designing a maintenance support system interface:...

59
Linköpings universitet/Linköping University | Institutionen för datavetenskap / Department of Computer and Information Science 30 credits, Master thesis | Cognitive Science Spring term 2016 | ISRN: LIU-IDA/KOGVET-A--16/008--SE Designing a Maintenance Support System Interface: Using the Activity Checklist Combined With the Sequence Model Jonna Magnusson Academic supervisor: Mattias Arvola Supervisor at Saab: Mattias Holmström Examinator: Arne Jönsson ISRN: LIU-IDA/KOGVET-A--16/008--SE

Upload: doanthuy

Post on 28-Aug-2018

219 views

Category:

Documents


0 download

TRANSCRIPT

Linköpings universitet/Linköping University | Institutionen för datavetenskap / Department of Computer and

Information Science

30 credits, Master thesis | Cognitive Science

Spring term 2016 | ISRN: LIU-IDA/KOGVET-A--16/008--SE

Designing a Maintenance Support System

Interface: Using the Activity Checklist

Combined With the Sequence Model

Jonna Magnusson

Academic supervisor: Mattias Arvola

Supervisor at Saab: Mattias Holmström

Examinator: Arne Jönsson

ISRN: LIU-IDA/KOGVET-A--16/008--SE

Upphovsrätt

Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare – under 25 år från

publiceringsdatum under förutsättning att inga extraordinära omständigheter uppstår.

Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner, skriva ut enstaka kopior

för enskilt bruk och att använda det oförändrat för ickekommersiell forskning och för undervisning.

Överföring av upphovsrätten vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan

användning av dokumentet kräver upphovsmannens medgivande. För att garantera äktheten,

säkerheten och tillgängligheten finns lösningar av teknisk och administrativ art.

Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i den omfattning som

god sed kräver vid användning av dokumentet på ovan beskrivna sätt samt skydd mot att

dokumentet ändras eller presenteras i sådan form eller i sådant sammanhang som är kränkande för

upphovsmannens litterära eller konstnärliga anseende eller egenart.

För ytterligare information om Linköping University Electronic Press se förlagets hemsida

http://www.ep.liu.se/.

Copyright

The publishers will keep this document online on the Internet – or its possible replacement – for a

period of 25 years starting from the date of publication barring exceptional circumstances.

The online availability of the document implies permanent permission for anyone to read, to

download, or to print out single copies for his/hers own use and to use it unchanged for non-

commercial research and educational purpose. Subsequent transfers of copyright cannot revoke

this permission. All other uses of the document are conditional upon the consent of the copyright

owner. The publisher has taken technical and administrative measures to assure authenticity,

security and accessibility.

According to intellectual property law the author has the right to be mentioned when his/her work

is accessed as described above and to be protected against infringement.

For additional information about the Linköping University Electronic Press and its procedures for

publication and for assurance of document integrity, please refer to its www home page:

http://www.ep.liu.se/.

© Jonna Magnusson

i

Abstract

This thesis aims to investigate how a new interface of the Maintenance Ground Support System

(MGSS) can be designed with insight generated from combining the Activity Checklist with the

Sequence Model. Through observation of an expert user, deficiencies and qualities of the system

should be raised. The combination of the activity checklist and the sequence model generated

fruitful insight over the deficiencies leading up to a troublesome event in the interface. A rich

understanding generated of what led up to a certain activity and how that action could be improved.

The activity checklist explains the users’ behavior and why they are. It provided beneficial

knowledge over the whole operation in the system and not just one functions. The insight from

these two methods set up a ground pillar for a solid design, where the deficiencies are improved

and the qualities are isolated from an untroublesome event. With the knowledge from the analysis

a prototype of a new interface was made in Microsoft Blend. The new interface was evaluated with

surrogate users at Saab AB. Though the result was not significant the result showed a higher SUS-

score and faster total mean time of the performing of the tasks. The result showed a trend with a

better usability and can be generalized and interpret that it would be the same output with real end

users.

ii

iii

Acknowledgement

This thesis would not have been done if there was not a good cooperation with Saab AB. A big

thank you to all that have been supporting me with your knowledge and inspiration. A special

thanks to my supervisor at Saab Mattias Holmström, for teaching me all about MGSS and all your

answers to my questions. I will also give a big thank you to Peter Bohnsack for the opportunity to

visit you at F17 in Ronneby, without the observation the knowledge would have been mostly

wrong! Mattias Arvola my academic supervisor at Linköpings University should also have a big

grateful thanks for the guidance throughout the thesis. You have given me inspiration and self-

confidence during this spring, thank you! The biggest thank you goes to my family and friends,

without you I would never have come this far, I love you all!

Linköping June 2016

Jonna Magnusson

iv

v

Contents

1 Introduction ........................................................................................................................... 1

1.1 Aim and purpose .............................................................................................................. 2

1.2 Delimits and limitation ..................................................................................................... 2

1.3 Saab AB ......................................................................................................................... 3

1.3.1 Maintenance Ground Support System - MGSS .............................................................. 3

1.3.2 MGSS walkthrough .................................................................................................... 4

2 Theoretical background ........................................................................................................... 9

2.1 Behavior of expert user ..................................................................................................... 9

2.2 Affordances ..................................................................................................................... 9

2.3 Signifier ........................................................................................................................ 10

2.4 The designers’ obligation ................................................................................................ 11

2.5 Activity theory ............................................................................................................... 11

2.5.1 Activity checklist ..................................................................................................... 13

2.6 The Sequence Model ...................................................................................................... 15

3 Pre-study ............................................................................................................................. 17

3.1 Needs analysis ............................................................................................................... 17

3.2 Observation ................................................................................................................... 17

3.2.1 Procedure ................................................................................................................ 18

3.2.2 Data analysis ........................................................................................................... 18

3.2.3 Observation summary ............................................................................................... 18

3.3 Result of the Activity Checklist ....................................................................................... 19

3.3.1 Decompositions of goals ........................................................................................... 19

3.3.2 Troubleshooting strategies ........................................................................................ 20

3.3.3 The use of artifacts ................................................................................................... 21

3.4 Result of the Sequence Model .......................................................................................... 22

3.5 Result of combining the Activity checklist and the Sequence Model ..................................... 25

4 Design process ..................................................................................................................... 27

4.1 Idea generation .............................................................................................................. 27

vi

4.1.1 Design decisions on theory adaption ........................................................................... 27

4.2 Design result ................................................................................................................. 28

4.2.1 Prototype walkthrough ............................................................................................. 28

4.3 Evaluation of the prototype ............................................................................................. 34

4.3.1 Measurements ......................................................................................................... 34

4.3.1.1 SUS ..................................................................................................................... 34

4.3.1.2 Time on Task ........................................................................................................ 35

4.3.2 Pilot study ............................................................................................................... 35

4.3.3 Procedure ................................................................................................................ 35

4.3.3.1 Participants ........................................................................................................... 36

4.3.3.2 Ethics .................................................................................................................. 36

4.4 Result of the evaluation .................................................................................................. 36

4.4.1 SUS-score ............................................................................................................... 37

4.4.2 Time on Task .......................................................................................................... 38

5 Discussion ........................................................................................................................... 41

5.1 Result ........................................................................................................................... 41

5.2 Interface ........................................................................................................................ 42

5.3 Method ......................................................................................................................... 43

5.4 Limitations .................................................................................................................... 45

5.5 Future research .............................................................................................................. 46

5.6 Implications .................................................................................................................. 46

References ............................................................................................................................. 49

Appendix ............................................................................................................................... 51

1

1 Introduction

Saab AB is a Swedish company developing aerospace and defense solutions. Their most famous

product is the fighter aircraft Griffin, which is a multirole aircraft. To keep a low cycle cost and a

high sustainability the knowledge about the wear of the aircraft is crucial. In order to generate a

longer lifecycle Griffin is using advanced techniques to provide the ground crew with insight and

understanding of the aircraft wellbeing. Over 6000 variables are logged every cycle to keep track

of the wear of the aircraft. There are a several types of cycles, for instance a ground cycle, when

the aircraft has started but has not left the ground, or a flight cycle which is a logged cycle when

the aircraft is in the air. The variables that are recorded are for instance flight hours, height, speed,

pressure and temperature. These variables are later analyzed by the ground crew staff through their

maintenance system called: Maintenance Ground Support System (MGSS). This system, explained

in detail later in the thesis, provide the ground crew staff with information that helps them improve

the wellbeing of the aircraft. The system enables the staff to perform advance troubleshooting in

the logged data. If any abnormal events have occurred during a cycle MGSS present the fault codes

that represent the appropriate actions for the ground crew staff. The priority of the maintenance

system is to offer a longer mean time between failures and decrease the mean time to repair (Saab,

2016). The results from the analyzed data contributes to a secure and well-being aircraft with low

cost over a long lifetime.

