a small team of supervisors became full-time Form-87 monitors. Therefore, each CHL site has lessons to share that will be valuable for the 2011 Census when it builds on the 2006 experience.

    Results by site
    CHL 16. These vary because of the differences described earlier and comparisons must be made cautiously.  

    Moncton
    CHL 17. About 10% of operators were not monitored at all by June 7. Most operators would have started by May 2, 2007, so they should have had about 4-5 monitoring events by June 7.  Using Moncton data, the average over this period was 1.4 times. These results improved over time, to approximately once every two weeks. Moncton developed its own tool to assess improvement towards the benchmark.

    Toronto
    CHL 18. Toronto monitored on average 1.5 to 1.6 times over a 5½-week period. This is about the same as in Moncton, where the rate was 1.4 times. The monitoring rate did not improve in June or July: in fact, it was best in May, when just over half of all the monitoring occurred.

    Edmonton
    CHL 19. Twenty-two percent of operators were not monitored sufficiently in the May-June period. This is a conservative number, since we did not access all the records, given how they were stored. Evidence of weekly monitoring exists to an extent sufficient to show that a weekly monitoring benchmark is attainable.

    CHL 20. Compared to Toronto and Moncton, Edmonton appears more successful in accomplishing monitoring. Edmonton was the only site that required supervisors to hand in a daily checklist, which included monitoring, at the end of a shift.

    CHL 21. Several factors explain why Form 87 monitoring did not meet the benchmark, in addition to high call volumes around and after census day described earlier. While organizational differences mean different emphases for each factor, most were shared by at least two sites.

    • CHL 22. Operators were not trained to a production-ready level and were learning on-the-job: supervisors were answering their questions, especially at the start of operations, not monitoring.
    • CHL 23. Supervisors did not always have solid supervisory skills. Recruiting was accomplished in tight time frames and finding staff was challenging. Edmonton made expectations clear to supervisory candidates and suggested that this is an important success factor. An after-the-fact assessment in Moncton supports this. Identifying candidates with an interest in CHL work rather than field (LCO) work through an improved STC web site would improve the hiring process for supervisors according to at least one site manager and census management may wish to consider this for 2011.
    • CHL 24. Supervisors’ performance expectations did not include a requirement to quantify how much Form 87 monitoring was being done, although in Edmonton supervisors were required to hand in their daily duties checklists which made clear that various monitoring levels were expected (not Form 87 monitoring, however, perhaps because this was “weekly”. However, this could be accommodated in a revised checklist).
    • CHL 25. Form 87’s design used a “pass-fail” approach which many supervisors disliked since it made it harder to deliver feedback. It diverted energy from the monitoring task because it was seen as unfair: it provoked debate and in the case of Moncton, necessitated the formation of a separate, full-time monitoring team. Operators that were considered as doing a very good job were sometimes being failed because they did not follow the script to the letter. Supervisors require time together to develop an implementation standard to avoid this sort of result. If supervisors are not experienced, they risk being too tough—failing an operator for saying “ya” not “yes”, for example, with an unproductive knock-on effect. Some dimensions of Form 87 were considered more important and automatically lead to a “fail” but some supervisors disagreed with this evaluation—rather, all dimensions are important—provoking debate. This was quite strong in some quarters, with one monitor declaring that completing this form was the worst part of the job. While this is an extreme view, it serves to illustrate the passion provoked and energy diverted because of it. Moncton failed more operators, suggesting that they applied standards more stringently, leading to more controversy.
    • CHL 26. Tools that would facilitate Form 87 monitoring were not available: supervisors could listen in on a call, but could not see how operators were using the operator reference system (ORS). It was difficult to locate operators (at the start of operations, even recognizing them is a challenge if a team-based approach is not used) and took time to switch over to access an operator for monitoring purposes. Phone systems used old technology and varied by site so there is no overall “one way to do it”. 
    • CHL 27. Organizational logistics do not facilitate monitoring. First one must identify who is to be monitored. If this is not the supervisor or the supervisor changes, then determining this is not obvious. Ideally, one wants to monitor a complete call.  The monitor needs to determine this, either visually (the operator must be within eyesight) or technically: find the phone to tap into and know that the call is just starting. Then the supervisor needs time to do the monitoring—and not be interrupted by others. Supervisors monitoring on the floor so they can see how the operator reference system is used are prone to interruption. Once the monitored call is over, feedback is delivered: ideally, immediately for most effect. This can be challenging since the operator might have started another call, gone on break, etc. If feedback is sensitive, the supervisor needs a private place to deliver it. Such spots were not readily available.
    •  CHL 28. The time required to monitor one individual once under the circumstances described above (without effective tools) is too long. Snags along the way will mean that an event is aborted until later. The steps above show real potential for this to happen.
    •  CHL 29. Monitors in Moncton did not have access to CARS, so were not able to easily retrieve start-work dates for employees to include in their spreadsheet that was directing their monitoring efforts. Start-work date is a key piece of information to maximize the probability that an operator is monitored early.

