MLS data quality

[Iris]

I am a researcher studying questions of data quality. Recently I've been working on a project related to real-estate data. I am looking for data, general assessments, impressions, etc. about the quality of the data in MLS databases. Can anyone help?

Iris

[langard]

The data is ugly times two. Imagine a couple thousand uncoordinated MLSs saving data from a few million real estate agents, in 20 different types of databases with no standardization whatsoever, all of which have different abnormal schemas...Then multiply that by the lack of error-loops times the corruptions times the duplications...And you have an incomprehensible mega-blob of erroneous rubbish that you couldn't sort out with a pitch fork.

So it's pretty much like government data. Garbage in, garbage out.

[TCarr]

I'd have to agree with langard - it's not real good.

While each MLS is different, some tend to be better at maintaining the quality of the data input. For example, an MLS is a larger town neighboring us has a policy of fining agents who input bad data (such as wrong area, wrong schools, etc) but ours doesn't.

As a result, we have listings entered in the wrong MLS area, wrong subdivisions, wrong schools and more. Some agents are a lot more conscientious than others.

Plus, lots of fields are open input fields. Like ACREAGE - I would like to give my visitors the ability to search homes by acreage, but for a home with one acre, for example, the field might have "1", "One", or "1.0+/-".

I would love to be in charge of my local MLS and standardize every field and fine the heck out of agents that continue to just put whatever into the system.

And don't even get me started on agents who enter sold data 3 months after the home closed!

[james.ash]

Poor and Poorer, that is how I'd describe it. It is so bad that ARMLS recently insituted a REPORT VIOLATION button, anyone viewing can click on the button to report a violation and the offending realtor either gets taken to the wood shed or is fined. (Ouch)

[drpatt]

I'd have to agree with langard - it's not real good.

While each MLS is different, some tend to be better at maintaining the quality of the data input. For example, an MLS is a larger town neighboring us has a policy of fining agents who input bad data (such as wrong area, wrong schools, etc) but ours doesn't.

As a result, we have listings entered in the wrong MLS area, wrong subdivisions, wrong schools and more. Some agents are a lot more conscientious than others.

Plus, lots of fields are open input fields. Like ACREAGE - I would like to give my visitors the ability to search homes by acreage, but for a home with one acre, for example, the field might have "1", "One", or "1.0+/-".

I would love to be in charge of my local MLS and standardize every field and fine the heck out of agents that continue to just put whatever into the system.

And don't even get me started on agents who enter sold data 3 months after the home closed!

The Upstate (Anderson) Board is terrible about enforcing their own rules. Look at how many "Owners of Record" are entered as "RECORD." Many don't have photos. Greenville catches those things more often I think. Agents are terrible, but enforcement of well-known rules can be very lax.

The acreage thing is a nightmare. Searching is a real mess with so much improperly-entered data. At least for areas, towns and subdivisions we have to pick from a list. Yes, many seem to work hard at choosing the wrong items from the lists.

Greenville would get more correct "area" entries IMO if they would get rid of this politically correct policy of not allowing us to see the area names on the entry screen. Absolutely ridiculous. How could someone be offended by "Six Mile" or "Pickens?"

Doug

[AaronSOLD]

The RMLS of Minnesota, serving the Twin Cities, does a great job with the data. There's a few things here and there that are wrong but all-in-all, the data is clean. Fields are very restrictive in the input they'll accept and there's required fields for much of the data. Not to bad for a system with data input by 1000's of agents.