Over the years MGSS has been updated with new functions but the interface design has not been

in focus during this time. The existing interface of MGSS is perceived to be too complicated for

easy control and it is difficult to get an easy overview of the data. Systems that are perceived as

having a poor interface design have issues with loss of efficiency, productivity, and most

importantly, safety (Stone, 2005 p.9-10). The interface should encourage users to carry out their

required tasks in an easy and engaging interaction (Stone, 2005 p.6). In order to create a well-

designed interface of MGSS knowledge about users’ actions and intuition are crucial. This

knowledge will affect the interface design and let the system support rather than limit the users

(Preece, Rogers & Sharp, 2016 p.410). Not only is it crucial to know which actions and activities

that are needed, but an understanding of how these activities are related to each other in order to

see the entire development process is equally important (ibid.). This thesis will investigate if it is

possible to get an understanding of the users’ actions and their relations by combining the Activity

2

Checklist with the Sequence Model. The Activity Checklist is an analytic tool used to get a better

understanding of the user’s activity in the design and provides a guidance and structure for

empirical work on design (Kaptelinin et al., 1999). In order to analyze how the actions are related

and how they unfold over time the actions are modelled, step by step, and analyzed with a Sequence

Model (Beyer & Holtzblatt, 1998 p.89).

The goal is to examine if the troubleshooting of Griffin can be more efficient with a new interface.

The design will focus on usability and not just adding additional functionality, as in the past, to

create a quicker and smoother workflow for the user (Stickdorn & Schneider, 2011 p.82-83). Higher

usability would provide faster and more confident troubleshooting and reduce ambiguity in the

interface for the ground crew staff.

1.1 Aim and purpose

The aim of this thesis is to develop a prototype of a new interface of MGSS, by using knowledge

and insights from the Activity Checklist and the Sequence Model. Can these two method be

combined and generate useful knowledge in the field of design. The main questions in the thesis

are the following:

● How can an interface for maintenance system for aircraft be designed to increase

the usability?

○ Which are the qualities and deficiencies in the existing system?

○ In which way can desirable qualities be improved or implemented?

○ How can the daily work routine be more efficient for the user?

● How can a combination between the Activity Checklist and the Sequence Model

contribute when analyzing a workflow in an interface?

1.2 Delimits and limitation

MGSS is a complex system, to consider the system as a whole in a prototype was not possible.

Delimitations of the thesis were therefore crucial. Agreed upon with the supervisor from Saab AB,

the focus of the thesis should mainly consider the primary use of a daily work routine by the ground

crew staff.

3

End-user of MGSS is not the customer. The Swedish Defense Materiel (Swe. Försvarets

materielverk, FMV) supplies the Swedish armed forces with equipment, hence FMV is the

customer to Saab, but they are not end-users. For this thesis one visit to one of the end-user was

possible, to do an observation. Going back to test and evaluate the prototype was not possible. The

prototype is therefore only evaluated with surrogate users, the development team of MGSS at Saab

AB.

1.3 Saab AB

Saab AB was founded in 1937 and is a Swedish aerospace and defense company. Today they

administrate solutions in fields of air, land, naval, security and commercial aeronautics (Saab,

2016). Saabs main focus is the aircraft Griffin which is a multirole fighter aircraft capable of

performing different missions of air-to-air, air-to-surface and reconnaissance. Four versions, one

launching 2016, have been produced over the years. The focus for this thesis is to redesign the

complex maintenance system for Griffin called Maintenance Ground Support System (MGSS).

1.3.1 Maintenance Ground Support System - MGSS

MGSS is mainly used by the ground crew staff for Griffin to conduct advanced fault analysis after

a cycle. During a cycle over 4000 variables of data are recorded in Griffin, a few examples of

variables are flight hours, height, speed, pressure and temperature. Analysis of these data points

provides control over the aircrafts well-being. For instance, after a certain time of flight hours some

components need service or replacing. Knowing the wear of the aircraft creates control and the

ability to achieve low operating costs, and a longer life cycle by only replacing necessary parts.

The data is recorded on a data transfer unit (DTU) placed in the cockpit during a cycle. The DTU

is a portable device, and after a cycle the DTU is disconnected from the cockpit and connected into

a computer with MGSS. The process is similar to using an usb-stick for transferring data between

two computers, though the DTU is bigger and more robust. When the transfer is finished the system

will unpack the files and provide the ground crew users with big datasets from the cycle. MGSS is

mainly used after a cycle with Griffin to analyze the data and confirm that the aircraft is okay to

perform another cycle. If a fault event occurs during a cycle the pilot receives information about it

directly in the cockpit in real time. All fault events can later be seen in MGSS and the ground crew

staff can troubleshoot the events. After an analysis the staff can verify if the aircraft is okay or if

something needs to be fixed. Some fault events are more common than others and the staff learns

from experience what these events mean and what action they warrant. Sometimes an unusual code

4

appears and further troubleshooting to find what the code means is needed. Today, this is

complicated since there is no easy way for the user to know where to find correct information about

fault codes. Fortunately, majority of cycles does not have any fault events. The data however still

needs to be analyzed by one of the ground crew staff in order to make sure that the aircraft is

functional.

1.3.2 MGSS walkthrough

In this section MGSS will be presented and explained along with pictures from the interface. MGSS

can contain secret information and cannot presented with an accurate walkthrough, the pictures in

this thesis are approved by Saab AB.

MGSS is designed as a traditional window application, instead of changing the view of the interface

a new window opens up containing new information. The start view of MGSS consist a lot of

information of all cycles (Figure 1), one row represents one cycle. A row consists of 19 columns

containing summed up information from the cycle such as time, cycle type and station. Farthest to

the left is a box showing the ID-number of the different aircrafts, allowing the user to only show

data from one select aircraft. In the start view there is also information about the different

administration actions, six small columns indicate different information depending on the letter. In

Figure 1 there are a few rows presented in a red color which means that the cycle has been created

manually. If a cycle has not been logged on the DTU for some reason, the ground crew staff creates

a cycle manually so that at least known variables can be documented, such as flight hours.

5

Figure 1. Start view of MGSS

Column number six is called “Fault Events” and contains the events that occurred during the cycle.

As the system is today there are fault events in almost every cycle, although they are not critical.

The user need to analyze to interpret the criticality of the events and to get more information about

a cycle the user needs to double click on the corresponding row. Then, because of the traditional

window application, a new window opens up containing more information from the cycle (Figure

2).

6

Figure 2. Presenting the new window displaying cycle information

The new window in Figure 2 includes nine tabs, each showing different information from the cycle.

The start view of this pop up window is the second tab Events, containing information of data

collected over time in a cycle presented in a sequence of events. The user gets information of when

the event happened, the event number and name of the fault event. It also provides the user with

the estimated criticality of the fault event, which type of event and a MG number that represent to

which material group the event belongs to. A trained eye can in this view see the event number and

interpret the fault event from that, but sometimes an unfamiliar event number occurs and the user

need to troubleshoot further to get explanation of what cause the event. For instance, at the time

260s in Figure 2 there has occurred a fault event with event number 10, with the name ‘Fault Report

DTU’, and has an Event Type of ‘Fault Report’ and MG number 7. All events with the event type

‘Fault Report’ is listed under the tab ‘Fault Reports’ (Figure 3).

7

Figure 3. The same window as in Figure 2, showing another tab for more information

of a fault event from the cycle

From the view in Figure 3, the user gets the same information as in the previous view in the left

section and new information to the right. Here are three tabs which contains additional information.

In order to interpret the information shown to the right a separate document called Aircraft

Maintenance Publication (AMP) is needed. In this version of MGSS the user does not get any

further information of what the fault event means, it is only providing codes that can be interpreted

with help of the AMP.

8

9

2 Theoretical background

This chapter explains the theoretical framework and background needed to understand the result in

the thesis.

2.1 Behavior of expert user

From start all individuals are at some point novices, but expertise develops over time as a person

matures with training and education in a certain field (Cross, 2004). During the development from

a novice to an expert something happens. Studies has shown that there are differences in how

experts and novices categorizes problems and how they reflect about them have been shown. It

also has shown differences in the brain activity between novices and experts (Sheridan & Reingold,

2014). Further results shown differences in perceptual skills while playing chess, an expert can

distinguish between relevant information faster and therefor spending more visual time on these

areas (ibid.). The novices tend to focus on shallow parts of the problems, whereas the experts tend

to study problems more profound (Chi, Feltovich, & Glaser, 1981). An expert is aware of

limitations and the criticality to distinguish between important from the less important problem.

The experts take advantage of explicit artefacts and restructure the problem into smaller sub

problems to easier solve the task at hand. Additionally, experts tend to be more capable to

generalize knowledge and information in a higher cognitive process than novices (Cross, 2004).

Because of this, experts can recognize and detect underlying principles, rather than focusing on the

shallow parts of the system (Cross, 2004: Ball et al., 2004). The novices on the other hand tend to

use more general principles (Ball, Ormerod, & Morley, 2004).

