Polling with Mobile Devices on University Campus

8/11/2019 Polling with Mobile Devices on University Campus

1/19

Polling with Mobile Devices on University Campus

Denis Andrasec, BSc

Assoc.Prof. Dipl.-Ing. Dr.techn. Martin EbnerDipl.-Ing. Thomas Ebner

Institute for Information Systems and Computer Media (IICM),Graz University of Technology

A-8010 Graz, Austria

Evolaris Next Level Gmbh,A-8010 Graz, Austria

November 2013

Abstract

Mobile devices have become ubiquitous in our daily lives, and they can be used in novel ways to improve orenhance existing use cases and offer new solutions for unsolved problems. One such use case is performingstudies, questionnaires and polls. With their unique capabilities and sensors, mobile devices nowadays cancontribute to those elds and even offer new ways of data analysis and methodologies.

The combination of technologies in just one device, namely the touchscreen, sensors, web-connectivityand build in GPS tracking, make them a perfect tool for different scenarios and contexts. Some of those, inparticular touch-screen technology, can be found in the literature about computer assisted questionnaires andinterviews, where the technology was found feasible and in a large number of cases even better accepted as

classical methodologies like paper-and-pen questionnaires.Drawing from those experiences, and with current trends in mobile development in mind, a prototype

application for conducting interviews and questionnaires is developed. The main research question is, howthe conjunction of automatically tracked geolocation data and questionnaires can be used in a scenario thatexamines quality of life on campuses across the university. By creating a heat map of location and questionnaireanswers, dependencies can be visualized, making them easily detectable. It can be concluded that with the helpof mobile devices questionnaires can be enhanced in an implicit way to give a deeper insight in dependenciesaccording to the local places the questionnaires are happening (in our special case on university campus).

Draft version - originally published in: Andrasec, D., Ebner, M., Ebner, T. (2014) Polling with Mobile Devices on University Campus.Stavros, A.V. (ed.). Advances in Communications and Media Research. Nova Publisher, pp. 99-118


2/19

Contents

1 Introduction 21.1 Interviews, Questionnaires and Polling with Technologies of Mobile Devices . . . . . . . . . 31.2 Qualitative vs Quantitative Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.3 Tablet Computer Assisted Personal Interviewing (TCAPI) . . . . . . . . . . . . . . . . . . . . 5

2 Method 62.1 Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.1.1 Requirement 1: Useful for both qualitative and quantitative methods. . . . . . . . . . 62.1.2 Requirement 2: Support a wide range of question types. . . . . . . . . . . . . . . . . 62.1.3 Requirement 3: Perform interviews and surveys ofine. . . . . . . . . . . . . . . . . . 62.1.4 Requirement 4: Use special capabilities that only mobile devices support. . . . . . . . 62.1.5 Requirement 5: Focus on ease of use. . . . . . . . . . . . . . . . . . . . . . . . . . . 72.1.6 Requirement 6: Keep close to the paradigms of the operating system. . . . . . . . . . 7

2.2 Ecosystem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.3 Development Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.4 Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.4.1 Import Questionnaires . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.4.2 Start, Abort, Continue Survey or Interview . . . . . . . . . . . . . . . . . . . . . . . 82.4.3 Questionnaires and Results on Device . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4.4 Many Question Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4.5 Record and Play Audio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4.6 Export Results in CSV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4.7 Import and Export of Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4.8 Question Status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.4.9 Full Screen Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102.4.10 Location Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

2.5 Database . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102.6 User Interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.6.1 Home . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.6.2 Questionnaire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.6.3 Interview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112.6.4 Full Screen Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132.6.5 Export . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3 Discussion 14

4 Conclusion 16

Bibliography 17

1


3/19

Chapter 1

Introduction

Mobile devices have become ubiquitous in our daily lives. They come in the form of smartphones like theGoogle Nexus 5, tablets like the Apple iPad, and similar devices from various hardware manufacturers. Thesedevices have a modern mobile Operating System (OS), with Android from Google, iOS from Apple, Windows8 and Windows Phone 8 from Microsoft beeing the most popular representatives today [ Worldwide SmartphoneSales to End Users by Operating System in 2Q13 2013 ; Worldwide Devices Shipments by Operating System2013 ]. Others are Symbian from Nokia and BlackberryOS from RIM.