    CHL 30. When operators talk to respondents they are Statistics Canada’s public face and it is important that this public receives professional service. Further, operators contribute to census quality results by doing their job correctly. As such, Form 87 monitoring is a key control. While an absence of monitoring does not mean that operators are not doing their jobs effectively, it does mean that supervisors are not in a position to improve the way in which the work is done. Further, knowing that an operation is meeting expectations is a control itself.  Monitoring would identify inefficient calls handling (e.g., not using operator tools efficiently) insufficient politeness with respondents (for example, “tutoyer”), or significant deviations from script. The better operators are doing their jobs, the more effective the result will be.

  3. Effective controls
    CHL 31. In the section above, various monitoring types—and particularly Form 87 monitoring—have been discussed and the three sites described. This section focusses on the controls managers used to oversee what was happening at lower levels. The focus is on Form 87 monitoring control, and is considered in a larger context of managing operator performance generally.

    CHL 32. Managers had controls to assess monitoring results although there is some opportunity to improve them. Form 87 monitoring would be improved by better knowledge of its frequency so that the site manager can assess and those directly below adjust. Moncton developed a management control to assess Form 87 monitoring frequency and results, although even here, reporting to the site manager on results was not required.

    CHL 33. To assess performance management generally, managers used a variety of tools. They described the complementary nature of the component elements of good performance and created tools to manage performance overall. They emphasized the context in which monitoring took place: changing nature of day-to-day operations and the importance of flexibility to rapidly adapt. They identified ways to improve monitoring in future, both Form 87 monitoring, and more generally; changes that could be made even within the constraints imposed by the short-term nature of the work.

    Form 87 monitoring
    CHL 34. Managers identified Form 87 monitoring as important and knew in general terms when it was not getting done as frequently as it should. For example, May was identified as a problem month in both Toronto and Moncton, when operators were new and operations were in full swing. This was the time when monitoring would have been most useful to resolve issues early in operations but Form 87 monitoring took a back seat to more pressing operational concerns. Nevertheless, better information on the extent of this monitoring might have provided impetus to improve, as it did in Moncton, where management had a control on Form 87 monitoring. Having this control allowed the site manager to take steps to improve efficiency. This would have occurred much earlier if reporting on monitoring results had been required from the start.

    Other monitoring
    CHL 35. Managers at each site were creating or borrowing different tools. As noted previously, daily monitoring was emphasized in Edmonton through the use of completed daily duties checklists that supervisors handed in. The Interviewer Reporting System was used, borrowed from the Western Regional Office and it logged performance issues. In Moncton, operators submitted daily logs of their production—a management control introduced after operations started and production levels needed improving. In Toronto, supervisors submitted weekly summaries showing daily operator production.

    CHL 36 In general, managers reported using standard techniques such as: establishing benchmarks independently at sites; giving priority to problem cases; manager meetings and e-mails; one-on-one performance meetings to manage issues identified through monitoring; and competition through the posting of targets and results achieved to spur production. Managers noted that providing an integrated performance picture to the operator is preferred (production rates, quality of that production, attendance). This removes room for misinterpretation of performance messages, feedback on Form 87 being only one.

