.png)
Data Quality and Corrections FAQ
Note: Many answers here are guided by the FY 24 HMIS Data Standards from HUD. This is currently not available online and is in the process of being updated.
General Guidance on data collection from FY 24 Data Standards
“To fully understand the personal stories behind that data and ultimately improve system effectiveness and experiences for those navigating the system, qualitative data must also be collected. Communities are encouraged to work with people with lived homeless experience to determine what qualitative data would improve how we serve and expand the public understanding of homelessness. Data collection, data-informed processes, and the homelessness system design planning must always be approached with empathy and with consideration for how you would want to be treated if you were experiencing homelessness.”
“If a project is experiencing a high rate of client refusals as compared to similar projects, CoCs should consider implementing training around interviewing or trust-building techniques to support client engagement. A deeper engagement with clients may lead to more rapid movement off the street and placement in housing, consistent with meeting federal goals to end homelessness and improvement on HUD's System Performance Measures.”
“Shelters [and Street Outreach programs] may have high rates of missing Destination data. Often, in this type of shelter, a client is exited after a period of not coming into the shelter, at which point the opportunity to ask clients where they are going is lost. HUD and other federal partners strongly encourage shelters, even large-scale shelters, to consider themselves to be a part of the community's system working to end homelessness. Any steps these projects can take to establish relationships with clients, focus on moving clients into more permanent housing situations, or collaborate with service projects that do so, will improve a system's functioning, data quality, and client outcomes.”
When to Make Corrections
Q: If there is a data inconsistency, is there a hierarchy in which we should consider one data element more correct than another?
A: In cases of inconsistencies within a single project enrollment, the best method to resolve is through interview/communication with the client. Where this is not possible, additional fields within the enrollment can be used to make corrections in the following instances
| Disabling Condition | If “yes” is selected for any health condition listed and it is expected to be long term, then “Disabling Condition” should be corrected to “yes”. If “Disabling Condition” is “yes” and all following fields are “no”, please review with the client or other internal records. |
| Project Start Dates | This element is important for determining length of time homeless and utilization for housing and shelter programs, so it should only be corrected if an agency has other records believed to be more accurate or in cases where an agency is asked by the HMIS team to correct (ie when the client’s Project Start date is earlier than the project itself began). Bed nights or services can be used in some cases for correcting start dates. Household members on a single enrollment can have different start dates in cases where they joined the hh after enrollment. |
| Project Exit Dates | As above, bed nights or other services can be used in some cases to correct Project Exit. Exits should be on the day of the last recorded bed night, continuous project stay, or service provided depending on the project type. |
| Destination (Data Element 3.12) | Can be corrected if there are other internal agency records considered to be more accurate or if staff receive corrected information. Prior living situation data on a future enrollment should NOT be used to correct a client’s Destination. See note on Destination in General Guidance section above. |
| Prior Living Situation (Data Element 3.917) | This is used to determine if a client has experienced chronic homelessness, but data quality is expected to vary by project type. Responses should reflect the location of the client immediately before Project Start. For Services Only or other non-residential projects, the Prior Living Situation may also be the Current Living Situation. Corrections can be made based on additional enrollment information. In rare circumstances, this may be corrected based on information from other agencies. |
| Income and Sources | If “yes” is selected for any income source listed on that enrollment, then “Income from any source” should be corrected to “yes”. Other or future dated enrollments should not be used to correct these fields. |
| Non-cash benefits | If “yes” is selected for any benefit source, the first question should be corrected to “yes”. Benefits data should be recorded only for benefits that are current as of the 'Information Date' (i.e., have not been terminated). The field is only intended to identify regular, recurrent benefits. Services and/or gifts such as phone cards and vouchers that are provided by a project during enrollment are not considered benefits. |
Q: If a client is simultaneously enrolled in Coordinated Entry and an HMIS Project and the client exits the HMIS Project, should they be exited from CE?
A: Yes, if the client has exited to a permanent housing destination outside of a placement from the Community Queue (ie Rentals with or without subsidy, staying with family/friends on a permanent basis, etc.) the case manager for the HMIS Project is responsible for also exiting the client from Coordinated Entry.
Q: If a client who is missing an enrollment data element and in a future enrollment within at the same agency, should we backfill the preceding enrollment with the complete enrollment's information? If so, is there a date in the future that this would no longer be appropriate? Does project type change the process?
A: Enrollment data in other projects and in other agencies should almost never be used to backfill data or make corrections, especially where the date of collection of the information exceeds 36 hours. Data and client circumstances are expected to change over time. Capturing that change is part of the function of HMIS. Corrections based on data collected by other agencies will typically only be considered when there are overlapping enrollments in two residential projects.
Q: Should my agency use data entered by another agency to correct an enrollment?
A: No, data entered by other agencies should almost never be used to correct enrollment information. Other enrollment or agency information should only be used to correct client demographic info where duplicate records are identified.
Q: Should Coordinated Entry data be used to correct an enrollment error in my agency’s HMIS project(s)?
A: Unless directed to by HMIS staff, CE project information should not be used to make HMIS project corrections. Exceptions may be made when data is collected within 36 hours of each other.
Q: Is there a timeframe after engagement where it would no longer be appropriate to ask a staff person to recall client enrollment data? For example, if a client has received no services at my agency in the last 30 days, with what confidence level can we ask a staff person to recall that client's intake information?
A: This will be dependent on the individual, however staff should not be pressured or coerced to provide data they don’t have clear memory of. Data quality improvement is not intended to fabricate data, only to clean and correct where possible.
HMIS Project Types
Q: How should different project types be handled differently, if at all?
A: Data element definitions and how calculations are made vary across project types, so the corrections needed will also differ. The data quality reports built into HMIS and sent by HMIS staff account for these differences.
Approximate Date Homelessness Started
Q: How should we correct inconsistencies in homelessness start date? For example, if the approximate start date of homelessness is more than three years before the project start date, and the client stated they had multiple episodes of homelessness in the last three years which element should be considered more accurate?
A: This is part of Data Element 3.917 (Prior Living Situation) and can be corrected based on other HMIS information deemed accurate. In our system, the data for approximate date homelessness started and the number of times homeless both currently cascade for new enrollments- they pull in the value from the enrollment immediately preceding the new enrollment- as long as it occurred within 180 days or less.
Corrections on active enrollments or on clients with a currently active enrollment at a partner agency, can be corrected based on conversation with the client or case manager.
To correct errors on closed enrollments and where there are no opportunities to check with a case manager, users may check prior enrollments with caution within 180 days for the following :
1. to see if there were any enrollments in residential programs (TH, RRH, PH, PSH).
or
2. if there are prior exits to permanent destinations. If neither of the data points above can be identified, the Approximate date can be assumed to be more accurate. The “Number of Times on the streets” question can be corrected.
Other notes:
👉During data collection, should be reviewed with the client and updated with each new enrollment. It should be made clear to the client that they are being asked for “this episode” of homelessness which includes moving between unsheltered and shelter locations.
👉“Episodes” are broken up if a client was in an institution for more than 90 days (3 months) or if they were in Transitional Housing, temporary, or permanent living situations for 7 or more days.