All those devices share certain characteristics, which sets them apart from other mobile devices like hand-helds and Personal Digital Assistant (PDA) devices, and those are their operating systems, the touchscreentechnology, the form factor and extended location capabilities. While other devices may offer some of thosetechnologies, the combination of them is what makes modern mobile devices outstanding, and their successwith consumers is a testimony to this.

Touchscreen technology, where a user can manipulate software with the touch of his ngers, has improved

to the point where multiple points and touches are registered. Called multi-touch, it is part of most modernsmartphones and tablets. It enables users to manipulate content on their devices through gestures like swipes,taps, double-taps, pinch-to-zoom and press and hold. These are paradigms which are very different from thosefound on desktop computers where the mouse and the keyboard are the main input devices. The dynamicsof user interaction become fundamentally different. Therefore, interaction models for software need to beadapted when design applications for mobile devices [ Ebner, Stickel, and Kolbitsch, 2010 ]. There are morepossibilities, but also numerous drawbacks. For example, touch-inputs do not have the precision of a pointingdevice, making it necessary to create larger touch targets which are suited for human ngers. Balancing thisneeds a challenge with the constrained screen size often found on mobile devices, due to their smaller formfactor.

Furthermore, mobile devices are, as their name suggests, very portable. The screen sizes are typically

between 3.5 and 5 inches for smartphones, and 7 to 10 inches for tablets computers. Both device classesoffer long battery life and small weight. Computing power, albeit constantly rising, is still constrained whencompared against their desktop and Personal Computer (PC) counterparts, which limits their possible usagescenarios if computational capabilities are in question. Also, using larger computing resources negativelyaffects battery life, making it essential to constrain Central Processing Unit (CPU) usage if this is a concern.When designing a User Interface (UI) for a device with a small screen, it is not enough to shrink everythingdown. A different approach is needed, or as Dawson et al. [ 2012 ] puts it, "Boil the application down to the mostcritical functions and content, and then lay them out strategically in the available screen space. For example,action buttons should go in the lower third of the screen, where they are most easily tappable."

Mobile devices offer extensive mechanisms to determine location, which is the key component for offeringLocation Based Services (LBS). Different technologies can be used to determine the location of a device. For

example, most smartphones and a rising number of tablets offer Global Positioning System (GPS) capabilities.This is possible if the device receives radio signals from at least three GPS satellites, calculating the positionwith signal timings, strength and triangulation. Another approach is triangulation with cell towers or wireless

2


4/19

1.1. INTERVIEWS, QUESTIONNAIRES AND POLLING WITH TECHNOLOGIES OF MOBILE DEVICES 3

networks. Combining the aforementioned technologies can offer more accuracy [ Vaughan-Nichols, 2009-02 ;Dutzler, Ebner, and Brandner, 2013 ]. In conclusion, if a device offers one of these wireless technologies, it canestimate its current position rather accurately, making LBS possible. Examples of these services are mapping,directions, nearby recommendations, and social networks. This feature of mobile devices is seen as the one

with potential for innovative approaches for usage, and [ Vaughan-Nichols, 2009-02 ] concludes that demand forLBS will strongly increase if the technology is adapted wide enough and good business models are found.

LBS can also change users behavior in their neighborhoods. For example, Foursquare lets users check-in atpoints of interest, with the incentive to collect points and get coupons and deals [ Foursquare 2013 ]. While thisis leading people to explore more around them, this type of "volunteering" the own location gives private orga-nizations extensive information about their users, essentially enabling them to create detailed proles which canlead them to anticipate user behavior [ Wilson, 2012 ]. While essentially all social networks collect informationabout their users, organizations can gain even more detailed information when LBS are activated.

