HCI: Lesson 3

Now that we introduced the perception elements and the history of human computer interaction, we can speak about usability engineering. This part will link the two aspects that we saw before, in order to help adding cognition to the design and prototyping of sofwares and applications.

To start working on a design for an application, we must develop the definitions. A user-centered design is an iterative design process in which designers focus on the users. This latter is different from the human-centered design, that is an approach to interactive systems development that aims to make systems usable and useful by focusing on the users. In fact, the term "User" removes the emotional side from the human. Also, during the design, we must think about the user, but also about all the humans that will be involved (maintenance, selling,...). This human-centered design is the goal to reach.

Here are the different steps to follow:

Specify context of use

The first step to design this kind of application is to understand the user and the context of use. Different kinds of methods can be applied:

- qualitative: we make holistic analysis of a phenomenon and interpret selected data that are often small samples. These are inductive.

- quantitative: we precise measure, make statistical analysis and test on a large sample of data. These are deductive.

Now, we can give some examples of specific methods that belongs to one of this category.

Ethnography

This qualitative method is near to the user because researchers observe and interact with them in a real environment. The observations can be done with documentation (images, videos, reports), or by interviewing people. This tool is used early in the process and works well for complex problems. It is useful to identify new problematics. However, the observations can be biased in the frame of the research.

Interviews

- structured: questions created in advance, same order of questions, close questions

- semi structured: parts of interview created in advance, some questions can be reworked

- unstructured: free flowing conversation, open questions

The first one is really useful to do some statistics, the last one for having new and deeper ideas.

Focus Groups

This is a qualitative method that focuses on interviewing several people at the same time. This is like ethnography, but the discussion between all these ones can deepen the topics. It can also help them to exit the researching frame, for them to be more natural. However, social dynamics and behaviours can affect the results.

All theses methods involve an interview by an human being. In fact, the reaction of the one who ask the questions can also affect the answers, so that's a problem.

Specify user requirements

Now that we specified the context, we must identify the granular user requirements of the product.

Persona

In this phase, we create fake characters with dinstinguished behaviours, and we try to understand their need (and not only our need) according to their simulated personalities.

Scenario

These persona will be put inside a scenario with a defined structure:

- background: who are the users ?

- motivations: what are their goals ?

- tasks: what must they do to reach these goals ?

- context of use

- environment: where will they try to use it ?

- challenges: when they use it, what can get their way ?

Design/Implementation

Now that everything is in place, we must design some prototypes that will be used after in the evaluation phase. In this period, we can have two types of prototypes: the low fidelity ones (storyboards, sketching or paper) and the high fidelity ones (softwares, hardwares or fabrications). The prototyping phase is long, but is a succession of ideas and tries that can make the project reach the final goal.

Low Fidelity

This kind of process have quick visual designs in order to focus on the user experience.

Sketches

Sketching is a tool to help expressing, developing and communicating ideas. This part helps generating ideas and elaboring designs. We just need a pen and a piece of paper. These are composed of drawings, parts of text with ideas,...

Storyboards

This method is similar to the previous one, but here we work on the persona and their scenarios. We focus on some functionnalities they can need by using our tool.

Paper Prototypes

These are physical mockups of the interface, mostly made of paper. Here is a creative part where we can use whiteboards, papers, glue or post-its.

Wizard of Oz

This one is the most immersive method. Here, we create a fake facade to give the user the illusion as if the systems exist. It can involve acting and a control panel.

High Fidelity

Now, this part is about focusing on the main functionalities, and on the interaction aspects.

Interactive Wireframe

Unlike storyboards, form builders use actual working widgets, not only static pictures. This only works for the software side of the project. We can use Power Point or Adobe XD to create the designs.

Software

It is like the previous one, but with all the functionalities included. We need fast results and not nice code guidelines. Unity3D or Android Studio are references in the domain.

Hardware

This is similar to the software one (we don't want to have a product), but here we have material things. We can use microcontrollers (Raspberry Pi, Arduino,...), sensors, breadboards, or even ductape.

Fabrication

Here, we can by ourselves fabricate connectors or actuators that are useful to connect two hardwares. Different methods can be used: the additive (stereolithography, selective laser sinterings) and the substractive one (laser cutter or CNC Mill).

And what about prototyping augmented and virtual reality ? Low fidelity is a hot topic because it is hard to rapidly create some models. We can create some real prototypes (with actors and real objects) or paper prototypes. However, high fidelity is currently the most dominant used method, with for example Unity3D where we can have VR tools.

Evaluation

This phase is the one where we collect data on how a real user is interacting with the system. Here again, we can distinguish two types of methods: the quantitative and the qualitative one.

Quantitative method

We can find different methods and studies: in a laboratory (artificial situation created to obtain specific results), a field study (real situation where we try the tool) and surveys (questions).

During the lab experiment, we must make hypothesis (quantifiable, measurable ?), create independent variables (what we can change by hand) and dependent variables (what we measure). We must adapt all these parameters and consider unknown variables.

At the end, we can obtain an internal validity (refers to whether the effect we see on the experiment outputs was actually on the changes we made to the inputs, or caused by unknown variables), an external validity (refers to whether the effect we see can be generalized to the world outside) and a reliability (refers to whether the effect we see is repeatable).

We have also two different cases:

- between subjects: we split the sample into two groups, one using our solution and the other not, and we compare the results

- within subjects: all the people try both solutions, and give notations

Qualitative method

This method here focuses on recorded observations and interviews resulting in a large amout of unstructured data.

This human-centered design process seems to be really efficient, but has some problems. For example, the lack of innovation: any major thing has been created thanks to this method. In order to resolve that, the design thinking process appears.