    CHL 37. Individual sites adapted tools, introduced new ones and recognized the importance of tools that integrate performance information. Focussing improvement for 2011 on adapting individual practices generally offers a sensible approach based on experience.

  4. Appropriate action taken
    CHL 38. Appropriate action was taken when results were unacceptable. One key mechanism to provide feedback to CHL operators was Form 87. When a supervisor completed one, it was signed by the operator, showing that operators were routinely given feedback. As noted earlier, some supervisors found it easier to give positive feedback and identified language of the form—pass/fail—as a stumbling block.  In addition, managers used other tools such as monitoring of other performance-related reports to help them take appropriate action. In Toronto, for example, strong emphasis was placed on using a team approach and building performance metrics that were easy to use and understand.  In Edmonton, supervisors documented performance matters in a database for ease of communication with other supervisors who would cover an employee on another shift and to track resolution of issues. Moncton set up a spreadsheet to help monitor the frequency and results (pass or fail) of Form 87 monitoring. It included space to document when a more senior manager became involved.

    CHL 39. Managers did identify the need for an additional tool—escalation procedures when performance was not acceptable. Detailing these in advance would smooth operations in 2011. Sharing tools and associated details about when and how they were used across CHL sites in 2011 offers opportunity to facilitate further the ease with which appropriate action is taken.

  5. Complaints handled

    CHL 40. There are management practices in place to handle public complaints concerning the quality of operators’ work. There are written procedures for difficult calls and for complaints found in the manuals for operators, supervisors and managers, along with the supervisory procedures located in these manuals complement one another and constitute reasonable controls to ensure that any complaints about operators will be effectively handled.  Managers were consistent in their message that complaints were not an issue. Management practices related to escalated calls are appropriate.

  6. Evaluations stored in cabinets with built-in locks
    CHL 41. Form 87 evaluations were stored in cabinets with built-in locks. This cabinetry does not meet Statistics Canada’s internal standards but it does meet governmental standards for the storage of protected B information. At the Toronto CHL site, supervisors were provided with their own filing cabinet and keys in which to store their team’s evaluations. In the other two CHL offices, records were stored centrally.

    CHL 42. In evaluating the Moncton site, the CHL manager noted that there were insufficient cabinets with locks, and that there were no guidelines or estimates on how many were needed. In Toronto, the manager noted that in future, arrangements should be made in advance to supply each supervisor with a 2-drawer filing cabinet.

    CHL 43. While sites have other security measures to minimize the possibility of those without a ‘need to know’ having access, for example perimeter security, observation, walled offices with doors that can be locked, some elements focus on outsiders, not other employees. Further, the extra risk incurred by not using cabinets with hasps and proper locks is that entry could be gained without notice—built-in locks being easy to pick. The other risk is that lowering standards to make built-in locks acceptable would have a negative knock-on effect within STC. A threat-risk assessment for 2011 Census field collection would be advisable to assess this.

    CHL 44. There would be a cost impact to locating cabinets with hasps that are second hand. (Hasps allow a lock that is not built-in to be used.) They would be more expensive to buy and perhaps more difficult to find, since they are better cabinets.

CHL-related recommendations

CHL 45. Recommendation 1: Improve efficiency of Form 87 monitoring through improved tools and information; and increase accountability of supervisors for monitoring

  • Improve the tools supervisors use so that monitoring events are accomplished more quickly. This includes improving the design of Form 87 (e.g., replacing “pass or fail” evaluation). Other potential areas for improvement are the phone system monitoring capacity, scheduling tools, and feedback methods that do not rely only on a monitor and operator meeting. Automating the form could be an alternative.
  • Site managers should not have to analyse which organizational alternative is best while in the field.  This should have been considered in advance with regard to the operating constraints and tools available, so that it is integrated as effectively as possible. Promoting a consistent approach across sites makes evaluation easier.
  • Pay attention to the extent to which the Form 87 control is working—managers should be able to answer this question quantitatively: to what extent is the monitoring benchmark being met?