1.1 Interviews, Questionnaires and Polling with Technologies of Mo-

bile DevicesWith the capabilities of mobile devices in mind, interviews, questionnaires and polling is examined in thissection. According to Caviglia-Harris et al. [ 2012 ] successful approaches where Computer Assisted PersonalInterviewing (CAPI) was used date back to the 1980s, and they are increasingly becoming standard in inter-view situations. However, they found the following concerns when using such approaches, "Potential concernsexamined in the literature include technical feasibility, respondent acceptance, interviewer acceptance, effectson data quality, and cost." In Computer Assisted Self-administered Interviewing (CASI) approaches, feasibilityand respondent acceptance was already found to be met, and this can also increasingly be observed in interviewsituations. One of the most prominent requirement for feasibility in interviews is weight and portability. In-terviewers have to be able to easily transport the computer equipment and, depending on the situation, shouldeasily use the device in sitting or standing positions [ DeBree et al., 2008 ; Fann et al., 2009 ; Larsson, 2006 ].

In their research, Caviglia-Harris et al. [ 2012 ] were focused on testing the feasibility of CAPI methodsin remote areas with the help of ruggedized laptop computers with touch-screens and GPS capabilities. Indeveloping countries, face-to-face interviews are used to gather information about population and developmentdynamics, and traditionally those things are done with Paper Assisted Personal Interviewing (PAPI) approaches.With the computerized approach, several advantages could be observed:

Data Quality The quality of the data was increased as lower error rates were reported because computer ques-tionnaires offer possibilities for skips, required questions, and data consistency checks. For example,elds that only allow numerical inputs within predened ranges could be implemented.

Reduced Travel Time With the help of GPS equipped devices, right locations in remote areas were easier to

nd, therefore, strongly reducing travel time. Prior to this, directions had to be found from neighbors.Greatly Reduced Data Entry The times needed for data entry were greatly reduced.

Easy Meta Data Collection With a CAPI approach, it is easy to collect metadata like interview time, durationand location.

However, Caviglia-Harris et al. [ 2012 ] noted that there were also signicant difculties because of the us-ability of computer systems, noting that programming or changing the questionnaires needed deep knowledgeabout the system and also programming skills, calling for trained personnel on site to perform those tasks if short term changes to questions were needed. The design of CAPI methods seems to be the most inuentialfactor when comparing the collected data to the one obtained with other modes. For example, Fuchs, Couper,

and SE [ 2000 ] found that interview duration is the same with PAPI and Computer Assisted Telephone Inter-viewing (CATI) approaches when questions are structured the same way. Concluding, that where computerassisted techniques are slower, this can be attributed to questionnaire design, rather than to technology effects.


5/19

4 CHAPTER 1. INTRODUCTION

When performing questionnaires on touch-screen devices, feasibility and comparability to other data col-lection modes has to be determined. If those two are met, all the advantages of computer administered methodscome into effect. Those are almost error free data, cost savings, and immediate availability of the gathered datafor analysis.

Several studies have been conducted on the feasibility of touch-screen computers for lling out question-naires and forms [ DeBree et al., 2008 ; Fann et al., 2009 ; Larsson, 2006 ]. A key point of examination is if the gathered data is comparable to classic methodologies like pen-and-paper questionnaires. Of course, it isdifcult to show the feasibility of those new approaches in general because of the context where they are used,and exposure to the technology of participating people has to be taken into consideration.

One eld of interest is the improvement of Health Related Quality of Life (HRQOL). The use of touchscreen computers in such an environment was examined by DeBree et al. [2008 ]. The goal is to captureinformation about health of patients over a longer period of time. Classic approaches of data collection, likepen-and-paper, are difcult in such a setting because of missing resources for questionnaire distribution, datacollection and analysis, or put more bluntly, it is a matter of cost. To tackle this problem, an approach wherepatients autonomously answered questions on a touch-screen computer on their visits to the hospital was tested.Most patients found the system easy to use, and compliance to ll out questionnaires on later visits was 67%.This seems low, but it could be increased by training the hospital personal accordingly. The collected data wasavailable in real-time for analysis through a database.

