How to Tackle Common Data Issues and Improve Data Quality

Jacqueline Janosko

What is your program’s data quality score from the last month or year? How about your agency’s data quality score? What about the same project types in your Continuum of Care (CoC)? How are your CoC’s overall data scores? Can you answer these questions or do you know who can? Why are we even reviewing these numbers in the first place?

Like it or not, these questions all speak to the reality of ending homelessness in a digital age. The Department of Housing and Urban Development (HUD) has not kept the importance of data quality a secret and the agency is moving toward a more data driven culture than ever. HUD has upped the ante again with the System Performance Measures. These measures help determine which communities are successfully ending homelessness and which ones are still struggling. CoCs have to report these outcomes to HUD as part of the NOFA. As a result, POOR DATA QUALITY AND PERFORMANCE HAVE REAL CONSEQUENCES (reading between the lines, this means funding related consequences)!

Before we all panic, let’s remember that the state of CT is actually performing very well when it comes to data quality. Now we need to push ourselves to do even better. To paraphrase a famous line from the original Ghostbusters, “We have the tools and we have the talent.” So how do we get even better? Glad you asked!

Our entire response to the homelessness crisis MUST be data driven. We need to set specific goals for our communities and measure our progress against them. Included in these goals should be:

  1. Assessing our existing crisis response system to determine what it’s actually accomplishing

  2. Review our homeless specific programs to figure out how many entries are coming from another homeless program

  3. Program lengths of stay

  4. Exits to Permanent Housing

  5. Returns to Homelessness

  6. Length of time a person was homeless before their crisis was resolved, whether through self-resolution or through a housing program in the community

Data quality monitoring is always a local process first and foremost. Agencies need to understand and respect the human process to get at quality data. Local user meetings around data quality and improvements will help encourage staff to input better data, as will reviewing this information on a CoC level. System administrators are also part of the process, and in CT CCEH works closely with Nutmeg Consulting to ensure CoC Leads receive data reports that highlight areas needing improvement. Additionally, agencies and system administrators need to keep their lines of communication open and friendly as we all rely on each other to report accurate data.

Commitment to excellent data quality can also be frightening to some agencies or CoCs. In actively monitoring the data and the outcome story it tells, it’s not uncommon to discover issues that you may not want to see. Also, when going beyond monitoring just the data completeness scores, the outcome analysis for projects may show agencies that have stellar data quality scores, but abysmal program outcomes. However, the purpose of having comprehensive data quality monitoring is to help agencies improve so that the entire CoC’s score stays high and the risk of funding cuts remains low!

What are some tips to improving our data? Again, I’m glad you asked!

  1. Real-time data entry: every best practice indicates that your best data comes from entering the data while you are working with your client. Not only will you be able to ask clarifying questions, but you will also be increasing agency capacity by approximately 20% due to the reduction in data entry done at a later time.

  2. Everything begins and ends with being able to de-duplicate data. Using a consistent method of asking a person for the information needed will lead to better results. For example:
    a. Name: “How do you legally spell your name?”
    b. Date of Birth: “What is your date of birth?”
    c. Gender: “What do you identify as your gender?” (This should always be asked, regardless of what the staff person thinks a person gender is.)

  3. Housing & Homelessness History: Determine the best approach for getting someone to provide as accurate a picture as possible. Use phrases that connect to key times that stand out to someone like their birthday or holidays.

  4. Exit Date & Destinations (this field is now critical to the System Performance Measures): Do your best to get an answer from your client, but don’t be afraid to ask others who socialized with the person if they knew where your client was going? In CT we ask a preemptive question at intake to shelter to determine where a person might go if they didn’t show back up for their bed. This field can help you get better exit destination data. Also, be sure to exit clients from programs when they leave. Do not let an enrollment stay open unnecessarily as it will affect the length of stay performance measure!

  5. When discussing data locally, require agencies to bring their own monthly data reports. Agencies need to have a feeling of data ownership and nothing removes that connection faster than having someone else run those reports.

  6. Publically report the data quality and performance outcomes for each project. Data improves dramatically when people see their results online!

  7. Use the tools you have: The APR and Shelter Utilization Report provide valuable information and the client level detail reports accessible to anyone who runs the reports will help improve your ability to fix issues and monitor quality and accuracy.

We must always remember that data is useful only to the degree it is being used!