CHL 46. Recommendation 2: Provide better management information tools related to operator performance

  • Provide integrated performance-related tools to supervisors and managers in which they can store and through which they can share performance-related information that they generate (e.g., Form 87 results, escalation information, attendance etc.) and which will summarize information. Build in the facility to record escalation information and establish appropriate escalation benchmarks.
  • Reduce the choice of production-related information available for supervisors and focus on only the most relevant, formatting the information in the most useful way: include very timely individual operator information.

CHL 47. Recommendation 3: Conduct a threat, risk assessment to determine if cabinets with built-in locks are sufficient

  • This may result in a unique Statistics Canada standard for use in the 2011 Census.
  • Give supervisors the facility to store personnel information under their own lock and key, respecting the principle of ‘need-to-know’.

Conclusion

Gen 22. Pay was processed in an accurate way, mainly in accordance with established legislation, policies and procedures. However, payments to employees were not always timely as streamlining and adjustment of some controls is necessary to show evidence of due diligence.

Gen 23. Personnel screening for new census Statistics Act employees met most of the requirements of the Government Security Policy (2002) and Personnel Security Standard (1994). Reliability checks were done before employees began work, although improvements to some tools and procedures are required for 2011 Census field operations. Employees swear the oath of office under the Statistics Act and have a reliability status before accessing a computing device connected to any Statistics Canada network.

Gen 24. Overall, management controls were in place related to monitoring of Census Help Line telephone operators either responding to the Canadian public or contacting them for information as part of failed edit follow-up, although their effectiveness should be improved.

Appendix A

Management Action Plan
Recommendations Action Plan or Explanation of No action on the Recommendations Responsible for Action Estimated Completion Date Status
On Pay
Pay 21. Recommendation 1: Reduce number and complexity of pay forms
Such measures are needed to reduce or avoid the long delays observed. Separating the pay and production functions on forms of their own would help. The pay process should be simplified and the potential for further automation should be explored.
Pay and production MIS were decoupled during the 2006 Census operations. The current planning assumption for the 2011 Census is to automate pay reporting and sign-off through the use of an on-line portal. We are currently investigating the possibility of moving field staff in all operations to an hourly rate of pay for 2011. Doing so would allow for further simplification of the pay process and system. Marc Hamel In time for the 2009 Census Test  
Pay 22. Recommendation 2: Streamline control related to supervisor signature
In cases where the distance between enumerator and supervisor is significant, signatures are difficult to obtain, causing some delay in the pay process. Electronic approvals of forms could be an option. On the other hand, this control (supervisor's signature on Form 33) could be dropped if the signature of the supervisor on the daily summary (Form 32) is judged to be sufficient.
Automated pay reporting with supervisor on-line review and approval would replace the need for a physical signature and create an electronic audit trail (as outlined in Pay 21). In the absence of an electronic option. Simplification as recommended can be made to the paper based process. Marc Hamel In time for the 2009 Census Test  
Pay 23. Recommendation 3: Improve follow-up of late claims
A control should be added to the pay process to prevent late pay forms from escalating. After a certain amount of time following a reference period, a follow up process should be activated. This process must be able to detect the problem, analyze it and find a solution. It is very important to document all the steps taken, as it will become evidence to show due diligence.
In 2006, daily MIS reports were produced in Head Office and sent to the regions. These flagged late pay cases for further follow-up. Most late pay claims from the 2006 Census were not the result of backlogs in the Local Census Office pay units, but more so due to late or erroneous pay claims from field staff. Based on the 2011 planning assumption it is hoped that an automated pay reporting tool (as outlined in Pay 21) will prompt field staff to submit claims in a more timely and accurate manner and allow for faster supervisory review and approval and reduce the requirement for pay reconciliation in the Local Census Office pay units. We will endeavour to include flags in the automated pay tool to be triggered when pay claims are late and to track the receipt date of the initial claim, if it was in error and returned for correction, date of second receipt, when it was processed in the pay unit and what management action is being taken on the file. Procedures will also be revised to further formalize the process to be followed for late pay claims. Marc Hamel In time for the 2009 Census Test  
On Personnel Screening

PS32. Recommendation 1: In keeping with the 2011 Census timetable, review the procedures related to personnel screening and clearly identify who is delegated as responsible to grant a reliability status, under what circumstances, and how this will be accomplished.  Reflect the results in modified procedures and tools, done with the benefit of consultation as elaborated below.