2.2 Affordances

The term affordance is coined by James J. Gibson within his theory of ecological approach to

perception. He argued that a physical object in the environment contain all information about how

an agent, for instance a person, can interact with it. Affordances refers to the relationship between

an object and a person, and it implies all the properties that the object has and all the possible

actions. In short, affordances determine what actions are possible (Norman, 2013. p.14, 145). The

relationship varies between individuals, the same object can have different affordance to different

people. A knee-high firm surface does have the properties to afford sitting on, but the knee-high

height is however individual, hence is the relationship of affordances (Gibson, 1979. p. 120).

10

Norman introduced the term affordances into the field of human-computer interaction (HMI). For

designers’ affordances provides strong clues to the operations of things and can therefore be fruitful

in design (Norman, 2013 p.12). For example, a door with a flat plate mounted on it affords pushing

and a door with a vertical door handles afford pulling. Users should be able to figure out what

actions are possible without the need for labels or instruction, for instance a label at the door saying

‘Push’ or ‘Pull’. When talking about affordance within interface design, a button that is designed

to protrude from the display do have affordance to pushing and not moving or editing (Gaver,

1991). When taking advantage of affordances, the user knows what to do by just looking and no

further instructions are required. The term was loved in the field of design, and became used to

almost everything, but according to Norman (2013) the usage of the word had nothing to do with

its original definition. An indication of where the user should touch in the interface is not an

affordance, it is just a way of communication where to do the touching (ibid.). Affordance of

touching is existing on the entire screen. So for example designing a circle in the interface is to

signify where the touch should take place, and that is not the same thing as saying what actions are

possible (ibid.). There have been discussions for a further development of the concept, introducing

such a powerful concept but limiting its scope is a failure of its full potential (Kaptelinin & Nardi,

2012). To disentangle the ambiguity Norman coined a new term, Signifier (Norman, 2013 p.14).

2.3 Signifier

Instead of using the ambiguous term affordances, Norman proposed to use the term signifier. A

signifier should not replace affordance but rather be seen as a complement, designers need both of

them. Affordances determine which actions are possible (e.g. display afford touching), and

signifiers communicate where the action should take place (e.g. the circle indicate where to touch)

(Norman, 2013 p.14). Users search for clues and signs in the interface that can help them understand

what to do. A signifier should provide the user with the information that is needed. A signifier does

not need to be intentional, it could be accidental for instance a visible trail in a field after people

had walked (ibid.).

11

2.4 The designers’ obligation

What the term is called is not crucial for the designer nor the users, the important thing is how the

users interpret the designers’ system. There is no need to talk about putting affordances and

signifiers in the interface, but rather of how the system speaks for the designer. Getting the interface

to talk for the designer is a way of metacommunication (de Souza, 2005 p.90). The designers are

trying to represent their understanding in a simplified way, so the users can determine what they

mean in the system, it is a one-shot message from the designers to the users (ibid.).

Norman underline the significance of cognition as central to the users’ activity, and de Souza

complement that we need to emphasize the fact that cognition is also central to the designers’

activity (de Souza, 2005 p.89). The fact that the designers says “I put an affordance there”, to

describe for example a circle to indicate where to touch is a conceptual mistake (Norman, 2013

p.13; de Souza, 2005 p.91). Whether the term is called affordances, signifiers or the designers’

deputy, it is the designers’ obligation to make the interface understandable for its users.

The design need to have clues to enhance the likelihood for the users to make the correct

interpretation. They need to have the not only the answer of “where can I go next”, but rather,

“where is the most beneficial next step to go”. The interface must follow the natural evolution,

rather than random choices of the users (Spence, 2007 p.160). If the next step is not clear for the

user, the interpretation involves a high-order cognitive process.

2.5 Activity theory

Vygostsky, a Russian psychologist, presented that human thinking is not a single mental entity set

apart from the outer world, it is a product of social interaction (Garbin, 2002 p.40). Our cognition

is situated within our social and cultural context. Activity theory is a general conceptual approach

that explains the activity which is a human interaction with the objective reality. An activity can

be anything from shopping in a store to working with a computer task. To understand the human

activity requires a complex analysis (Kaptelinin, Nardi & Macaulay, 1999). The basic unit of this

theory is an activity, which is composed of subject, object, action and operations. A subject consists

of an individual or a group engaged in an activity. An object, or sometimes referred as a motive, is

something pursued by the subject. It can be a material thing, but also more abstract like a plan, as

long as it can manipulate the individual's activity. The object is not a permanent structure, it can

12

change during the activity. An action is a goal-directed process that must be used to complete the

object. An operation refers to all the actions performed by an individual. According to activity

theory there are different artifacts employed in an activity. Individuals mediate their relationship

with the world through these artifacts (Garbin, 2002 p.40). Figure 4 shows a triangle that is

representing the relationship between subject, object and artifact.

Figure 4. The basic triangle of the Activity Theory

(Source: The Cognitive Use of Artifacts in Cooperative Process Management: Rescue Management and Underground

Line Control. (Linköping Studies in Arts and Science, 2002) p.40 Figure 3.1)

Activity theory pose that there are differences between internal and external activities (Kaptelinin

et al., 1999). As mentioned before cognition cannot be analyzed separately in isolation from the

external activities. There is a constant transformation between external and internal that stands as

the basis of cognition (ibid.). These transformations have two directions. The first is when external

activities are transferred to an internal, and is called internalization. An example of this is a person

going from having to look at the keyboard while writing on a computer to being able to write

without having to look. The second direction, externalization, is when internal activities is

transferred into external ones. This could be when a calculation is too hard for mental calculation,

and need to written down or solved with a calculator.

Activity theory have developed over the years and several branches of it has emerged. The original

three connection triangle in Figure 4 was deemed too simple to fulfill the needs of a theory

explaining the relationship between an individual and the environment in an activity. The Finnish

researcher Engström extended the triangle and developed an activity theory toward organizational

change and development (Garbin, 2002 p.41). He complemented it with more connection, by

including communities, social roles, and the division of labour (Figure 5). According to this

expanded view, interaction between a subject and the community is mediated by social rules.

Similarly, the interaction between community and an object is mediated through division of labour

(Garbin, 2002 p.41).

13

Figure 5. The expanded triangle of Activity Theory

(Source: The Cognitive Use of Artifacts in Cooperative Process Management: Rescue Management and Underground Line Control. (Linköping Studies in Arts and Science, 2002) p.41 Figure 3.2)

The concepts and the theoretical foundation of activity theory have been further developed and

oriented into the domains of Human Computer Interaction (HCI). This context makes it possible

for activity theory to be a powerful multi-disciplinary framework within cognitive, social,

technical, and organizational aspects and how these can be studied together (Garbis, 2002 p.41).

The activity theory however does not provide a designer with a solution that can be applied on

specific problems (Kaptelinin et al., 1999). The activity theory has been seen as a promising

framework to use when analyzing work settings of complex technologies, but have been found as

an extensive and difficult to use. It helps the researcher to orient thoughts, but can for instance be

too abstract when it comes to working on a design. For this reason, the Activity Checklist was

introduced, an analytic tool of activity theory (ibid.).

2.5.1 Activity checklist

The activity checklist is an analytical tool established from the activity theory to get a better

understanding of design activities, and is often used during the early design phases for a system, or

for evaluating existing systems (Kaptelinin et al., 1999). It provides guidance and structure for

empirical work on design and evaluation of computer systems in workplaces (Garbis, 2002 p.41).

Further the checklist can help the designer to identify the most important issues and potential

trouble spots (Kaptelinin et al., 1999). The checklist exists in two slightly different versions: the

evaluation version, and the design version, both used to help organize sets of item that potentially

influence the use of a system. This thesis used the design version from 1999 by Kaptelinin, Nardi

14

and Macaulay. The checklist has potential to identify the most important issues or trouble spots

that designers can encounter. It starts general, by examining the whole area of interest, then

focusing on identifying areas in as much depth as possible. When those areas are established they

are analyzed with a breadth-first consideration, followed by narrowing it down and isolate the

problems (Kaptelinin et al., 1999). This breadth of coverage in the checklist will help to ensure that

designers do not miss any areas that might be important to the understanding of the systems they

are working on. The checklist reflects on the core of activity theory but is intended to apply on the

analysis of how people are using, or will use, a system. For this reason, the activity checklist had

divided into four perspectives (ibid.): Means and Ends: Environment: Learning, Cognition and

Articulation and: Development.

Means and Ends:

This perspective involves to which degree the technology facilitates and constraints the

achievements and how it affects obtainment of the users’ goals.

Environment:

This perspective considers that all human lives in a social and cultural world. They achieve their

goals by active transformation of objects through activity. It is an integration target technology

with requirements, artifacts, and social rules of the environment.

