Ideally, a good data governance system will be in place before LDS development begins. However, if your organization has broken LDS ground without establishing an effective governance structure, it is never too late to start. And, even if your agency has implemented a governance structure, data governance is an iterative process that can always be improved.
Gauge your governance
How governed is your organization? The Data Administration Newsletter offers a data governance test to help you evaluate your agency in eight areas of data governance. The test is available at http://www.tdan.com/view-articles/10149 and is geared towards building the case for data governance to leaders of your organization.
Although no single approach is best for implementing a data governance initiative, some key action steps
should always be taken on the road to good data governance. Normally, chief information officers (CIO)
initiate the data governance plan. However they, or other leaders, come to realize their organization would
benefit from better data governance, they should begin with a careful and honest appraisal of the agency’s
approach to data management, weighing the enterprise’s current levels of coordination. For tools to help
in this period of reflection, see "Gauge your governance" above.
After this stage of self-assessment come some key steps. Firstly, high-level executive support should be sought, followed by the assignment of key responsibilities to staff members and the formation of several groups central to managing and implementing the initiative. Data governance structures range from basic to elaborate in terms of personnel who serve the various functions (see chapter 3). However, the key action steps in a minimal, core data governance structure are:
In this chapter, the terms "data steward" and "data owner" are used to describe staff roles within a data governance structure, and do not imply technical or legal ownership. However, these terms are sometimes used elsewhere with meanings different from those applied here. For instance, by "owner," real ownership of the data is not implied here in the way the Family Educational Rights and Privacy Act (FERPA) outlines. That is, at an organizational level, a state education agency (SEA) may be considered a data steward when managing data on behalf of a district, in this case the data owner. Or, similarly, a district might be referred to as a "data steward" for managing data from its schools, which are the "data owners" at this more granular level.
Appeal to agency leadership to gain buy-in for the data governance initiative. The specific leaders who should be on board will vary depending on the state’s size and staff. In small states, support might come from the education chief; while in bigger states the appropriate executive might be on the chief’s staff, perhaps an assistant commissioner or deputy. Communicate the costs of the status quo and the benefits that will result from greater coordination. Executives must understand the value of the agency’s data and the need to ensure effective management of those information assets, and to improve quality and security. To gain support, present qualitative and, ideally, quantitative costs of poor data, redundant efforts, and insufficient security. Stress the tangible benefits of a more strategic, enterprise-wide approach to data management that coordinates policies, processes, and architecture to improve data quality; aligns work across the agency and streamlines operations; more effectively protects the data; and shares information in a more systematic and timely manner (NASCIO 2008). In making your case for data governance, try the following compelling arguments.
Stress real or potential examples of loss of funding due to late or inaccurate reporting to the federal government. Explain how data governance can streamline processes and increase data quality; as well as ensure that the state and districts get the funding they require to meet students’ needs. For example, if the special education data were late last year, explain how this might have been avoided with clearer standards and requirements, better communication and collaboration with the districts, improved validation procedures, and the sharing of best practices among data stewards.
Explain that increased data accuracy and transparency possible with individual-level collections and data governance enable more appropriate allocation of resources. With an individual-level data collection, one can see where the numbers come from rather than just rely on a district tally. For instance, while last year’s submission might have included 50 English language learners, this year’s student collection may only include 15 individual students. While the aggregate number may be difficult or impossible to verify, the individual records show exactly where the numbers come from.
By improving communication between the state and districts, and facilitating collaboration around data issues, data governance reduces the time and money spent fixing bad data (through multiple resubmissions by districts, for example) or auditing districts with problematic data. Better quality data from the start will save time and resources later on.
Beyond the effects on the education system, education data may affect sectors of the state’s economy. For instance, a mistakenly high number of dropouts may decrease a school’s ratings and negatively impact real estate values in the area.
Through better coordination with districts, states are able to more easily identify districts experiencing similar difficulties. For example, if different districts use different vendors, the state might determine that a particular vendor needs to make improvements in a certain area. Or, districts might be targeted for professional development or the introduction of new validation procedures.