       

PS32a. Tools: A ‘delegation of security authorities’ chart for the 2011 Census would be a useful tool to concisely communicate this sort of information. A second tool is an updated Personnel Screening, Consent and Authorization Form (presently TBS/SCT 330-23E (Rev 2006-02)). A third tool is the implementation of the Security Screening Certificate and Briefing Form (TBS/SCT 330-47(rev.2006/02). The latter two tools are required by Treasury Board directive.

PS32a – A ‘delegation of security authorities' chart will be developed in conjunction with DACS. This will be integrated into the 2011 procedures and training materials. The 2011 Census Field Operations project will use the most current version of Treasury Board security screening forms within the confines of systems development and operational windows. The project will continue to work with DACS on this front. Marc Hamel and DACS In time for the 2009 Census Test and forms review prior to 2011 operations  

PS32b. Procedures: These should be more comprehensive (including regional security committees' procedures and standards of documentation, making obvious why individual tasks are necessary and who is authorized to do them) and cover appropriate forms completion and include a specific role for site managers to inspect and monitor the way in which consent and authorization forms are completed, if the authorizing security official is not the site manager.

PS32b – Procedures for the 2011 Census will be enhanced building on those which were in place for the 2006 Regional Security Committees. As in 2006, we will consult with DACS in the development of all security and personnel screening related procedures and practices. FOP will analyze the utility of new technologies such as ‘digital finger printing' in order to reduce the complexity and increase the timeliness of the screening process. Marc Hamel and DACS In time for the 2009 Census Test/2011 operations  

PS 32c: Consultation: In keeping with the 2011 Census timetable, 2011 Census and Data Access and Control Services Division officials should formally seek advice from Treasury Board security officials to determine if Statistics Canada is appropriately interpreting the draft personnel security screening standard (or its final version) as it will be applied in the context of the 2011 Census, document the results and make any changes as needed to the Statistics Canada Security Practices Manual.

PS32c –This recommendation will be adopted. Census FOP will work with DACS to seek advice from the Treasury Board. DACS and Marc Hamel In time for 2009 Census Test/2011 operations  

PS 33. Recommendation 2: Implement an additional control so that employees cannot be assigned work until their security screening is complete or an interim status is granted by the regional security committee. This control will detect any case that slips through earlier, preventive control points.

Current procedures will be reviewed prior to the 2009 Census test to ensure that the appropriate checks are implemented. Will investigate adding a flag to the Automated Recruitment System for cases where a final or interim security clearance has not yet been granted. Marc Hamel In time for the 2009 Census Test  

PS 34. Recommendation 3: Provide sufficient cabinets and keys and include instructions about appropriate key management.

This recommendation will be adopted for the 2011 Census Marc Hamel In time for the 2009 Census Test  
On CHL

CHL 42. Recommendation 1: Improve efficiency of Form 87 monitoring through improved tools and information; and increase accountability of supervisors for monitoring

  • Improve the tools supervisors use so that monitoring events are accomplished more quickly. This includes improving the design of Form 87 (e.g., replacing “pass or fail” evaluation). Other potential areas for improvement are the phone system monitoring capacity, scheduling tools, and feedback methods that do not rely only on a monitor and operator meeting. Automating the form could be an alternative.
  • Site managers should not have to analyse which organizational alternative is best while in the field. This should have been considered in advance with regards to the operating constraints and tools available, so that it is integrated as effectively as possible. Promoting a consistent approach across sites makes evaluation easier.
  • Pay attention to the extent to which the Form 87 control is working—managers should be able to answer this question quantitatively: to what extent is the monitoring benchmark being met?
The form 87 monitoring form will be reviewed and revamped prior to the 2011 Census. FOP will investigate the feasibility of an automated monitoring form. The current planning assumption calls for one Census Help Line site to be co-located with the Data Processing Center for the next Census. This should eliminate the issue of consistency across sites. FOP will recommend that a new supervisor/monitor position be added to the CHL organization structure to ensure autonomous and consistent monitoring feedback. FOP will also recommend that CHL operators are provided with productivity reports comparing their personal production to a benchmark or group average. FOP/DPC In time for the 2009 Census Test/2011 operations  

CHL 45. Recommendation 2: Provide better management information tools related to operator performance

  • Provide integrated performance-related tools to supervisors and managers in which they can store and through which they can share performance-related information that they generate (e.g., Form 87 results, escalation information, attendance etc.) and which will summarize information. Build in the facility to record escalation information and establish appropriate escalation benchmarks.
  • Reduce the choice of production-related information available for supervisors and focus on only the most relevant, formatting the information in the most useful way: include very timely individual operator information.