Learning, cognition and articulation:

This perspective involves how artifacts affects the mutual transformation between internal and

external activities, which also can be transferred into each other.

Development:

Activities undergo constant developmental transformation. All practice is a result of certain

developments under certain conditions.

The checklist is not a tool to be used in isolation, it is an appendage to the basic principle of activity

theory. Furthermore, the activity theory itself does not provide straightforward solutions that can

be applied directly on specific problems (Kaptelinin et al., 1999). The activity checklist interprets

as top-down approach, and is used to get richer knowledge. It is complemented with a bottom-up

approach, hence introducing the Sequence Model below.

15

2.6 The Sequence Model

One way to make an analysis from the observed data collected during the visit is to modelling it,

for instance with a work model. Work models are built to present and to analyze individual

perspectives of the observed actions. They are not intended to represent the whole organization and

everything that one person does (Beyer & Holtzblatt, 1998 p.89). The model used in this thesis is

called The Sequence Model, and analyses the steps people take to accomplish an intent in work

which unfolds over time - step by step. The action people take are not random they happen for a

reason and individuals do them for a purpose. These action reveal their strategy, and what intention

they have and what matters. Understanding the individual's real intentions are a key to improve

their work practice (ibid.). With this knowledge one can redesign, modify, and remove steps as

long as the user can achieve their underlying intentions. When changing the work tasks, the system

need to support all intentions concealed in the work, the goal is to make the work process more

efficient. A sequence model represents the steps that unfolds over time during work, it becomes a

sequence of actions - steps to achieve an intention. The visualization of the sequence model become

a detailed map to the work that is actually done. In this way the designer can easier see what actions

needed change, and what these changes can contribute with (ibid.).

A sequence model starts with the overall intent of work, and this intention triggers new actions in

the forms of the work steps. A list of all the steps that follows after each other are visualized, at the

detailed level of the collected data. Steps that are causing problems are labelled with a lightning

bolt. When the work of an individual is modeled the sequence model does not attempt to show

pattern or repetition, they will be identified during consolidation.

16

17

3 Pre-study

This section contains the methods and result from the pre-study. The observation and the procedure

are introduced first, followed by the result of both the Activity Checklist and the Sequence Model.

3.1 Needs analysis

In order to design a customized interface, an understanding of the users’ tasks and goals in the

system provides a solid base to stand on during the development. An understanding of which

functions they want and need in the system are crucial (Allen, 1996 p.55). The requirements in a

needs analysis elaborates from data collected of how the system is used by real users. The aim of

a needs analysis is to create a comprehensive understanding of the users’ needs and issues which

lays the foundation during the design development and design decisions (ibid.). In this thesis an

observation at F17 was made to collect as much information as possible of the daily usage of the

system for making a need analysis. The need analysis can be seen as a ground pillar of the

development throughout the entire design process.

3.2 Observation

To gather data of MGSS’s daily usage observations was made for two days at F17 in Ronneby

Sweden. F17 has since 2004 had the responsibility for The Swedish Air Force Rapid Reaction Unit

(SWAFRAP) and they always have a group of Griffin ready to fly. After an aircraft has been on an

operation the cycle needs to be analyzed to see if the aircraft is approved for another mission.

Although the pilot gets all fault events in the cockpit that appears during the cycle, if any critical

event occurs the pilot can tell the ground crew what happen after landing. A crucial event may be

analyzed directly, otherwise the fault search is in no hurry.

Through a contextual inquiry detailed perspectives can be elaborated on how the system is used.

During the observation interviews were simultaneously held with the expert. The interviews were

based on a semi-structured approach, hence it is difficult to follow a strict questionnaire

simultaneous as the situation develop. Observations and interviews are in this inquiry combined to

collect as detailed perspective over the work process or daily activity as possible (Arvola, 2014

p.50). To learn as much as possible during the observation an apprentice approach was applied. In

this approach the observer aims to learn from experts, like an apprentice learns skill from a master.

18

The observer wants to learn about the users’ work, and together define the functions that the system

need to address (Beyer & Holtzblatt, 1995).

3.2.1 Procedure

The observation lasted for two days by following the MGSS expert who has the responsibility for

the data analysis in MGSS at F17. The first day started with a guided tour of the wing, after that

the work with MGSS started. The expert started the daily work routine and simultaneous explained

all events and actions. Though the observation was made from an apprentice approach taking notes

at the same time was not always easy. The learning and doing approach took more time so there

was little time to the sit down and passively observe approach and taking notes. Notes were written

during the observation when the moment allowed it, or sometimes it was asked to get a moment of

silence to take notes. In order to get the information as organized as possible, the notes were

rewritten and analyzed once again later that day. For confirmations about the notes, the expert was

asked if the notes were understood correctly. If there was some misunderstandings or ambiguity

the expert went through the descriptions one more time. Detailed descriptions were always

encouraged during explanations or demonstrations of functions in MGSS. On the second day of

the observation it was the apprentice turn to make the daily routine work, with support from the

expert.

3.2.2 Data analysis

All notes during the observation was made in a notebook size A5 with lines. To get the notes better

organized and analyzed they were rewritten and analyzed in a mediated action sheet and a persona

sheet (Arvola, 2013). The persona sheet can be modified and function as a checklist for developing

themes during a contextual inquiry. This was made to compensate for the risks of losing important

data. To facilitate the analysis a draft timeline over the users’ work routines was created

continuously as the observation proceeded. To help the analysis further a drawn map analysis made

with notes of the user’s artifacts and what they were used to. The data was further analyzed through

the Activity Checklist and the Sequence Model.

3.2.3 Observation summary

The thesis focus on the daily work routine within MGSS, so the observation had the same focus.

There are probably differences from one MGSS user to another. In the observed expert’s daily

routine all cycles from the day before are analyzed the first thing in the morning, at least all cycles

19

that the pilot did not think was a critically fault event during the cycle. All events that the pilot

interpret as critical are analyzed as fast as possible, the regularity is though that no critically events

occur. Every cycle is logged on a stick in the cockpit called a DTU, after a cycle is finished the

DTU’s are stored in a safety locker waiting to be analyzed. Every cycle has an attached log sheet,

containing information from the cycle. On the log sheet the pilot has noted the time of the cycles

and how many cycles has been done. The information is confirmed through a signature by the

turnaround service manager and the pilot. The DTU contains data of all cycles that has not been

transferred into MGSS. When the DTU is inserted in a computer all data is transferred into MGSS

and the DTU clears automatically. It is rare but sometimes the cycle has not been logged on the

DTU. In these cases, the MGSS leader need to create a cycle manually from the information in the

log sheet. The MGSS expert stated that he needs to create a cycle manually every two months or

more rarely.

After the expert had transferred all data from the DTU’s into MGSS troubleshooting started. The

expert took one log sheet at the time, and analyzed all the cycles from that log sheet. To get the

correct information the expert chose to see data only from the current aircraft. The expert then

interpreted the fault events if there was something unfamiliar with the presented codes. The expert

does have a lot of the codes memorized and knows how interpret them. If the code is unfamiliar to

the expert a further analysis and troubleshoot in MGSS is needed. If the cycle was free from fault

events the expert signed the log sheet and moved on to another log sheet and started over. After all

cycles had been analyzed the data was exported into another system that saves the data for a longer

period than MGSS.

3.3 Result of the Activity Checklist

The activity checklist results in three categories, presented below, that was analyzed deeper. These

three categories together with the sequence forms the basis of the requirement that was used during

the design process.

3.3.1 Decompositions of goals

When talking about the goals with MGSS one is roughly talking about transferring data into MGSS,

make a fault analysis and then export the data. To understand what the user wants and need in the

system this big picture must be narrowed down into sub goals, which are presented in more detailed

in the sequence model.

20

A major sub goal in the system was the comparing the information on the log sheet and the data in

MGSS to ensure that they matched. This was the first thing the expert did, was to check the log

sheet to see how many cycles had been done, compered the information in MGSS, and then started

the fault search of these cycles. The expert thought it was much easier to select and sort the cycles

by aircraft name to get an overview. Then he knew that he cannot be mistake the data with another

aircraft. This activity seemed very important for helping the user to feel like they have control over

the situation, and became a requirement for the design process.

In the situations where the DTU did not log a cycle the user must create a cycle manually with the

information from the log sheet. The data is needed later for further analysis over the aircraft, such

as flight hour for the ability to calculate the wear of the material. The user must be able to

accomplish this activity, which lead it to a requirement.

Different kinds of exports are needed for various addressees, though MGSS sort out relevant

information on its own depending on which export is selected. The participant in this thesis did,

some exports more frequently like once a day and others once a month.

3.3.2 Troubleshooting strategies

If there is a fault event during a cycle it is tagged with codes. For instance, a fault event is tagged

with a MG number, an event number and also given a binary code. As it is today MGSS just shows

the user a fault event with different tags, but it does not give the user any information of what these

tags mean. To interpret the tagged codes, they use an external document containing all information

needed. Recently F17 gained access to have two screens, before they needed to tab between the

views on one screen. The participant thought this was a considerable improvement.

