The perils of selection bias in LBE development

Developing a location-based entertainment (LBE) center is fraught with landmines. Entrepreneurs move forward with their projects, often unaware of many biases that threaten their very success. One potential peril is known as selection bias.

To figure out the winning formula for a LBE, many entrepreneurs go out and look at successful operations and imitate what they are doing. On the surface, this plan seems totally logical. Unfortunately, copying what appear to be successful LBEs can be remarkably misleading and dangerous -- an actual route to failure.

An anecdote the statistical community frequently cites to illustrate selection bias concerns World War II statistician Abraham Wald, who was studying the vulnerability of airplanes to enemy fire. All the data showed that certain parts of planes were hit much more often than other parts by enemy fire. So military personnel concluded that reinforcing those parts would increase the planes' survival rate. Seems logical, doesn't it? Well, Wald came to just the opposite conclusion: the parts hit least often should be protected more.

Wald came to his conclusion because of his understanding of the selection bias inherent in the data. The data was from only the planes that returned. It didn't include any data from the planes that were shot down. Wald reasoned that a plane would be less likely to return if it were hit in a critical area. Therefore, the ones that returned, even when hit, had not been hit in a critical location. Wald argued that reinforcing parts of returning planes that had the most hits would be unlikely to have any benefit.

Here's another example of selection bias. Suppose you want to determine the attributes of successful entrepreneurs. You take a look at all the research done on them and conclude the top two key traits responsible for their successes are:

  • They persisted, even after initial failures, and
  • They had the ability to persuade others to join them.

Sure sounds reasonable. But again, the conclusion falls into the trap of selection bias. Think about it. Those exact same two traits are the characteristic of fabulously unsuccessful entrepreneurs. Incurring large business losses requires both persistence in the face of failure and the ability to convince investors to pour their money into a losing proposition.

Selection bias results from looking only at successful firms, not the entire universe that includes those that failed. Anyone who tries to reach conclusions about what the winning formula is for a business by studying existing companies falls into this statistical trap. If LBE developers study only successful centers, any relationship they infer between their design, mix and management and their success will be misleading and quite possibly wrong. Rather than copying a formula for success, they may very well be copying the formula for failure.

We see selection bias all the time with perspective clients who make generalizations about LBE success by studying existing LBEs. A would-be LBE developer will call us wanting to develop a center with a particular mix, based upon one or more LBEs they observed in other markets. One example we hear all the time is a children's entertainment center with soft-contained-play as its anchor, as they visited one that seemed successful in some large city. What they haven't researched is the soft-contained-play centers that failed, and there have probably been over a thousand. Remember Discovery Zone?

The center they observed may be an anomaly. It might be the one out of the 100 that worked. More likely, they may not have studied the center in enough depth to accurately extrapolate whether, for whatever reason it seems successful, that success will translate into the developers' market. Suppose the center is two years old. It would have cost a lot less to develop two years ago than it will to develop today, especially with recent construction cost increases. How about its rent? The center might be paying only $8 per square foot in rent, whereas the developer's location is $15 per square foot. Is two years really a long enough track record to infer success? Have the center's sales been increasing or declining? Is it really profitable? Looking at a center on a Saturday when it is busy does not mean it is profitable. It could very well be on its financial death bed.

Even with answers to those questions, the experience of over a thousand failed pay-for-play soft-contained-play centers tells us the odds of success are slim to zilch.

We see similar situations with other types of LBEs. Sometimes the 'successful center' a developer wants to model has already failed once and was sold for pennies on the dollar. The new owner, with greatly reduced capital cost, can now operate the center successfully. Or like in the case of the soft-contained-play centers, there are numerous failed examples the developer has no knowledge of.

Avoiding selection bias is one of the advantages of working with a veteran LBE industry consultant, someone who has been active in the industry for at least a decade or more and is aware of roadkill that has occurred. An experienced LBE industry consultant will have seen thousands of examples of both winning and losing LBE formulas and can correctly identify traits that separate the successes from the failures. Knowledge of both successful as well as failed LBE concepts is essential to making informed decisions that avoid the inherent peril of selection bias and result in successful LBE development.