FOP will recommend that CHL operators are provided with productivity reports comparing their personal production to a benchmark or group average. Current planning assumption call for more streamlined MIS with the goal of creating key indicators for managers and supervisors.

An escalation process will be specified and recommended to CHL management

FOP/DPC In time for the 2009 Census Test/2011 operations  

CHL 35. Recommendation 3: Conduct a threat, risk assessment to determine if cabinets with built-in locks are sufficient

  • This may result in a unique Statistics Canada standard for use in the 2011 Census.
  • Give supervisors the facility to store personnel information under their own lock and key, respecting the principle of ‘need-to-know'.

Will endeavour to only use secure cabinets with hasps and approved locks.

Given that supervision and monitoring of CHL operators is often shared across supervisors, centralized, shared cabinets with enhanced key control will be recommended for 2011.

FOP/DPC In time for the 2009 Census Test/2011 operations  

Note

  1. Employees swear not to divulge confidential information.
  2. Statistics Canada Departmental Performance Report, period ending March 31, 2004. p.20
  3. Failed edit follow-up is a census process whereby questionnaires that have not passed quality control checks are identified and the respondents contacted to get the correct information.
  4. Employees swear not to divulge confidential information.
  5. Field operations covers those attached to Census Help Line sites, Regional Census Centres, Local Census Offices (LCO) and sub-LCO
  6. Completed Forms 87 were to be stored in personnel files. We expected to find individual file folders for each employee that held these forms. We did not expect to find them married up with ‘the’ personnel file held in HR, since the forms have operational use after completion, at least in two sites.
  7. This is the auditors' residual risk assessment, based on audit results. It is providing the reader with a high-level overview of the impact for Statistics Canada.
  8. One could rate this as a 'caution'. However, it is our considered judgement to rate this as ‘OK' although there were authorization issues. This recognizes the inter-dependency between this criterion and the one above and in our view, balances the message without inappropriately overstating it.
  9. Forms 30, 33 (a to g) and 34 are weekly pay forms filled out by the employee. The type of work being done dictates the form to use. Form 1C was introduced later in the process.  It was also a pay form filled by employees. Form 32 is the supervisor’s daily summary.  Pay forms served a dual purpose: pay and production to feed MIS reports. Reconciliation between the weekly pay forms and the daily summaries was done weekly.
  10. Statistics Canada. The Companion Guide to the Statistics Act, December 1999.
  11. Government standards for cabinets were met. However, STC standards for cabinets were not met. We did not find cabinets with hasps and proper locks used. Rather cabinets with built-in locks were routinely used, in contravention of the Security Practices Manual, section 2.10 b). A threat-risk assessment could be done to determine if an exception should be made for census employee personnel and pay records.
  12. The audit could not confirm independently from Form 26B results that the employment history check was done. It is a separate check indicated on Form 26B. According to census documentation (Form 45-7), this check is part of the reference check. This is not a concern since the TBS consent and authorization form (TBS/SCT 330-23E (Rev. 2006/02) groups these checks together too.  However, it would be helpful to have a fit between the census form and the procedures.
  13. As was noted on p. 11, the census consent and authorization form did not contain space for authorization, only certification that the checks had been done.  Therefore, certification information acted as an approximation.
  14. There are other controls designed for more senior management levels. They are not in scope.
  15. This checklist was designed locally, containing more items than the daily duties checklist in Form80B.
  16. Informal monitoring through observation happens on a daily basis, by virtue of questions asked, walking around and seeing what operators are doing—looking over their shoulders for a brief period—and so on.