The tags from a fault event are represented with different tags, the MG number represent a material

group and the binary code can be translated through the external manual into errors. Knowing all

fault events is not possible, but an expert user learns the fault events that are more common. This

is an example of when an action has become internalize, as mentioned above within the activity

theory.

21

The user at F17 have on own initiative developed a cheat sheet with all of the most frequently fault

codes. They are using this cheat sheet as an external memory but also to avoid the complex

troubleshooting if it is possible. The cheat sheet however belongs by itself in next category the use

of artifacts.

3.3.3 The use of artifacts

To facilitate the fault analysis, the users at F17 use some artifacts. For instance, they use their

homemade cheat sheet mentioned above that contains a summary of the most frequently occurring

error codes. This cheat sheet was a printed on a piece of paper and placed in the physical regular

manual. The physical manual is almost the same as the digital AMP on the computer, but not as

complex. Most of the users do not use the physical manual only the cheat sheet and the manual on

the computer.

Another artifact, not only affected by the MGSS user, but also the pilot and the clarifying leader is

the log sheet. As mentioned above the pilot take some notes from the route, like aircraft number

and time in the air, and then sign it. The MGSS user compares this information with the data in

MGSS and confirms it is correct. Then the user also signs the log sheet confirming the information

is correct. For an easy overview of the information the participant chooses to only look at the

aircraft that matched the log sheet. To simplify and minimize the risk of looking at another aircraft

data, therefore it is a requirement to have the ability to sort out and just look at data from one

aircraft.

Another interesting artifact the participant had was a board on the wall in the workroom that helps

to keep track of all the different exports. After an export the user signed the whiteboard with which

export he had done and when. The whiteboard is an organizer over all the exports that are

distributed over time. In the existing interface of MGSS there is no information about which of the

different exports that had been done, nor when the latest export was. Feedback of the exports

became a requirement for helping the user by taking away concerns and doubt about which and

when the latest export was made. The user should in the new interface be able to find information

about the different exports, when it was done and which of the different exports are done or not.

22

3.4 Result of the Sequence Model

To get a more detailed analysis of the results from the activity checklist the data was also modelled

through a sequence model, described earlier. This to generate a better understanding of the

workflow over the daily work routines. For being able to create a sequence model all activity on a

workday was broken down into smaller steps of actions. From these smaller steps and with the

knowledge from the activity checklist sub goals appeared, and they are in organized in the sequence

model as they appear over time. The sequence model generates a further explanation of how crucial

every step is for the result of this thesis. Figure 6 - 8 are showing the result of the sequence model

analysis.

Figure 6. Result from the Sequence Model of the experts’ work routine, presenting the start of the morning routine

23

Figure 7. The routine of the daily analysis routine

24

Figure 8. The work process if any fault event had occurred during a cycle

25

3.5 Result of combining the Activity checklist and the Sequence

Model

A requirement specification was made from the data collected and analyzed from the observation.

Using the activity checklist and the sequence model combined gave a deeper insight of the sub

goals. It provided information of when they occurred in the daily routine and which steps and

activities that precede and follow them as well as how they may be effected by changes. Further, it

lets the designer see what changes does to related sub goals, an improvement for one step may

influence the outcome of other sub goals. The following requirements stand as a ground pillar

during the design process and has developed through the analysis:

● Have the ability to see information from one specific aircraft

● Clearly see fault events from a cycle

● Get a simple overview

● Export of data, through different exports

● Get feedback on which exports has been done and when

● Being able to create a cycle manually

26

27

4 Design process

In this section the prototype will be presented, along with explanation of the idea generation and

the design decisions.

4.1 Idea generation

The idea generation started from the need analysis, these aspects are important to consider during

the sketching. During the design development all of the requirements are not needed to be

considered in every sketch. The requirements are a good start to have them in mind, but they should

not disturb the creativity. The sequence model contributes with knowledge to consider for the

sketching as well, what actions that are important and in what order they are needed to appear in.

4.1.1 Design decisions on theory adaption

One major design decision was to take advantage of the moto for the design in the cockpit on

Griffin: Don’t need, Don’t Show. The current interface of MGSS do not have this mentality, it

shows all information that the user can be interested in at some point. This leads to an interface

containing a lot of data that is presented all at once, displayed before in Figure 1 – 3.

To show all information at once is not necessary in daily routine work, for any user. Even an expert

users of the system get overwhelmed with all the information which makes it hard to interpret, even

though expert users can distinguish relevant information from irrelevant better (Cross, 2004).

Making a clear distinction between new cycles from an already analyzed cycle is one improvement

that will remove the need to present all information at the same time in the interface. Splitting the

interface into two views will also help all users, experts as novices, to not be overwhelmed by the

interface. A cleaner interface would be simpler to interpret and also provide a more efficient usage.

The actions of a daily work routine were shown to be fairly straightforward in the sequence model

(Figures 6 – 8), and so should the actions in the interface be as well. An interface with the right

clues, affordances, can be as straightforward as the natural work flow. In an efficient interface the

user should be able to figure out which action is the next step to take (Gaver, 1991). The interface

must follow the natural evaluation, as shown in the sequence model, rather than the user’s random

28

search for the right next step (Spence, 2007 p.160). An easier interpretation involves lower cognitive

processes.

One other design decision that went back and forth was the question to have feedforward and/or

feedback on the dates of the exports visible in the interface. The exports are one of the significant

actions of the daily work routine. The routine of the exports probably differs a lot from place to

place, due to their work culture. The user in this thesis did one of the different exports every day,

other exports once a week. The unawareness of how other users act makes it difficult to generalize

the knowledge from the observation. This led to a decision to only using feedback in the interface,

though the feedforward probably differs too much between users.

The existing system do not have a characteristic color that the prototype need to follow. The color

of the system was an easy, and yet rather difficult decision. The argument of using a grey scale

throughout the entire interface are that grey is timeless. Using a background of grey and black text

is not too distinct to the eye, as black text on white background can be. A decision was made to use

different shades of grey to distinguish different areas in the interface. All boxes that are interactive

are in a darker grey, if they contain information with texts and so on the areas is in a brighter grey

to make it easier to read.

4.2 Design result

This section introduces the prototype and explains the design in the new interface. One

improvement that is presented is the possibility to choose between two different views, one simple-

and one advanced view. Further, introducing the interface as a whole with explanation. The

prototype is made with Microsoft Expression Blend 2013. Microsoft Blend is a production tool

created by Microsoft and is a part of Visual studio. Blend is a design tool suited to create graphical

interfaces. It produces Windows Presentation Foundation (WPF) applications, and is represented

in XAML.

4.2.1 Prototype walkthrough

During the observation it was noted that very few of MGSS’s functions was used on a daily basis.

The routine had an obvious sequential process, especially when no fault event appeared during a

cycle. To make the interface cleaner and more convenient to use the prototype has two views that

29

the user can choose between, one simple- and one advanced view (Figure 9). MGSS need to retain

all the advanced settings, but they do not need to be presented all at once. The idea to have a simple

view helps the user to easily do the daily routine; to check the cycle for any fault events, see what

the fault events means and the ability to do different exports. If the user would need to do a deeper

analysis or use some other functions they just need to switch to the advanced view. Two views

provide a simpler interface that is not overwhelming, which contributes with an easier overview

for the user.

Figure 9. The user can change between the simple and the advanced view in the upper right corner

In the beginning of the analysis the data from the DTU need to be transferred into MGSS. This is

done in same way as before, by docking the DTU into the computer and retrieve all files. Figure

10 shows the start view of the system after data is transferred into MGSS. Some qualities founded

in the observation are retained in the prototype. For instance, the ability to select and to see data

from only one aircraft is retained. However, in the prototype there is more focus on the ability to

choose between the aircraft than it is in existing MGSS. This also provides with a naturally distinct

bridge between the selected aircraft and the information to help the user know what cycles to

expect.

30

Figure 10. The start view of the prototype

With less information in the start view the prototype gives a more comprehensible overview, first

the distinction between the new cycles and the old ones. The new cycles are presented in first box

of the tab name “Events”, which is a tab connected to a box containing cycle history. The current

interface present both new and old cycles together in the start view with nothing to separate them

except the date tag. Another new feature in the start view is the indicators of the fault events.

Additionally, the current version represents fault events or alarms in text in one column at start

view. However, in the new prototype the information is summed up and categorized by its

criticality presented with three different symbols. The green square represents normal events that

are supposed to be logged. The yellow circle indicates events that are not critical, but needs to be

noticed. The red triangle indicates critical events that often needs to be dealt with.

