These theories are stated in terms of concepts such as likelihood of confusion, false or deceptive advertising, and secondary meaning, and are typically described in documents such as complaints.
Authorities increasingly rely on surveys for evidence about these theories. To provide evidence worth considering, surveys must always be reliable and often must establish causation. Both of these issues may require the use of controls. When discussing Lanham Act surveys, I am often asked questions about controls, so this post explains how and why controls are used in intellectual property surveys.
The Two Purposes of Controls
Controls are typically used in surveys either to distinguish an actual measurement from background noise, or to establish causality. Sometimes, but not always, those two purposes are identical. Let me explain each.
In terms of background noise, it is well-known that in a large group of survey respondents, some survey respondents may exhibit certain tendencies as they answer questions. They may be more likely to say “yes” than “no” to survey questions. They may harbor underlying beliefs or attitudes that affect their answers. For example, in a likelihood of confusion survey, some percentage of respondents may believe that every company is connected, even if the products have no relationship with each other and if the product categories are as different as automobiles and ice cream. One role of controls allows a survey to adjust for these tendencies.
A second job of controls is to establish causality. Surveys for Lanham Act disputes often involve questions of causality, which becomes important when claims imply a chain of events, such as:
- Is this trademark likely to cause confusion with another trademark?
- Does this advertisement cause false or misleading beliefs among consumers who see the ad?
Controls are Common in Other Types of Research
Controls are commonly used in other types of research, and we have all encountered controls or used products that were tested by use of controls. A medical example might help explain: if a company developed an active ingredient that they believed cured the common cold, they might test it by recruiting two groups of consumers. One group would get a pill with the active ingredient; the other group is called a control group and would receive a placebo, namely a similar looking pill but without the active ingredient. By comparing the results from the two groups the company would develop evidence as to whether the ingredient causes a reduction in symptoms from the common cold.
Controls are used frequently in many places. For example, a recent article in the New York Times discusses the impact of home video games on exercise among children. The article discussed a particular experiment in which some children received video games that involved exercise, while others received more passive video games. By comparing the two groups, the researchers could establish whether certain types of home video games are associated with higher levels of exercise.
How A Control is Used
In surveys that use a control, researchers typically recruit two groups of respondents: a test group and a control group. The control group receives a control stimulus, which is similar to the test stimulus, but has the potentially-objectionable elements removed. The respondents who receive the control are recruited identically to those who receive the test, with the assignment to test or control often conducted by a computer program using an essentially random process.
The researcher subtracts the confusion associated with the control product from that of the test product. In this way, the use of a control allows a survey to quantify the net or “true” level of confusion that is caused by the presence of the potentially-offending elements.
Sometimes, a suitable control stimulus can be located by using a product or trademark that is available in the marketplace. The researcher also can use products or trademarks that are modified to have as many elements in common with the test product as possible (without any objectionable elements). Alternatively, researchers may use fictitious products specifically developed for the research. For example, I was once retained to conduct a survey measuring likelihood of confusion in an apparel study. In this study, I used a control product that was similar to the potential infringer’s products in all respects, but I altered the logos on labels and hangtags to eliminate the potentially infringing elements. The control otherwise had all elements of the test product in their natural context – the same color scheme, the same font, and a name for the control product that conveyed the essential function and meaning of the test product name.
A control is not always required. The times when it is not required is a matter of some discussion, but controls may not be necessary where causality is not an issue and where baseline noise or biases are unlikely to affect responses, such as with open-ended responses. This short post cannot describe all the circumstances when a control is and is not needed, but the need for controls depends in part on the specifics of the matter for which the litigation survey is conducted.
Dr. Bruce Isaacson
MMR Strategy Group