Similarly, Fann et al. [ 2009 ] tried to show the feasibility of touch-screen equipped computers for depressionscreening of cancer patients. As a questionnaire, the Patient Health Questionnaire 9, or PHQ-9 was selected.It is very short with only nine items and is especially useful for detecting changes and treatments effects.Therefore, it can be used for longitudinal screenings, which take place over a long period of time, and requirea screening method that is both reliable and not cost intensive. Patients lled out the questionnaire in a waitingarea. While older patients needed longer, which is attributable to the lack of experience with computers, thecomputerized method of data collection was found to be feasible in this environment.

Using only a touch-screen computer in such a difcult setting was shown to be feasible. The next stepis showing that this method is also comparable to well known and tested methods, which was examined byLarsson [ 2006 ]. He compared a touch-screen computer assisted method to the classical form of data collectionthrough pen-and-paper, also in a clinical setting. Patients were asked to rate the quality of care they were givenduring cancer treatment. One half of patients performed the questionnaire with the touch-screen, the other half used paper. It was shown that the majority of ratings were comparable between the two modes. Generally,patients seemed to prefer the touch-screen version. The quality of data was excellent as no errors in the datawere reported. The questionnaire was designed in a way that all questions have to be answered. It is notpossible to move on with a question if it is not answered, but it is possible to go back and change answers.This has the effect that no data was missing, in contrast to the paper questionnaire where 14% of patients didnot provide one or more answers, even though they were instructed to answer all questions. Also, questionswere presented individually on the screen. One would assume that older people would use the computer moredifcult to use, but in this study elderly patients also preferred the touch-screen. The computer version of thequestionnaire produced better results, was easier to use, and preferred by the patients.

In another setting, Denny et al. [ 2008 ] found that usage of hand-held touchscreen devices for data collectionwas feasible in schools. Students had to ll out questions on touchscreen hand-held devices and laptops. Thelargest part of the students had no preference between the two data collection modes. However, the ones thathad one, strongly preferred the hand-held devices. The reason was ease of use and larger perceived privacy.

The literature showed two instances where mobile devices were used for surveys. The rst is from Klem-merer [ 2011 ], comparing a survey on smartphones to web and paper equivalents by building a prototype appli-cation for the Android platform. Participants lled out surveys in the three different modes, and the results werecompared to each other. The preferred mode was the web survey, followed after mobile and then paper. How-

ever, the mobile method was perceived as the most "fun" to use. Furthermore, participants needed more timefor the completion of the mobile survey compared to others, and this was, in part, attributed to the prototypalcharacter of the application with long loading times and several errors, as well as the small screen size.


6/19

1.2. QUALITATIVE VS QUANTITATIVE METHODS 5

Similar results were found by de Bruijne Arnaud and Wijnant [ 2013 ], also comparing the mode effect of surveys. A questionnaire was developed for mobile devices, with unique design considerations to account forscreen size and the touch screen interface. While the results of surveys performed with the mobile deviceswere comparable to the ones from web surveys, the completion time was found to be longer than on their larger

counterparts.

1.2 Qualitative vs Quantitative Methods

Qualitative and quantitative methods are both an integral part of empirical social sciences today. They are notexclusively used and can be combined, differing in the researchers role and level of standardized questions.

In qualitative approaches, the researcher asks questions, usually in the form of a face-to-face interview.Those are very open, and subjective impressions of the researcher are desired to gain more insights. Theresearcher asks people in a certain context, with the goal to gain insight by letting them answer very freely.Therefore, this form of approach is used when there is not much available information about a eld of researchand theories or hypothesis have to be build at rst, making it an explorative approach. A xed hypothesis prior toqualitative interviews is not seen as desirable because the interviewer would take the research in a predetermineddirection, limiting the extractable knowledge. Qualitative research was usually seen as a preliminary stage of quantitative research, but Atteslander [ 2010 , Page 2425] argues that qualitative research does provide insightson its own, although they are different from those in representative studies.

