I was listening to another gaming consultant the other day, and he made the comment, “I don’t understand why all the casinos aren’t using five spot Blackjack layouts on their tables.” He continued, explaining that his comment was based on a trial that was conducted at a Las Vegas Strip property several years ago. It was his conclusion, based on the outcome and validity of this trial, that five spot Blackjack layouts allowed the dealers to achieve a higher number of rounds per hour and many more decisions per hour, which increased the casino’s ability to generate more Blackjack revenue.
Although under most situations I would have jumped on this theory in a heartbeat, in this instance I couldn’t. My hesitance to embrace his comment came from the fact I knew this “five spot” experiment was flawed. The test trial was flawed in the same way that a number of trials have unknowingly drifted into the realm of being biased. The trial was established with the correct test parameters and the correct measurement tools; however, the test failed because it was loaded with evidence of bias that favored a positive result for the argument that five spot layouts were best. It involved the human element.
A number of decades ago, a similar trial failed. It happened with a company named Hawthorne. Management at Hawthorne was looking to see what level of lighting would increase work production in their factory. When management established the comparison process, they announced to specific factory workers that they were going to be involved in an important productivity experiment. What Hawthorne management unknowingly did was instill a high degree of motivation in the employees involved in the experiment, which resulted in their trial production exceeding normal expectations. Based on the results of this “bias” trial, Hawthorne switched over to the new level of lighting, satisfied that they had found the correct illumination setting. Were they right? Did this experiment’s results really reflect the true nature of the effectiveness of the true lighting? Was the trial actually a failure? [See the following website for more information: http://en.wikipedia.org/wiki/Hawthorne_Effect.]
The previously mentioned casino conducted a similar test regarding the use of five spot Blackjack layouts. They placed five spot layouts on several Blackjack tables and compared the number of player hands dealt over a specific period of time on these tables with that of player hands dealt on tables with seven spot layouts. The critical mistake made during this trial sampling was identical to that made by Hawthorne. Management informed the dealers participating in this experiment exactly what they were looking to achieve, a reason for instituting the use of five spot layouts instead of the traditional seven spot layouts. The dealers who participated in this trial were excited. Not only were they involved in the trial, but there was a chance that if the five spot layouts results proved to be more efficient, the casino might opt for all five spot layouts. The Hawthorne Effect motivated the dealers to increase productivity when they believed they were being watched and evaluated. In addition, the dealers’ original perception was that a five spot layout would require them to work less versus a seven spot layout because they would be required to deal to fewer players. This perception added to their motivation to produce on the five spot games. Of course, their trial results were overwhelmingly in favor of the five spot games. However, when the casino replaced the existing seven spot layouts with the new five spot layouts, overall production did not increase. Why? Because the dealers were no longer under the “spotlight” of the experiment and their accelerated motion levels shrank back to their previous levels.
Have Other Casino Trials and Experiments Failed?
Unfortunately, I don’t have information on the success of other casino floor experiments or trials, but I would guess there are a large number that produced flawed results. The reason I make this statement is that many casino executives fail to structure casino floor tests and trials logically, usually adding some unintentional form of bias. Why? Because they were never schooled in the particulars involved in doing these types of experiments.
The primary flaw comes from the executive’s lack of statistical understanding. This is an area where most slot executives excel. Slot managers work with statistics on a regular basis and understand the number of observed results they need before they can make a correct decision on the performance of specific slot and video game types. In most instances this requires 10,000 handle pulls on a single machine. Slot executives understand that fewer observations will result in too wide a variance range and/or incorrect findings.
Executives involved in live casino game trials fail to understand this simple statistical dilemma. For example, not long ago I was discussing the benefits of using a continuous shuffling machine (CSM) versus the dealer hand shuffle. The casino executive who initiated this discussion informed me that management had decided to experiment with several CSMs and had planned a parallel trial in one of their gaming pits alongside several hand-shuffled games. After asking him some of the particulars about this trial, he informed me that they planned to run this experiment for one month and then look at the numbers. I stopped him right there. In my opinion, one month worth of observation was nowhere near a large enough sampling. There are many issues that can flaw this sampling—everything from limited table utilization of specific trial games to drastic and unexpected (but normal) statistical fluctuation swings (also known as good and bad luck).