31

Figure 11. Information from one selected aircraft

In Figure 11 only information from the aircraft 222 is selected and viewed. The button that is

selected has become an arrow indicating for the user what to expect. The user clicks on a cycle to

get more information about it, this opens a row that contains events in sequential order as they

appear over time. For instance, the cycle in Figure 12 has no fault events, but all the other non-

critical events are presented as they occurred over time. When a fault event occurs it will be

displayed in like Figure 13, the events are marked with a red triangle and the given event number.

32

Figure 12. When a cycle is selected all events from it are shown

Figure 13. A cycle where two critical fault event has occurred

33

An expert user automatically understands how to interpret the most frequently occurring event

numbers, and do not need more information about the given fault. Using the three new symbols is

a way to peel the interface by removing unnecessary perception input. If a fault event is critical it

will be marked by the red triangle, and the user do not need to interpret the criticality of a code by

itself. If an unfamiliar fault event number appears, the user can now search for more information

directly in the system, by using the search box to the right in the interface (Figure 13).

When the fault analysis is done and the cycles confirmed, the data needs to be transferred into

another system. As mentioned previously, there are different exports needed to be done for different

systems and purposes. In the existing interface of MGSS the exports are not gathered, they are

distributed over three places in the interface. In the prototype the exports are gathered in the same

view and, in order to get a natural transit from the “Event” tab to the “Export” tab they are placed

next to each other (Figure 14). The exports concern all aircraft and therefore the ability to choose

only one aircraft is faded. The user can export just one cycle manually directly from the view of an

event if that is needed.

Figure 14. The exports view, all exports are gathered and contains feedback from previous exports

34

To help the user feel in control of all the different exports, the prototype presents the user with

information from previous exports. It presents when the last export was done as well as how many

days ago this was, this is done to lower cognitive strain for the user. The box at the bottom of the

interface contains history over all the exports that has been done.

4.3 Evaluation of the prototype

The new interface needed to be evaluated against the existing system, using System Usability Scale

and Time on Task to present the differences and to see potential improvements in the new interface.

Because there was no opportunity to test the prototype with an end user, an evaluation was made

with surrogate users. The participants in the evaluation was the development team of MGSS at

Saab AB in Linköping.

4.3.1 Measurements

This section describes the two methods used in the evaluation: System Usability Scale (SUS) and

Time on Task. The result from which was analyzed in the program SPSS version 22 developed by

IBM. Both tests were analyzed through Levene’s Test for Equality of Variance, to see if the

variance is equal between the groups. Further investigation to see if the results are significant was

made through paired sample t-test.

4.3.1.1 SUS

SUS is a popular method for measuring and comparing usability in different interfaces. SUS is

often referenced as a quick and dirty method, which means that the method is cheap and easy to

utilize (Bangor, Kortum, & Miller, 2008). John Brooke introduced SUS in 1986, and it is a

questionnaire with ten questions, half are negative formulated and half are positive. All the

questions have an answer of a scale between one and five. These scales are then transformed into

scores, and is calculated into a total score that generates the overall usability of the interface

(Brooke, 1996). The calculation generates a SUS-score between 0-100, where 100 is a result for a

perfect system. A score under 50 is a non-acceptable interface, a score that lays between 50-70 is

on the marginal to fairly good interface, an acceptable interface should have a SUS-score over 70

(Tullis & Albert, 2013 p.139). The calculation of the total score follows: the score from the positive

formulated questions takes the score from 1 to 5 that the participants have grade the questions and

then takes the score minus one to get the SUS-score. The questions of a negative formulation takes

35

minus five on the given score. For a total SUS-score all these new values are multiplied with 2.5,

and the total score ends up between 0-100. The higher the score is, the higher one can say is the

usability in the interface is.

4.3.1.2 Time on Task

Another usability evaluation made in this thesis is Time on Task, it is simply the time the user need

to complete a task. This is, in most situations, a good way to measure the efficiency of a system

(Tullis & Albert, 2013 p.74). Reducing the time is particularly important in these kind of systems

like MGSS, where tasks are performed repeatedly by the same user. The more frequently a task is

performed in the system by the same user, the more important efficiency becomes (ibid.).

Collecting data with Time on Task is rather easy, it is simply timing the user when they start a task

and stop the time measure when the user has completed the task.

4.3.2 Pilot study

A pilot study was made with one participant from the development team for test the setup of the

evaluation of the two systems. Formulating the tasks was challenging, it is a big difference between

having knowledge about the system and knowing how the daily routine is done. All tasks were

rephrased except the last task. The tasks in the evaluation was formulated to resemble a work

situation, for instance task number three in the prototype was before the pilot study formulated as:

“Do a clarification of the aircraft 222 latest cycle”. This formulation was experienced as confusing

because the development team does not have knowledge about how the daily work routine is done

and how they determine clarification of an aircraft. This led to formulate the tasks to be more

exactly of what actions are expected of them in the interface. Task number three now reads: “Did

any fault events occur during aircraft 222 latest cycle?”. Except for the tasks changes the pilot study

went well, and no other changes with the setup of the computers was made.

4.3.3 Procedure

The evaluation was made with one participant at time, in small room close to the participants work

desks. During the evaluation two computers was needed, one that had MGSS, and one computer

with the prototype. It was not possible to use the same computer to both tasks because of the

security needed for the existing system. The laptops were similar in size and did not differ in

performance for these evaluation. Both laptops had an external mouse connected. The participant

performed four isomorphic tasks in the two system (see Appendix). Before the participants started

36

the evaluation they received information about the procedure of the evaluation. Information about

how they were going to test four similar test in both systems and the time for each tasks was going

to be measured. After the test of one system was finished the participants were asked to fill out a

SUS-survey, and then do the same procedure in the next system.

During the evaluation the participants received the tasks one at a time printed on a paper. They

were asked to tell when they had read and understood the task so that the time measurement started

first when they started the task. Finally, they were asked to tell when they had finished the task to

stop time measurement. Before the participants started the evaluation in the prototype they got an

explanation of the interface. This is because the system is not supposed to be used by novices, the

actual end users are experts and gets education in the system before they use it. They obtained an

explanation of the three new symbols and the idea behind them, and what their purpose in the

interface are. Furthermore, the participants got information of the new possibility to search for

more information about the event number directly from the interface and that no external AMP is

needed.

4.3.3.1 Participants

The participants in the evaluation was no ordinary end-user and was so called surrogate users. All

six participants were employees at Saab AB. Some information in the existing MGSS is secret and

can only be seen from people with permission to use MGSS, which limits the number of

participants. All participants in the evaluation are from the development team of MGSS. They do

have a great knowledge about the systems functions, but not how the system is used in a daily

routine. Among the participants there was one woman and five men.

4.3.3.2 Ethics

Before the evaluation started all participants received a short introduction over the evaluation

scope. They obtained information about the data collection and that all data was handled

confidentially, that no data could lead back to the participant. A verbal approval was then made

before the evaluation started.

4.4 Result of the evaluation

Results are presented as they has developed over time, first a short summary over the results is

presented, after that follows the summary of the evaluation. Different qualities and deficiencies in

37

the system emerged during the observation and are presented in the result of the sequence model

and the activity checklist. These results generated a short requirement list. The most crucial

information that emerged from the observation was the overview of the daily work routine. It

provided insight and knowledge about which events that needed a shortcut or which events that

was experienced as problematics. All this fruitful information helped the designer to not implement

the same deficiencies in the prototype once again as well as the ability to keep the good qualities.

The sequence model also provided information about how straightforward the proceeding in the

interface is. Knowing the flow of the work routine also provide knowledge about what information

to show in a given context rather than showing all information again, which can provide for a

cleaner interface.

4.4.1 SUS-score

Levene’s Test for Equality of Variances of time on task showed no significant result. This verifies

the assumption for homogeneity in the group. The total mean of the SUS result, which is analyzed

after SUS-survey instructions, is shown in Figure 15. The prototype got a higher mean on the SUS-

score (n = 6, M = 71.25 & SD = 13.1) than MGSS (n = 6, M = 44.6 & SD = 23.3). MGSS’s mean

SUS-score was below 50, which interprets as a non-acceptable interface. For the prototype the

SUS-score precisely got over the threshold, 70, for being an acceptable interface. A further analysis

with paired sample t-test did not show any significant results between the SUS-score (p > 0.05).

Figure 15. The mean of the SUS-score for the two systems, error bars with confidence interval at 95%

38

4.4.2 Time on Task

Levene’s Test for Equality of Variances of time on task showed no significant result. This verifies

the assumption for homogeneity in the group. A paired-sample t-test showed a total mean time on

25.4 seconds in MGSS, (n = 24, M = 25.4, SD = 18.17) and for the prototype the total mean time

was 21.9 seconds (n = 24, M = 21.92, SD = 26.1), see Figure 16. Further analysis between the total

means did not show a significant result (p > 0.05).

