Methods

The following methods serve as a kind of toolbox we draw from depending on project needs. We are not tied to a particular method and are likely to adapt existing ones, and incorporate or develop new methods.

Cognitive walkthrough

Cognitive walkthrough requires an HCI practitioner to ask a set of usability questions regarding a proposed interface. Cognitive walkthrough requires a well specified prototype and task as it is designed to analyze each step in a task to determine whether or not the user is likely to succeed. For example, does the user have the right goal and will the user notice the correct action is available?

Contextual inquiry

Contextual inquiry is an adaptation of ethnographic methods for HCI. The inputs to the methods are similar to ethnography. It places the HCI researcher in the role of apprentice in order to observe and participate in the user’s real work. The outputs of the method are engineering style, box and arrow diagrams that capture and communicate the various aspects of the work (e.g. information flow, physical environment, etc.).

Ethnography

Ethnography is a staple of cultural anthropologists that places the researcher squarely in the research context as a participant-observer.. The broad focus uncovers issues not directly tied to user goals or tasks but that could put the project at risk if not understood. For example, failing to understand data security is important to a group of science users even if the set of tasks appears public.

Heuristic evaluation

Heuristic evaluation requires an HCI practitioner to compare of a given interface to a to set of heuristics or best practices that capture the key points elements of successful interfaces. One heuristics is, “does the system match the real world?” For example, does the interface terminology match the user’s terminology or does it use system centric terms? Heuristic evaluation does not require a specified task and is often considered a discount usability method because it is quick to perform and identifies a large number of problems.

Human Performance Modeling: Model Human Processor, GOMS

The human performance modeling methods generate a priori predictions about time on task for skilled human operators doing routine tasks. This family of methods relies on the composition of tasks from low operators (e.g. a cognitive cycle to send the command to a hand or eye is milliseconds according to the literature).

Personas

Personas are a way of synthesizing a concrete individual instead of an ‘average user’ that is difficult to conceptualize and design for. Often a persona is developed for each type of user of system user and one is highlighted as the hardest to serve. For example, between the flight attendant, business traveler and vacationer the hardest to serve is the vacationer because the will never become an expert user of the system. Personas require their designers to get very specific about who they are from their computer experience to the type of car they drive.

Process Analysis

Process analysis focuses on the larger work process a software application fits into. Ideally, process analysis is done as part of user research. If this is not possible, (e.g. the process has not be established or there are no users) process analysis can be done as a system is being prototyped or after it is in production such that the next version is adapted to fit the process. Based on an understanding of the existing software tools, procedures, and user roles, it can also be effective to propose process change which is often less costly and can be applied more quickly.

Prototypes

Prototypes are mockups of a proposed interface. The purpose of a prototype is to simulate user interaction, often in the context of a user test, with a system that does not yet exist. Based on the kind of data required and time allotted, different types of prototype are appropriate. For an early stage, informal discussion low fidelity paper or an HTML storyboard are often effective. For a detailed test of a highly interactive system, a high fidelity prototype is necessary build in a tool like Director or Flash. Effective prototyping requires an iterative design and evaluation cycle.

Scenarios

Scenarios are created based on user research. Designers will have identified a set of common tasks. Scenarios are specific examples of these tasks. Often designers will test their early stage interface concepts against a set of scenarios to make sure those particular tasks can be completed with the proposed interface.

Storyboards

Storyboards are a visual representation where a step or action in a task is represented by set of consecutive panels much like a comic book. This method is a low fidelity way to flesh out the interaction design required to support an existing scenario and effectively communicate it to users or stake holders.

Surveys

Surveys used in HCI are conducted according to standard research design methods. Their purpose is often to gather data about user populations or measure user satisfaction in a quantitative manner.

Task Analysis

Task analysis is a method used to understand the user’s current task at a fine grain size. Task analysis often involves a hierarchical decomposition that can be used independently to redesign the task or as input to another method like human performance modeling.

Think Aloud Usability Tests

Think aloud usability tests involve users of the system working to complete a well specified task with a prototype or implemented system. The HCI practitioner then analyzed the data collected during the user test, often as video or notes and identified critical incidents based on criteria identified beforehand. These criteria are based on project needs and can vary such that on one project spending 5 or more minutes to recover from an error is a critical incident. On another project a critical incident may be spending 30 or more seconds.

Use Cases

Use cases are a more rigorous, engineering style version of scenarios. They focus on information exchange between a system and a user by specifying who the user is, exactly what information they have, and the system’s response at every stage.