Finally, invoke unpleasant memories, such as phone calls from irate constituents and school staff about bad data.
Then explain how these problems could have been avoided had a data governance process been in place.
The importance of executive support for a data governance initiative cannot be overemphasized. Strong and continued commitment from organization leaders will not only provide needed resources but, more importantly, will support the culture change needed in a data governance effort, applying pressure from the top and providing the authority needed to enforce contentious decisions. For instance, a data governance group may make a decision that prompts backlash by a key staff member (perhaps responsibility for a data element is shifted from one staff member to another; or a program area’s data are included in the LDS without the department’s full support). Without pressure from leadership, staff resistance could undermine the process. Sometimes, just getting staff to show up for data governance meetings can be challenging if leaders do not stress the importance of participation or even make attendance mandatory. If possible, ask leaders to require data governance group members to send qualified substitutes in their stead when they are legitimately unavailable.
The good news is, your agency’s leadership is unlikely to willfully resist a data governance plan. Commonly, executives just need to be shown how data governance will benefit the organization. At the very least, staff need to inform leaders of the initiative; ideally, executives will support and participate in the entire process. If leadership support cannot be quickly won, however, agencies should start or continue their data governance initiatives anyway, while continuing to seek high-level buy-in. Inevitably, this support will prove crucial.
LDS Lore: The meeting mandate
Yori, the data governance coordinator, paced back and forth across the floor of the nearly empty board room. He didn’t know why people weren’t coming to the agency’s second Data Governance Committee meeting. The first one seemed a great success at the time, but here he was with only two thirds of the group, waiting for other members to arrive. He knew that the data governance plan would derail if participation was low. Nevertheless, Yori put on his game face, took roll, and got down to business with those in attendance. After the meeting, he strode down to the education chief’s office.
After a compelling appeal, Yori closed the chief’s door with a sense of accomplishment. He knew he’d sold her on data governance. The case was not even hard to make. Data governance, from his perspective, was a no-brainer: It was a common sense business solution for improving the department’s data. He had laid the case out clearly and simply, contrasting the department’s data management status quo with the potential benefits of the data governance initiative. Making the case even easier was the fact that, other than maybe some coffee and donuts for meetings, the costs to the state would be nearly zero.
That afternoon, the chief sent out a brief email:
"Participation in all data governance meetings is required. Staff assigned to these groups should attend all scheduled meetings, making them a priority over all other activities. If you are unavailable, please send a qualified substitute."
Sure enough, the next meeting of the Data Governance Committee packed the board room. Yori smiled as he called the meeting to order, confident that the plan was back on track.
The CIO should convene a group of executive management staff that includes the education chief (or other high-level agency staff member), the data governance coordinator, and executive leaders from each program area with a data steward. Rather than creating a new group, identifying an existing group that includes these members and asking them to occasionally focus on data governance may be preferable. This group’s main roles will be to establish the data governance policy; and to address issues that require executive support, such as those that affect multiple program areas and/or affect major agency reports or deliverables. (For a more detailed description of this group, see chapter 3.)
