*By Stefan Manojlović, Anika van der Sanden & Charlotte Song*

**1. INTRODUCTION**

Domestic appliances, especially washing machines, have dramatically changed their design features over the years. Designers and engineers have been developing several models of washing machines trying to highlight the evolution and importance of specific functions, such as introducing ecologic, energy-saving programs. Nowadays these kinds of appliances have improved automatic controls in order to reduce supervision.

The reflection we use as a focus point is: *If simple everyday actions involving technology should be one of the easiest, why is it then that the usage of such appliances one of the most difficult and confusing?*

HYPOTHESIS: Design can be a tool to demonstrate that time and action are directly connected. If applying this to washing machines, we state that to complete a task on a simplified interface less time for the same action is needed, rather than on an common interface.

**2. DESIGN PROCESS**

We imagine that washing machines are a today’s necessity but most of the actions are confusing and leading to misinterpretation of different programs. Our goal was to make a clear interface to get rid of all the buttons and turning wheels, to make a more understandable and pleasant experience. Since the introduction of smartphones people are really used to a touchscreen.

We started with the assumption that the use of existing domestic appliances for users is confusing, in particular washing machines. Our design process started with the design of two paper prototypes: the first one (*Picture 1*) is representing an existing washing machine, the second one (*Pictures 2a and 2b*) is representing a simplified design made by the team.

2.1 THE DESIGN

Washing machine interface (*1*) represents a common washing machine interface. It has a turning knob to select the type of program and buttons like spinning frequency and/or start button.

*(1) Existing washing machine interface*

The interface of the washing machine (*2*) is a touch screen based on icons that represent the various programs. After selecting the program, the details of the washing program would appear by a simple transition.

*(2a) New design interface – menu 1* *(2b) interface with transition – menu 2*

2.2 EXPERIMENT SET UP

To test our assumptions we decided to set up a user test among industrial design students with a total number of 8 participants (4 male and 4 females) between 18 and 24 years old (*Picture 3*). The participants have been already familiar with washing machines, to see if the design of the interface of the second washing machine will affect the interaction.

All the participants needed to perform 2 tasks with each interface. With this approach, we wanted to explore the familiarity of interacting with an object and to see the time they need to complete a task. We measure the time it took to complete a task and analyze it afterwards to support our hypothesis.

Tasks:

- Wash
*coloured* clothes at *40**°C* with *800 spinning* cycles.
- Wash
*black* clothes at *60**°C* with *800 spinning* cycles.
- Wash
*handwash* clothes at *30**°C* with *800 spinning* cycles.
- Wash
*white* clothes at *95**°C* with *1500 spinning* cycles.

After compleating all the tasks, the participants were asked to fill in a questionnaire. Our aim was to get quantitative data in order to compare the interfaces (experience, difficulties, design and interaction).

*(3)** User testing*

3. ANALYZING DATA AND DRAWING CONCLUSIONS WITH *ILLMO*

After collecting the data we used the software *Illmo* to visualize the statistical data.

3.1. FITTING THE MODEL

When importing the data into Illmo, we noticed that the model was a good fit (*Picture 4*), but we wanted to idealize the fit of the model. In order to do this, we used a BoxCox transformation and set the adjusted value to -0.6000 (*Picture 5*).

We observe that after adding an adjusted variable for the Box-Cox transformation, the LLC improved and the AIC went down. Afterwards, we used a private variable of -0.5700, which was calculated by the software.

*(4) Normal model (5) Model with BoxCox transformation*

3.2. SUPPORTING THE HYPOTHESIS

To support our hypothesis we took in consideration the measured time. We used it to explore the relation it has with “*getting familiar with an object*” and “*simplified interface – faster action*”.

Nonetheless, we analyzed some parts from the questionnaire (such as *experience*, *difficulty* and *design*–*interaction*) and imported them into *Illmo* to support our hypothesis.

3.2.1 Getting familiar with an object

Our first assumption was that people get familiar with various interfaces if used several times. To prove this assumption we tested the same paper prototype twice.

In *lllmo* we inserted 4 conditions:

- First time interface number 1 (
*Picture 1*)
- Second time interface number 1 (
*Picture 1*)
- First time interface number 2 (
*Picture 2*)
- Second time interface number 2 (
*Picture 2*)

a) Δ1 & Δ2

(6) Model Log- Likelihood Profile Δ1and Δ2

From the model *6* we can see that the value 0 lays outside the confidence interval. This implies that there is a significant difference between the measured times Δ1 and Δ2.

**CI(2-1) = [-14.5572,-1.21234] (LLP)** |

We can support the significant difference with the CI(2-1) value; since the value 0 is not within the resulting CI we can reject the null hypothesis that both conditions share the same distribution.

b) Δ3 & Δ4

(7) Model Log- Likelihood Profile Δ3 and Δ4

From the model *7* we can see that the value 0 lays outside the confidence interval. This implies that there is a significant difference between the measured times Δ3 and Δ4.

