HMIS Data: Getting Measurable Results Through Business Intelligence

HMIS Data: Getting Measurable Results Through Business Intelligence

“The goal is to turn data into information, and information into insight.”
Carly Fiorina, former Executive, President, and Chair of Hewlett-Packard Co.

Efforts to end homelessness have become increasingly sophisticated over the past few years, due in large part to improvements in HMIS utilization. As HMIS utilization increases, so does the success rate of efforts to end homelessness.

HMIS reporting techniques have shifted as a result of these improvements—simple ‘output’ reporting no longer holds the clout it once did. Instead, performance measurement reporting has become necessary to achieve outcomes, serving as the key measurement tool that CoCs can use to measure which efforts are working, and which efforts are not.

Therefore, CoCs must shift their focus from ‘output reporting’ to ‘data exploration’. For example, learning that one CoC housed 350 persons in the course of a year (output reporting) is no longer sufficient, nor helpful, information.

In contrast, the ability to drill down on this number to the granular level to explore the patterns that lie beneath is invaluable. For example, of those 350 persons, which cohort took the longest for housing placement? Or, which programs within the CoC were the most successful, and which services within these programs contributed to this success?

Now, this is valuable information. This is information that shows patterns within that HMIS data.

Data alone cannot end homelessness; however, understanding underlying data trends, and responding to these trends accordingly, can.

In this article, we’ll discuss how to use business intelligence (BI) to understand your HMIS data. We’ll look at:

  • Why business intelligence applies to your CoC
  • The importance of understanding your HMIS data
  • How to use business intelligence to understand your HMIS data
  • Example scenario
  • 3 important questions you need to ask
    • What are the characteristics of your data?
    • What do you want to see?
    • What are the characteristics of your audience?

By the end of this article, you will be equipped with the knowledge and tools you need to better utilize BI and HMIS data to inform your policies, procedures, and/or programs to end homelessness.

 

Why Business Intelligence Applies to Your CoC

Funding is the ultimate goal of reporting—your CoC needs sufficient funding for each project within the CoC. Without proper funding, a CoC cannot develop the programs and services they need to end homelessness in their community.

Simple as this may sound, it’s actually quite complicated—each project has unique sets of clients, services, and functions, so it’s difficult to appropriate the correct amount of funding to each project. Therefore, you need to know the unique performance details for each individual project, right? This is difficult, to say the least.

For example, one Transitional Housing program serving homeless youth might house 75 persons over the course of the year, while another Transitional Housing program serving victims of domestic violence houses 5 persons over the course of the same year. Even though both are Transitional Housing projects, they have drastically different performance benchmarks due to the differing characteristics of their client population.

To make things more complicated, even if you are armed with this information, how do you present it to the decision-makers who disperse the funding? How do you make them understand why one project needs more funding than another project, or why your CoC needs additional housing vouchers for a particular population, for example?

The answer to both of these questions is the concept of business intelligence.

Although ‘Business Intelligence’ may seem to be a misdirected term to use in homelessness services, it’s actually a precisely accurate label.

Business intelligence (BI) is defined as a data analysis process aimed at boosting business performance by helping corporate executives and other end users make more informed decisions.

If you replace ‘business’ with ‘project’ and replace ‘corporate executives’ with ‘CoC Administrator’, then it becomes clear that BI is fully applicable to homelessness services. (You can also replace ‘business’ with ‘CoC’, ‘agency’, or ‘service’ and arrive at the same conclusion.)

 

The Importance of Understanding Your HMIS Data

Business intelligence enables you to understand your HMIS data, allowing you to unveil the hidden data trends existing beneath its surface.

Oftentimes, stories are hidden within HMIS data, undetectable by even the most sophisticated reports. Many communities and CoCs have ‘hunches’ regarding the patterns they believe lie within their data, but these patterns remain elusive.

For example, a community may have a hunch, based on day-to-day client interaction, that a program’s recidivism rate is higher for adult clients struggling mental illness, while it is substantially lower for homeless youth. But their standard reporting measures do not substantiate this belief. However, once they use BI techniques, these hidden data ‘gems’ are exposed and the knowledge garnered can be implemented in efforts to achieve outcomes.

Using BI techniques to understand these stories helps to create the programs and services necessary to end homelessness. The rich data sets stored in the HMIS can enable you to learn about the clients you serve, and thereby develop more effective policies and programs to meet their needs.

 

How to Use Business Intelligence to Understand Your HMIS Data

As mentioned above, performance cannot be measured using simple output reporting measures. Although such simplicity would be optimal, the interplay of far too many variables render simple output reporting methods inadequate.

The time has come for CoCs to allow BI tools to shift their focus from output reporting to data exploration. Differences between the two are discussed below:

 

Output Reporting

Output reporting provides a simple piece of information on a single variable. An example of output reporting is: How many homeless females compared to males has the CoC housed in the past year? Output reports will only provide a raw number—245 females and 315 males were housed this year, for example.

 

Data Exploration

Using BI techniques, data exploration can detect the interplay and correlations among different variables that make up a raw number. So, using the same example as above—data exploration will help to determine why more males were housed than females. Perhaps a significant portion of the female population were victims of domestic violence, making finding safe, stable housing more difficult? Or, another scenario could be that substance abuse was more prevalent in the male population, and the CoC happens to have particularly successful housing programs for persons struggling with substance abuse.

 

The possibilities are endless. The point is, business intelligence techniques can help to reveal the root causes of success and failure in relation to desired outcomes. Whether the success is associated with a particular project/service, BI is the secret to detecting not only which project/service is the most successful, but, most importantly, why this project/service is so successful?