Figure 16. The mean of the total time it took for the participants to complete all tasks, error bars with confidence interval at 95%

The result for the time on every task individually is seen in Figure 17. Throughout the evaluation

the participants mainly finished the tasks faster in the prototype than in the existing one. Only task

two took longer time for the users to complete, but as mentioned above the mean time to complete

the tasks was faster in the prototype. Paired sample t-test was also conducted between the total time

between each task. The result from task one showed that the prototype was significantly faster (t(5)

= 2.97, p < 0.05). In task two existing MGSS did not show a significant faster total time (t(5) = -

2.24, p < 0.05). For task three the prototype showed a significant faster time (t(5) = 3.00, p < 0.05).

In the last task, task four, there was no significant result between the two systems (t(5) = 0.955, p >

0.05).

39

Figure 17. The mean time on individual tasks

40

41

5 Discussion

This section is divided in different sections, first it focused on the result, method and interface. It

also contains discussion about limitations, future research and implication.

5.1 Result

The results of the SUS-score was higher in the prototype than in the existing interface, which points

towards that the prototype do have a higher usability than the existing system. Using a paired-

sample t-test did however not show any significant result, this could be a result of having few

participants. The result of the prototype is more consistent, a lower standard deviation (SD = 13.1),

compared with the SUS-score of MGSS where the standard deviation varies greatly between

participants (SD = 23.3). Even though the participants are surrogate users the prototype seems to

have a higher usability according in SUS measurements.

The prototype got a faster total mean on time on task. To save time on tasks that are performed

frequently by the same user is an important improvement for the usability and the efficiency of the

system (Tullis & Albert, 2013 p.74). MGSS is used daily, and the troubleshooting is almost always

done with the same routine. As mentioned before the more frequent a user performs a task, the

more important efficiency becomes (ibid.). The participants did accomplish the tasks faster in the

prototype, though with no significant result of the total time. A further analysis on the tasks

individually showed significant result on all tasks except the fourth. Task four did not take many

seconds for the users to finish, and with so few participants it is difficult to get a significant result.

The time on task two however was the mean time significantly faster in MGSS rather than the

prototype. The task concerned interpretation of fault events, the faster time can be explained by

that the participants are familiar with the existing MGSS and knew were to get the right

information. The prototype however does have new symbols that indicates the fault analysis, which

the participant was not familiar with. To accomplish task three similar actions are needed from

both programs and the prototype have a significant faster time than the existing system. With this

result it is argued that there is a very low learning curve in the prototype, once the user knows how

to operate in the system it is faster to use and accomplish the daily routines. The users did however

finish all of the tasks on a faster mean time in the prototype, which indicates a more efficient

interface. This prototype is one way to make the daily work routine more efficient to use.

42

The result shows both a higher SUS-score and a faster mean on time on tasks in the prototype.

Even though the evaluation was made with surrogate users the results indicates a trend of a more

efficient interface with a higher usability in the prototype compared to the existing system.

5.2 Interface

One of the most important improvements in the new interface is the single view functionality.

Existing MGSS is designed in a more traditional windows application, instead of changing the

view, a new window opens up. For instance, if the user wants to see the fault events in the existing

MGSS, a new window opens up containing further information. The single view application in the

prototype contributes to actions being more fluent and natural for the user. For example, to get

more information about the fault event in the prototype the row expands with a click and shows all

information in the same view. To create a more natural flow of the actions in the interface the

exports in the prototype are visible directly from the start view, and not distributed as tool functions

in the menu structure as it is the existing interface. The export in the interface now follow the

naturally flow of actions, rather than random searching the menus for the right export (Spence,

2007 p.160). If users know from the interface where to go next, they will experience less cognitive

load (Sheridan & Reingold, 2014). To tell the user what to do next the prototype uses affordances

(Norman, 2013 p.14) For instance, the button for the exports does not look active in the start view,

but do have the affordance of the possibility to be active (Gaver, 1991). An expert user is also better

in riddle information and can more rapidly distinguish important from less important information

(Cross, 2004). Which is the argument for only showing the fault code and not the description when

a fault event occurs. They know how to interpret the most frequently common codes, when an

unfamiliar code appears they need to search manually for a description. Scaling the information is

an improvement when it is certain what the users can and what they cannot. Taking away

information that is obvious for an expert user creates a cleaner interface which provides an easier

overview that is more efficient for an expert user to use. Although an expert user is better on

distinguish relevant information from irrelevant information (Cross, 2004: Sheridan & Reingold,

2014) the interface can help the user to focus more on the interpretation of the fault code rather than

using cognitive processes on sorting out important information.

The interface of the prototype was needed to be less complex, but access to all the functions of the

system is still important. This led to the decision to have two different views that contains levels

of functionality. During the observation the expert user showed the daily routine which mainly did

43

not include the advanced functions. The daily routines in MGSS showed to be straightforward and

not that complicated, as is presented in the sequence model described in section 3.4. A simple view

however does need to have access to the more complex view, when more advanced troubleshooting

is needed. Which led to have a switch at the top of the interface (Figure 9) so the user easily can

change between the views.

The vision to create simplicity in the interface requires decisions about what information is

important in the given context. MGSS which is a maintenance system for Griffin need to adapt the

mentality that Griffins has: “Don’t Need, Don’t Show”. The design in the prototype is built on

sequential actions, to make navigation in the interface as clear as possible. The sequential actions

are based on the daily routine that proved to be straightforward during the observation. For instance,

the expert user first chooses which aircraft he wants to analyze, and then selects to see only

information from that one. This led to the design decision to have a more distinct area for the ability

to do this action. The new design does also clearly show which aircraft that is selected with the

button changing into an arrow pointing at the data in the adjacent area.

Another design decision was to keep the interface clean from all inputs in existing MGSS and to

distinguish new cycles from the older already analyzed cycles. Though the old cycles need to be

saved within the system for a longer period to make export at a later time possible. The data

however do not need to be clustered together in the same view as in the existing interface.

During the observation the need of overview of the exports was noticed. Today the user had all the

latest export noted on a whiteboard in the workroom to keep track of all different exports. Every

time a new export was made the user manually noted which export it was and the date on the

whiteboard. To help the user keep track of the exports the prototype shows when the latest export

was made, with the date but also with a counter telling how many days ago it was to minimize the

cognitive load for the user. History from exports is also shown in the prototype to help the user to

feel confident and to be able to double check that no exports have been missed.

5.3 Method

The thesis did not only have the purpose to create a prototype of a new interface, it also analyzed

the potential advantages of combining the Activity Checklist with the Sequence Model. Combining

these two methods led to a lot of profitable insights of the system. First the activity checklist was

44

used during the observation, for question inspiration and later for a more fruitful analysis of the

collected data. The activity checklist was needed to be analyzed iterative for a rich understanding,

every iteration gave more useful information and helped to organize sets of items that did not

surface earlier (Kaptelinin et al., 1999). Analysis of the activities the user performed during the

daily work routine gave information and hints of what actions the interface needed to support. The

observed user was an expert of the system and knew where all the functions was therefore the

collected data transmitted knowledge about the functions qualities and deficiencies and how they

could be improved. Furthermore, the sequence model pictured the work flow and gave a clearness

over the actions, and where in the system the actions seemed difficult and troublesome. The

sequence model generated a solid ground pillar to lean on during the design process. The model

showed what step the user would take next, expanding on this knowledge provides the interface

with a seamless flow with a natural evaluation (Spence, 2007 p.91).

Combining these two methods gave a deep understanding of the users’ actions and behaviors in the

system. The activity checklist gave inspiration during the observation over small details, it also

helped to find the deficiencies that needed to be changed and qualities that needed to stay in the

interface. Later modelling of the usage process in the sequence model gave a richer and an easier

way to understand all the information that appeared from the activity checklist. All sub goals that

was found in the activity checklist can now be seen in the sequence model in section 3.4. It also

provided a modeling of all actions that is leading up to a sub goal, and which of these actions that

seems awkward for the user. Using these two methods together was very useful, it gave insight and

clarified both deficiencies and qualities of the existing system. Knowing what functions needed

after each other, (i.e. looking at the operation in the sequence model) made the prototype

development easier. The activity checklist also gave insight in all sub goals, the troubleshooting

strategies and also how the user use artifacts for an easier operation in the system. Every iteration

of the two methods gave more fruitful information about the users’ behaviors.

As mentioned earlier the more frequently a task is performed in a system by the same user, the

more important the efficiency becomes (Tullis & Albert, 2013 p.74). Hence, measuring the time is

important for MGSS, it is used daily by one user. The result for the mean time demonstrated that

it took longer time for the participants to finish all the tasks in existing MGSS, even though the

participants never have seen the prototype before the evaluation. This result can be interpreted to

mean that an expert user with years of experience of the new interface should be able to perform

the tasks even faster.

