One of the most common questions I get is about how many participant should be included in an eye tracking study. Well, it is not a simple question to answer. I think it is impossible to come up with a specific number of participants that will work in all cases, but there can of course be recommendations for specific situations. I think it is very important to remember that eye tracking can be used in both qualitative and quantitative studies and that the results should be interpreted and used completely differently depending on this. I have put together some questions I usually have people think about when they ask me how many participants to include in a study:
1. What kind of study will you do?
A) Qualitative research: Usability studies, design studies, focus group studies
– The purpose of the study is to find specific problems, errors, issues that can be improved to make the usability or design better and to get feedback. One occurrence of a problem is enough for it to be counted.
– Only a small number of representative participants from a defined population are required
– The time used per participant for both the test and the analysis is high
– Adding many more participants will not help you find more usability or design problems. With 5 participants you can find 85% of the usability problems, with 15 you can find all. See: http://www.useit.com/alertbox/20000319.html
Results: Eye movement data is only used to describe in depth how the participants in this study behaved while performing a task and which problems were found. Eye movement data is used in combination with other data from interviews, observations etc to help describe the problems and the behavior. Results cannot be generalized to be valid for the population, but they can be used to improve the product.
B) Quantitative research: Usability studies, design tests, package tests, advertisement tests, experiments etc
– The purpose of the study is to describe typical behavior, to find typical problems or issues and to describe how often these occur, and to find differences or similarities between different groups of people. The more a problem occurs to more severe is the problem.
– A large number of participants should be sampled from a defined population
– The time used per participant for both the test and the analysis is low
– The sample size is depending on the following factors:
– The subgroups (cells) that will be analyzed independently like women/men
– The degree of risk involved in the decisions being made based on the results
– The amount of available resources (time & money)
– Which statistical tests will be used for analyzing the data
– Margin of standard error of the mean and significance levels accepted in the results
– Adding more participants will improve the quality of the data; the standard error for 20 participants is much higher than for 50 people.
Results: Eye movement data is used to describe how the typical person or group of people in the population behaves while performing a task. Eye movement data is usually the main source of information but also other data is included. The results are generalized to be valid for a specified population, and sub-groups within that population.
2) What kind of stimulus and tasks will you include in the test?
A) Stimulus complexity
– How complex is the stimulus and what will you analyze? This has an impact on the saturation of the eye movement data.
– Complex stimulus: more people
– Simple stimulus: less people
– How visible are the individual objects important for your analysis?
– High visibility: less people
– Low visibility: more people
B) Task complexity
– How complex are the tasks performed by the participant during the test?
– Complex tasks: more people
– Simple tasks: less people
– No task: more people (most times)