My recommendation was for him to run the experiment for no less than three months, more optimally six months. The executive said that extending the trial any longer was out of the question. Management wanted an answer by the end of the month so they could decide whether or not to add leasing the machines to the next year’s budget. What are the chances this trial provided management with skewed results?
In this situation, the problem element is time. The casino was willing to make the effort to put together the test trial, but it wasn’t able to meet the time commitment necessary to provide the best feedback on trial performance. The “time” is usually secondary to the “human” problem; however, they both contribute greatly, either individually or combined, to the failure of most casino floor live game tests.
Is There a Customer Effect?
The customer may also have to go through a type of “learning curve” during your floor trials. The customer needs to become comfortable with whatever changes have been made to establish your casino experiment. Remember, they’re usually the most important element in the correct data collection process. If the players aren’t familiar with the new process, they may tend to slow down the playing of the game, or they may initially shy away from gaming equipment that is unfamiliar. The customers need some time to adjust to the new situation before the data you collect is true and correct.
For example, if you pull a Blackjack table off the floor and replace it with a house-banked Poker game, it will take some time before the customers come over to the new game and test the waters. This period of awkwardness can be mitigated through game play examples provided by floor supervisors and clear and simple game rule explanations conducted by dealers on those tables [Note: This is another overlooked area in live gaming: establishing employee instruction on what to tell the players.]
When instituting a trial of a known game or procedure with experienced gamblers, the “learning curve” period will be shorter in duration than when the gaming public is naïve to the process. Also, destination resort casinos have a different dilemma. If they decide to wait for the players to grasp the concept of the new game or procedure, they may have to wait a long time. Since their customer base is primarily tourists, the customers will not learn the new game or process until they have become standard throughout the entire gaming market. Destination resort casinos may need to push forward and understand that the “learning curve” factor is the norm, and their trial decisions need to take this into account, since the uneducated customer situation will not change.
How Can You Prevent These Problems from Happening?
In order to reduce the human false motivation factor, place the test equipment, game type or procedure into service with as little “fanfare” as possible. This doesn’t mean reduce the amount of required training. It means to get the trial up and running without making a big “to-do” about it.
Next, once the trial begins, wait for a period of time before you start collecting your data. If the trial concerns a situation or procedure that is common and not unusual, then a short period of time—maybe only a week—is needed for things to quiet down. If it’s a major experiment that will attract much attention, or if the experiment requires a high degree of training by a specific group of dealers and supervisors, it might be most prudent to wait 30 days or more. Once you have collected your data during the more “quiet” period of the trial, there is nothing wrong with going back and looking over the numbers previous to the collection period. This way you can gauge the impact, if any, the Hawthorne Effect had in the beginning stages of the trial.
Of course, time was a big element in nullifying any problem caused by false motivation; it now becomes a bigger element in developing optimal trial data. Again, it’s important that you, as an executive, stress the need for a long enough trial period whenever conducting tests or experiments that involve different variables like chance and human traits, including the effect on your customers.
After you have made the decision to establish a field trial experiment involving new or different gaming equipment to determine the effect of different game procedures, or other gaming floor trial possibilities such as table lighting, music levels and comfort issues, be sure to determine the experiment’s logical progression first. Logical issues might involve duration of the experiment, the minimum number of observations, the influence and possible effects on the experiment due to the human element (both employee and customer), and anticipated data levels to confirm or deny your trial assumptions. Once the data has been collected, take some time to digest the information. Be careful to use an open mind while searching the data for proof to deny or confirm the trial’s success. Remember, the person developing the experiment will cast some bias on the project based on his or her original stance on the outcome of the experiment at hand. Look for possible elements of experiment bias, and if these elements are discovered, either correct the problem or reject the trial data completely and develop new test parameters. If you establish the experiment with an open mind and use care in developing the trial steps and parameters, you stand an excellent chance that the results will be true and accurate.
As a former Nevada Gaming Control agent, casino operator, professional card counter and present gaming consultant, Bill Zender has been involved in various areas of gaming and hospitality since 1976. He can be reached at wzender@lastresortconsulting.com.
Comments
Post new comment