The ‘why’ is the key to acquiring the appropriate amount of funding needed to replicate the successful attributes of these successful programs/services throughout the CoC. Thus, BI is a critical cog in the process of ending homelessness.

 

Example Scenario: The Story Behind Average Length of Program Stay

An example scenario is provided below. In this example, the CoC wants to determine what factors affect average length of program stay.

In this example, the initial report provided the average amount of days a client remained in a program before being housed—117 days.

Example of getting measurable results by using business intelligence with HMIS data.

But what variable(s) affect this average length of program stay?

Is it possible that certain cohorts remain in programs longer than other cohorts before they are housed? Does the designation of chronic homelessness make a difference?

After using BI techniques to drill down on chronic homelessness, it becomes apparent that persons who are chronically homeless have a higher than average program stay compared to those who are not chronically homeless. More specifically, programs take an average of 141 days to house a chronically homeless person (blue) as opposed to an average of only 117 days for non-chronically homeless persons (green).

Second example of getting measurable results by using business intelligence with HMIS data.

Now that we know that programs are having more difficulty housing chronically homeless persons than non-chronically homeless persons, it would be helpful to know what types of characteristics might be contributing to this difficulty.

This is accomplished by putting additional variables into the mix. One such variable could be mental illness. This would answer the question of: “Are mentally ill chronically homeless persons likely to spend more days in a program before being housed than chronically homeless persons who do not report mental illness?”

The example data below indicates that this is a true statement. Programs take an average of 54 more days to house chronically homeless individuals who report mental illness (green), than chronically homeless persons who do not report mental illness (blue).

(NOTE: Any data for non-chronically homeless persons have been removed from the chart.)

Third example of getting measurable results by using business intelligence with HMIS data.

Now, it’s time to understand which types of programs are most successful at housing mentally ill chronically homeless persons. In the graph below, it’s apparent that Transitional Housing programs (turquoise) have a greater average program stay than Rapid Re-Housing programs for chronically homeless persons reporting mental illness. (NOTE: Programs that are not applicable to housing have been removed.)

rrh vs th

There are many directions to go from here, depending on the question you are trying to answer.

Examples include:

  • How does drug/alcohol abuse affect the total?
  • Does Veteran status play a role?
  • What types of demographic patterns are present?

Again, the possibilities are endless.

The scenario above is a simplified example of how business intelligence techniques can help you understand the trends hiding within your data. If the data analysis had stopped at output reporting, then the only piece of information available is that it takes an average of 117 days to house a client. However, now we have additional information on cohort (mentally ill chronically homeless) and also information on which type of program is most successful at housing this particular cohort (Rapid Re-housing).

There are additional advantages accompany BI techniques. One such advantage includes the ability to monitor data quality and detect any potential problems with data collection.

For example, if you are interacting with your data to determine the patterns present in clients with disabilities, you may discover, based on the graphs, etc. that some of your agencies are not collecting information for substance abuse. You can then address this issue at its root cause.

 

3 Important Questions You Need to Ask

Answering these questions, and applying the proper business intelligence techniques will enable you to present your data in interesting ways and in formats that your audience can easily understand. It will also help you to detect those hidden stories that lie within the HMIS, stories that oftentimes have a substantial impact on the amount of funding a community receives, or does not receive, for that matter.

 

1. What Are the Characteristics of Your Data?

What is the size and cardinality of your data? Cardinality is the uniqueness of the data values contained in a data set. High cardinality indicates a large amount of unique values. Low cardinality means a set of data contains a large percentage of repeat values (i.e. gender or yes/no veteran status). Here are several examples: Client Record Unique Identification Numbers—this would be high cardinality as each item is unique. Client Gender or Veteran Status—this would be low cardinality as there will be a high amount of repeat values.

 

2. What Do You Want to See?

Determine what you are trying to visualize with your data. Are you trying to view the demographics characteristics of your data? Are you trying to understand causation or causality? Or perhaps you are trying to determine change over time. Understanding what you want to know will help you to know the steps to take to get there (e.g. what chart(s) to use).

 

3. What Are the Characteristics of Your Audience?

Who will be digesting your data? Will it be CoC board members? A collection of stakeholders? Or perhaps you will be presenting data at a community meeting? It’s also important to ask: What information does the reader need to be successful? And similarly, how much detail does the reader need? Knowing your audience, and understanding how they typically process visual information, will enable you to choose the visual representations of your data that best convey the information in the simplest form for your audience.

 

In Sum…

The HMIS is the critical conduit that funnels and processes the tireless efforts of homeless assistance providers, churning out the reports that accurately inform policy-makers at the local, state, and Federal level. In short, HMIS data is a primary way to develop policies, procedures, and programs based on measurable results.

However, HMIS data, regardless of its level of quality, is not useful unless it is understood.

As mentioned, each piece of HMIS data is telling a story—it’s explaining how and why certain programs/services are more effective than others for certain populations. The data shows the characteristics and antecedents of their homelessness, unveiling patterns that are critical to ending homelessness in your community.

Therefore, BI techniques enable you to explore your HMIS data to find patterns and correlations. It also then allows you to present your HMIS data in clear, digestible formats so that policy-makers can fully comprehend your CoC’s needs and allot funding accordingly.

In sum, the benefits of BI are not exclusive to large corporations, they’re highly applicable—if not critical—to local and nationwide efforts to homelessness. CoCs nationwide would benefit from applying BI to their CoC operations. More importantly, business intelligence can play a significant role in the efforts to end homelessness.