Eviction Prediction Study


A little over a year ago I embarked on a large project with the help of some interns to accumulate eviction data from our applicants  in hopes of getting a better predictor of when someone will be evicted or how likely it is for them to get evicted.  Even though VeriScreen has been doing affordable housing screening for over 20 years and feel like we have intimate knowledge of our applicants and this particular market segment, I wanted to find out more.  I see our role in providing background checks for affordable housing as not only important in keeping our customers’ delinquencies and evictions down but I also see it as a moral issue.  The way I look at it, every time we let an unqualified applicant into one of our client’s communities, we are probably putting a very qualified single mom with children on a waiting list.  The question is how long can she go without getting housing assistance?  Isn’t that the whole idea behind the program?

The single mom with children can go to another community that does not do background checks the way we do them and possibly get in.  The problem is that I feel another purpose of doing such a comprehensive criminal background check is to provide a safe environment for all residents and keep criminals away from influential kids or teenagers.  It defeats the purpose to move innocent, impressionable children into a bad environment that are more than likely going to become criminals themselves or get stuck in the system with bad influences throughout their community.

Now that we have established our desire to find all criminal records, how did that come into play when trying to build an eviction prediction model.  The study gathered several pieces of information to help us determine or predict the likelihood of an eviction as follows:


  • What is the likelihood that someone will be evicted once they have a prior eviction?
  • What time frame will someone be evicted again after being evicted one or more times?
  • What correlation does criminal activity have on eventually being evicted?
  • Breaking down the above hypothesis even further, what is the correlation between felonies and misdemeanors looked at separately on the likelihood of being evicted?

The above questions were what I wanted answers to.  Let’s face it.  If a property manager could predict evictions costing an average of $4,000 each wouldn’t they want that information and just avoid evictions almost entirely and at the same time provide a more crime free community for people to live in?  I assume there will always be circumstances whereby someone does not have a high eviction prediction indicator and for many reasons end up getting evicted anyway.  We all go through life with major happenings that makes us do something different than we normally would.  Therefore, I don’t think our comprehensive study and predictor would ever totally eliminate evictions.

The basics of our study was to put in a database a representative sample of evictions from our data and record all the evictions by applicant and when they occurred and the same with felony and misdemeanor criminal records.  Then using independent statisticians develop a predictor of how likely evictions would occur.  Now it’s been a while since I took statistics in college but I do remember you could have a statistically valid sample with as little as 67 subjects.  I was also an CPA/Auditor with a major CPA firm for several years and we relied on statistics very heavily on that line of work.  I also remember that the more random sample subjects you have the more statistically valid the results will be.  Since we don’t do anything halfheartedly, we took a sample of a little over 4,000 applicants for affordable housing to include in our sample.  This took many hours of data entry to record the number of evictions and crimes and time period between each.  We finally accumulated the data and sent it to a statistician with a PhD in Neuroscience to analyze and give us the predictors we wanted.

The results were staggering in my opinion.  We had no idea the correlation between getting evicted once or getting convicted of a felony tied so closely to getting evicted again or even getting evicted the first time after a felony conviction.  We were equally surprised at the correlation of misdemeanor convictions and eventual eviction.

Our results were so consistent that we have included an eviction costs savings calculator on our website at the bottom of the solutions tab.  All you have to do is enter the number of units you manage and the % turnover you have in your properties and we will immediately tell you how much you can save by switching to the VeriScreen solution.  The assumptions are listed above the calculation but I would be happy to review the calculation with you via the telephone if they are not clear to you.

I would also be happy to email you the results of our revolutionary study if you just call the office and ask for the President/CEO of the Company.

I’m sure you will find the savings substantial and the results of the study eye opening.



Mark Harmon





Leave a Reply

Your email address will not be published. Required fields are marked *