As we work toward more data-driven decision making, it’s implied that our strategies will be less and less subject to human error and prejudice.
As it turns out, that assumption is only somewhat correct. Data is not deployed in a vacuum, therefore it’s still subject to poor interpretation and mishandling. Even if you have a system of checks and balances for data cleaning (which most organizations don’t), there is still room for a slip-up.
Bias can creep into any company’s portfolio data and have a sizable impact on real estate strategy. When you’re making decisions off of the wrong assumptions, your decisions are misguided.
No one is completely immune to this data bias phenomenon. However, identifying various biases and how they could manifest for your team is a step in the right direction for self-correcting.
Confirmation bias arises when a person is trying to prove a predetermined assumption. If you want so badly to be right about something, you might manipulate or interpret the data only in ways that support your preference (this is what we deem a human bias). When you’re in charge of picking the restaurant, you might choose to see only the positive Yelp reviews. You turn a blind eye to the negative reviews, because you’ve already made up your mind to go there.
Maybe you’ve also decided prematurely that you need to renew that lease in Philadelphia. When presenting your business case to your team, you present only biased data that reinforces your preference rather than showing the full picture.
Selection Bias, or sampling bias, manifests when a sample set is chosen subjectively rather than at random from the general population.
Let’s say you’re polling employees to gauge their thoughts on moving the office to the suburbs. You decide only to poll upper management, who can more easily afford the costs of a) living in the city and b) the commute into the city if the move to the suburbs does, in fact, happen. You’ve chosen a sample set that doesn’t accurately reflect the entire population. Because of this selection bias, the data you’re using to make a decision won’t be accurate and you might arrive at the wrong decision.
Outliers in your data can cause for skewed data, which in turn poorly reflects on interpretation and resulting strategy. When you fail to consider the way in which outliers might be affecting your data analysis, you’ll be misled.
Awareness of uncharacteristically large or small data points is the first step in avoiding this type of data bias. Additionally, taking various measurements (e.g. mean, median, and mode) will allow you to look more critically at the data and analysis outcome.
When you’re benchmarking your portfolio against your peers, you might encounter some of these outliers in your dataset due to factors below surface-level. For example, we recently worked with a client who was spending significantly more on build-out than others in their industry cohort. We were able to help them determine that this higher-than-average cost was due to their company-wide standard for building spaces. Outliers should not be ignored our counted out; rather, they should be investigated with an analytical lens to discover their origin.
When working with a commission-based service provider, it’s worth noting the data gathering techniques at work. While brokers are an integral part of a holistic real estate deal team, your internal department should ultimately assume data ownership.
When employing a provider who works on a commission, there’s a conflict of interest. The properties you’re shown might reflect only those that have a large price tag, high commission rate, or immediate availability for maximum number of deal closures.
You’ll need access to market data that’s completely unbiased in order to lock in a strategy that’s optimized for cost savings and ultimately best for your unique organization’s needs. Seeking out an objective third-party software provider for CRE data and analytics will ensure that you’re getting visibility into the entire market – not just deals that are convenient for someone else.
Data bias in real estate is not 100% avoidable, and they are rarely the result of attempts to deceive. It’s rare to make decisions in the complete absence of bias. However, there are steps you can take to help ensure more sound analytics, and in turn, CRE strategy.
Taking the proper steps to clean and validate your dataset helps prevent certain types of unwanted bias from creeping into your portfolio data and analytics. This involves making sure data is error-free, eliminating duplicate data, and standardizing the so-called language of your data. If your team doesn’t have time for this heavy, foundational lift, you can outsource this data cleaning effort to minimize biased models.
While data cleaning and validation serves as a preventive measure, constant monitoring is necessary in order to neutralize statistical bias after it presents itself in a decision-making scenario. When you fail to detect latent bias in your portfolio, the trend will continue and have a negative impact on your strategy. Awareness is the most effective way to ensure that your portfolio, market, and benchmarking data are fit to use.
For assistance in proactively identifying and eliminating data biases in your CRE portfolio, schedule a meeting with RefineRE’s Data Foundations Team.