Assigning responsibility for overseeing the data governance initiative to one individual is absolutely critical. This staff member, the data governance coordinator, should be the "catalyst" for coordinating the data governance initiative, setting a cohesive action plan, and tirelessly pushing the process forward. Significant culture change must occur for data governance to take hold and make a difference in an agency. In fact, many initiatives fail because no single individual is in charge of making sure that the roles, responsibilities, and processes of the initiative are being followed on a daily basis. In the search for a coordinator, leaders may look for the following:
Identify areas of data and assign a data steward to each one. Articulate specific responsibilities, making each and every data element the responsibility of a single steward. For example, each data steward should be responsible for data elements, rather than for databases. Data stewards should “own” specific contents of the system regardless of where the data physically reside (e.g., on a desktop, in a database, or in a central LDS). A clear process should be created for identifying data stewards. For instance, leaders may look for staff who
Form a cross-area group of data management staff that includes the data governance coordinator, data stewards, and other key staff members. This group will be the core of the data governance process, and will effect most of the collaboration and decisionmaking. The data governance coordinator should chair the group and oversee the agenda. Early in its formation, the group should collectively agree upon a mission statement, which should include the identification, documentation, prioritization, and resolution of critical data issues; and to core goals and objectives (see appendix D). The committee should then meet frequently, perhaps monthly, to fulfill its stated mission. The coordinator and the data stewards should contribute agenda items for these meetings. Agenda items may include
Every data governance detail should be documented. For example, who’s responsible for what? When are deliverables due? What are the critical data issues and what’s the status of their resolutions? What are the standard procedures? Documentation helps keep the work on track, prevents confusion, and allows staff to replicate activities from year to year and in spite of turnover.
Master data management
Master data management (MDM) refers to the ongoing process of identifying the authoritative source of data
and ensuring that this source is consistently used to feed other data systems, or to populate the agency’s
central data store; as well as for reporting, dissemination, and analysis. In this way, it is the answer to
the “collect once, use many times” challenge. When key data elements are collected and used by multiple data
systems, MDM is the process that determines which single source is authoritative. When integrating data from
multiple sources into a central data warehouse, for example, authoritative (i.e., “master”) sources are
identified for each element. And when new elements are collected, an authoritative source is assigned for
each. When populating the data store with historical data collected before the MDM process began, the
authoritative sources need to be determined for those older data items as well. MDM also keeps track of the
data collected and maintained throughout the agency to ensure that common standards (data element names,
definitions, codes, formats, etc.) are being used. When all of the agency’s past and present data are
addressed, MDM’s will then focus on handling new data elements.
LDS Lore: Integration-inspired indigestion
Adam dug his fingertips into the armrest. He, his supervisor, and other program area staff were in
a meeting with the data governance coordinator and the CIO to discuss the agency’s plan to phase its
data into the LDS. Yori, the coordinator, was going over the LDS project’s goals and timeline, and had
just told Adam and his colleagues that their program area’s data would be part of the first phase of
data migration into the LDS. Adam did not like this idea one bit. “That’s my data,” he thought. He felt
he was losing control of a data set he’d managed for years and he didn’t want them dumped into some communal
data store. As the meeting continued, he worried the data would be at risk in the LDS and he wasn’t
swayed by the argument that the move was necessary to give users greater access to the records.
“They already get enough access to the data,” he thought. Would it be harder for him and his staff to
work with the data? He was accustomed to a good deal of autonomy, and he didn’t want to have to coordinate
with other areas of the department. Suddenly, Adam’s stomach tensed. Was this the first step in phasing out,
or scaling back, the staff? Would his job be at risk when the integration was complete? Then, twisting his
feet around the chair’s legs, Adam realized that integrating “his” data into the LDS would probably expose
the records to scrutiny. Errors might be discovered. He and his staff would be held accountable for
poor quality data. This was bad…
As discussed earlier, a key principle of data governance is collaboration across the program areas of the organization. Many data issues affect multiple program areas but, without data governance, these issues are unlikely to be addressed in a comprehensive way. According to Chatis Consulting, when an issue arises, the Data Governance Committee should form a working group of stewards to collaboratively address the problems and craft a solution. Within the group, one data steward should be ultimately responsible for ensuring the working group resolves the issue. Groups should identify the problem and pinpoint its original source; define the goals of solving the issue; set up a clear and detailed strategy for a resolution; report back to the Data Governance Committee on progress; work with IT to implement the business solution; and, finally, communicate the resolution to all relevant stakeholders. (See appendix D for guidelines on how group meetings can be structured.)
Track data problems to their source
The identification of a data problem’s source is essential if the issue is to be resolved once and for all. If staff only treat the symptoms, the problem will likely surface again in the future.