**CI(4-3) = [-8.20413,-5.83805] (LLP)** |

We can support the significant difference with the CI(4-3) value; since the value 0 is not within the resulting CI we can reject the null hypothesis that both conditions share the same distribution.

c) Differences in Averages

*(8) Model Thurstone Average 1, 2, 3 and 4*

**1 = 45.8428 s**
** 2 = 39.4901 s**
** 3 = 33.6864 s**
** 4 = 13.4496 s** |

From this model we can deduce that the averages 2 and 4 (second measured time) are significantly smaller than the averages 1 and 3 (first measured time). Overall the participants were faster with preforming a task the second time. It implies that, *higher is the number of times a user interacts with an object, less time is needed to complete an interaction. *

3.2.2 Simplified interface – faster action

Our second assumption was that a simplified interface would help the user to perform an action faster. In this part, we analyze the difference between the second-time-experiences, (when an action was not performed for the first time), of the interface 1 (*Picture 1*) and 2 (*Picture 2*).

a) Δ2 & Δ4

(9) Model Log- Likelihood Profile Δ2 and Δ4

From the model 9 we can see that the value 0 lays outside the confidence interval. This implies that there is a significant difference between the measured times Δ2 and Δ4.

**CI(4-2) = [-13.0618,-10.7822] (LLP)** |

We can support the significant difference with the CI(4-2) value; since the value 0 is not within the resulting CI we can reject the null hypothesis that both conditions share the same distribution.

*(10) Model Thurstone Difference **Δ2 and Δ4*

From the model *10* we can read that on average the participants are cca. 13 seconds faster. If compared, is higher than the average time it takes to complete the tasks with the interface number 2 (average time 3 is *13,8725s* and average time 4 is *5.58875s*). From this we can conclude that *the same* *action takes less time with a simplified interface.*

3.2.3 Experience, Difficulty and Design-Interaction

Other assumptions we took into consideration were measured through a questionnaire. The data used below is discrete data.

a) Experience

(11) Thurstone Model of Experience

To measure the experience we asked the participants to answer the question “What was your overall experience with washing machine 1 & 2?” with respective answers:

repulsive (0), confusing (1), smooth (2) or desirable(3).

In model *11*, we can visualize that there is hardly any overlap between the two conditions (blue – *interface washing machine 1* and green – *interface washing machine 2*), which means that there is an apparent difference. The peak of the first condition is between 1 and 1.5 and from this we conclude that the overall experience is* confusing*. The peak of the second condition is between the value of 2 and 2.5, which means that the experience is overall *smooth*.

*The model shows that the simplified interface is more positively experienced than the existing interface, which supports our hypothesis. *

b) Difficulty

(12) Thurstone Model of Difficulty

To measure the difficulty the participants were asked to answer the question “How difficult was it to use washing machine 1 and 2 to perform the task?” on a 5 scale, where 1 represents easy and 5 represents difficult.

The peak of the first condition (blue – *interface washing machine 1*) lies between 2.5 and 3.5, and the second peak (green – *interface washing machine 2*) lies between 0 and 1. *Users implied that the second interface is easier to use.*

c) Design – Interaction

*(13) Thurstone Model of Design – Interaction*

To measure the relation between design and interaction 2 separate questions were asked: a) “From a design point of view, how was the difference between interface 1 and 2?” and b) “From an interaction point, how was the difference between interface 1 and 2?”, with respective answers: minor, moderate and radical.

At the beginning, we assumed that design and interaction are strongly related. Whereas, if there are radical changes made in design we also expect radical changes in the interaction.

After analyzing the data, we realized that the model was not supporting our assumption. We can clearly see that the two conditions are not overlapping, thus *radical changes in the design are moderate changes in interaction.*

** 4. CONCLUSION**

Our aim with this project was to demonstrate and understand how can design be supported by data. Our hypothesis stated that design can help understanding the relation between time and action. In our project we related to the interface of a washing machine, because we state that “traditional/existing” interfaces are confusing to use. This can be seen in the section *3.2.3 Experience, Difficulty and Design-Interaction, b).*

Our hypothesis was supported by several parameters, such as:

– Number of times an object was used: *h**igher is the number of times a user interacts with an object, lower is the time needed to complete an interaction. *

– Simplified interface: *less time is needed to complete an interaction.*

– Experience: *the simplified interface is more positively experienced than the existing interface.*

– Difficulty:* users implied that the second interface is easier to use.*

– Design – interaction: *radical changes in the design are moderate changes in interaction.*

As shown, several parameters support our hypothesis “Design can be a tool to demonstrate that time and action are directly connected. If applying this to washing machines, we state that to complete a task on a simplified interface less time for the same action is needed, rather than on an common interface”.

After this short course, we realized that Analyzing Experimental Data can benefit in most of the design processes. Design has mostly been based on subjective assumptions. Our design approach was a common one: brainstorming – concept development – prototyping – user testing – conclusions. What is different in our design approach is the use of data to support our assumptions using a software that visualizes data.

We state that this process should be integrated in common design approaches to help designers support their hypothesis.