45

The idea of the measuring with SUS was to get a quick validated insight to see which system the

participant experienced to be the most usable interface. The participants did perceive the SUS

survey a bit tricky, for instance with the question: “I think that I would like to use this system

frequently”. They were then asked to answer how the interface spoken to them, and the experiences

of the systems. The SUS survey was used after both evaluation, so the user that did find the

questions tricky did it with both systems which concludes that none of the systems influenced

negative from this aspect.

The evaluation was made with surrogate users, since it was not possible to test with real end-users.

Neither the tasks nor which system they started with were balanced in the evaluation. The tasks

were not randomized with the argument that the evaluation should resemble a real work situation,

and the flow of actions is in a certain way as the sequence model showed. Hence the order of the

tasks starts with check for fault events and finishes with an export. The argument for not balancing

the systems are the notion of no effect on it. All the participants are developers of the existing

system and have a great knowledge of the system and they are working with it all day. The

evaluation was made during work hours and they were all working with the existing system before

they came in and did the test. Switching the start system on half of the group would not give any

difference in result, because their ability to generalize their skills (Charness, 1976) to the new

interface would not change the result. So the bigger struggle was to get the participants to

understand the question, because they do not have a knowledge about the work routine in the

system. This lack of knowledge could probably provide a misleading time for the participants that

started the evaluation with the prototype.

5.4 Limitations

The most crucial limitation in the thesis is the missing of end-users during the evaluation of the

prototype. Using surrogate users with minor knowledge about the daily routine is not optimal.

Though the evaluation was made with the development team which led to a more generalizable

result than using participants with no experience of the system. Although using random participants

was not an alternative since the existing system is not censored of the secret information, which

means that only confirmed people are allowed to use the system. This is also the reason that there

were only six participants in the evaluation.

46

5.5 Future research

This thesis showed an example of an improved interface of MGSS, but the need of evaluation the

interface with real end-users are crucial to the ability to confirm the improved usability of the

interface. The idea of having two views, one simple- and one advanced view, need to be tested and

evaluated with end-users. This thesis did however not concern the advanced troubleshooting view.

A future research of how the interface in the advanced view should be designed is needed.

It would be interesting to combine this kind of study over different fields, it is not only aircraft who

use and need these kind of complex maintenance systems. To do an over field study would maybe

contribute with profitable insights.

A broader inspection about what every possible outcome, what can go wrong in every step in the

sequence model would give a fruitful insight of needed improvements if the system ever needed in

crisis (i.e. a situation of war). This analysis could be done by for instance with Functional

Resonance Analysis Method (FRAM) by Erik Hollnagel (2012). Which is a method for describing

what has happened or what may happen.

5.6 Implications

MGSS helps Griffin keep a low maintenance cost with high reliability for a longer life cycle.

Rigorous safety requirements are crucial in tense situations. Time is also crucial in a situation of

war, therefore saving time for the aircraft troubleshooting should be prioritized. Air force is a pillar

in military defense and in war every second counts. Griffins have a good rumor about the ability to

be repaired, now it is time to make the pre work (i.e. troubleshooting the aircraft) more efficient.

A new interface can provide less troublesome and confusing information in the system, which can

provide a quick and confident troubleshooting.

47

6 Conclusion

The thesis presented a new way to think when designing a maintenance system. It presented an

interface with simplicity, in the purpose for the user to not be overwhelmed with information. To

help the user to interpret data correct is crucial when handling big data sets. This maintenance

system is used frequently by the same user that in one point becomes an expert user of the system.

Expert users do not need to receive information repeatedly, they can interpret the situation from

the context. They memorize and categorize information differently from novices and by supporting

this expertise the work time can be reduced. It is shown in this thesis that a simpler interface tends

to provide a faster time to finish the daily routine work. Saving time can be crucial in a situation of

war. The evaluation of the prototype showed that the prototype has a higher usability than the

current interface. This new interface the user tended to finish the daily work routine faster and they

experienced it to have a higher usability. These two improvements are important when dealing with

a system that is used in similar way frequently by the same user.

The combination of the sequence model along with the result from the activity checklist provided

with what information that is crucial within different context. Together these two methods

generated an easy way to understand how behaviors and actions elaborates over time in the system.

To see which actions that solves a certain sub goal, which can be seen as qualities. The combination

of the two methods also provided information of which actions that needed improvement, but also

which actions that were not needed at all. The usage of these two methods together gave a solid

insight and helped to define both deficiencies and qualities in the system.

48

49

References

Allen, B. (1996). Information tasks: toward a user-centered approach to information systems. San

Diego, Calif: Academic Press, 1996.

Arvola, Mattias. 2014. Interaktionsdesign och UX: om att skapa en god användarupplevelse. n.p:

Lund: Studentlitteratur, 2014, 2014.

Arvola, M. (2013). The Mediated Action Sheets: A Framework for the Fuzzy Front-End of

Interaction and Service Design. In Crafting the Future: The 10th International European Academy

of Design Conference. Gothenburg, Sweden, April 17 - 19, 2013.

Ball, L., Ormerod, T., & Morley, N. (2004). Spontaneous analogising in engineering design: a

comparative analysis of experts and novices. Design Studies, 25, 495-508.

Bangor, A., Kortum, P., & Miller, J. (2008). An Empirical Evaluation of the System Usability Scale.

Intl.Journal of Human-Computer Interaction, 574-594.

Beyer, H., & Holtzblatt, K. (1998). Contextual design: defining customer-centered systems. San

Francisco, Calif: Morgan Kaufmann Publishers, cop. 1998.

Brooke, J. (1996). SUS-A quick and dirty usability scale. Usability evaluation in industry.

Chi, M., Feltovich, P., & Glaser, R. (1981). Categorization and representation of physics problems

by experts and novices. Cognitive Science, 5, 121-152.

Cross, N. (2004). Expertise in Design: An Overview. Design Studies, Vol. 25, No. 5, pp. 427-441,

2004

De Souza, C. S. (2005). The semiotic engineering of human-computer interaction. Cambridge, MA:

MIT Press, 2005.

Engeström, Y. (1987). Learning by expanding: an activity-theoretical approach to developmental

research. Helsinki: Orienta-konsultit, 1987.

Garbis, C. (2002). The Cognitive Use of Artifacts in Cooperative Process Management: Rescue

Management and Underground Line Control. Linköping Studies in Arts and Science.

Gaver, W. W. (1991). Technology affordances. In Proceedings of the SIGCHI conference on

Human factors in computing systems (pp. 79-84). ACM.

Gibson, J. J. (1979). The ecological approach to visual perception. Boston, Mass: Houghton Mifflin,

cop. 1979.

50

Hollnagel, E. (2012). FRAM, the functional resonance analysis method. [electronic resource] :

modelling complex socio-technical systems. Farnham, Surrey, UK England; Burlington, VT:

Ashgate, c2012.

Kaptelinin, V., & Nardi, B. (2012). Affordances in HCI: toward a mediated action perspective. In

Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (pp. 967-976).

ACM.

Norman, D. A. (2013). The design of everyday things. New York, NY: Basic Books, 2013.

Preece, J., Rogers, Y., & Sharp, H. (2016). Interaktionsdesign: bortom människa-dator-interaktion.

Lund: Studentlitteratur, 2016

Saab AB (2016) Griffin Global Support – High availability greater operational effect. Retrieved

2016-03-23, from http://saab.com/air/support-solutions-and-services/airborne-platform-support-

solutions/Griffin-support/

Sheridan, H., & Reingold, E. M. (2014). Expert vs. novice differences in the detection of relevant information

during a chess game: evidence from eye movements.

Stickdorn, M., & Schneider, J. (2011). This is service design thinking: basics, tools, cases.

Hoboken, N.J. : Wiley, 2011.

Stone, D. L. (2005). User interface design and evaluation. San Francisco, Calif.: Morgan

Kaufmann, cop. 2005.

Spence, R. (2007). Information visualization: design for interaction. Harlow: Pearson/Prentice Hall,

2007.

Tullis, T., & Albert, B. (2013). Measuring the User Experience: collecting, analyzing and

presenting usability metrics, second edition. Waltham: Morgan Kaufmann.

51

Appendix

The tasks for existing MGSS:

1. Find out when the aircraft 3707 did its last cycle?

(define date and time, it must not be a manually created cycle)

1. During aircraft 3407 latest cycle a FM-Alarm occurred with Event No 563, find MG-number and

the binary code for this fault event.

2. Did it occur any FM-Alarm during aircraft cycle with time and date:

2010-08-02 08:14:40?

3. Do an OSIS export.

In the prototype the participants were asked to perform following four isomorphic tasks:

1. Find out when the aircraft 222 did its last cycle?

(define date and time, it must not be a manually created cycle)

2. During aircraft 224 latest cycle two critically fault events occurred, which was the first fault event

that occurred?

(define MG-number and the binary code for the event)

3. Did it occur any fault event during aircraft 222 latest cycle?

4. Do a Fenix export.