Contrary to qualitative approaches, subjective observations of the researcher are undesired in quantitativeresearch. The data collection is very standardized, and collected data is statistically analyzed and compared.This approach is about quantifying information and variables. The order of questions is in large parts pre-determined, and it is desirable to have identical circumstances in every situation where the data is gathered.Standardized approaches are usually used if there already exists a strong hypothesis or theory and they serve toprove or disprove it. [ Qualitative Forschung - Ein Handbuch , Pages 2425]

1.3 Tablet Computer Assisted Personal Interviewing (TCAPI)

With the research goal in mind, the question remains which new methodologies and benets can be offeredby mobile devices. To examine this question, a prototype application for Tablet Computer Assisted PersonalInterviewing (TCAPI) is build, offering useful features for qualitative and quantitative methods. This paperexamines the development process of the application, showing how features of native applications for inter-viewing, surveys, questionnaires and polling can be implemented. Furthermore, the overall research questionof this work 1 is: "How was the mobile application for qualitative and quantitative approaches build and can itoffer new possibilities for campus surveys?"

1The paper was created using the survey template of Andrews [2012 ].


7/19

Chapter 2

Method

This chapter is devoted to the development of the Tablet Computer Assisted Personal Interviewing (TCAPI)prototype application. Current trends in mobile design and the special capabilities of mobile devices were takeninto consideration over the whole development process. In Section 2.1, requirements for the prototype arepresented. section 2.2 presents the ecosystem and section 2.3 the details of the development process. Followingthis, the implemented features that support the requirements are described and discussed in section 2.4. Thedatabase implementation is described in section 2.5 . Lastly, the UI of the prototype application is described insection 2.6 .

2.1 Requirements

2.1.1 Requirement 1: Useful for both qualitative and quantitative methods.

The initial project plan envisioned TCAPI as a tool to assist interviewers and conduct partly structured inter-views. One integral part of this strategy is to foster mixed method approaches by the use of new technologiesto create new opportunities for method innovation. This makes it crucial to envision a system that can also beused for quantitative methods. Hence, making it a requirement to support both methods and their combination.

2.1.2 Requirement 2: Support a wide range of question types.

The more question types an application or tool offers, the stronger are the benets for standardized methodsas more complex questionnaire designs are possible. Such a tool gets more useful if researchers can choosebetween a wide variety of possibilities. Mobile devices have some capabilities which are stronger than in othercomputer products, like their weight and portability, and some that are weaker, like text input and screen size[Findlater2011 ]. Those need to be taken into consideration when implementing different question types.

2.1.3 Requirement 3: Perform interviews and surveys ofine.

Even though mobile devices have extensive capabilities for wireless communication, there are times when auser is not able to connect to the internet. For that purpose, the application should be able to operate completelywithout an internet connection. Communication capabilities should only be seen as a natural extension, notas the only means of performing interviews and surveys, or collecting and analyzing data. This requirementalso incorporates to keep as much functionality as possible on the device. For a tool to be useful anywhere andanytime, users should be able to do as much of the offered functions on the device itself.

2.1.4 Requirement 4: Use special capabilities that only mobile devices support.

Fully using the capabilities that mobile devices is an essential requirement for this work, and therefore also astrong requirement for the development process. A native approach must be chosen for development, because

6


8/19

2.2. ECOSYSTEM 7

of access to the hardware Application Programming Interface (API) that is provided by the OS. Most of thoseare, for now, only available for native applications. The downside to this approach is that a native applicationcannot be run on other platforms, which limits its audience. However, there are hybrid approaches that can beused to create applications for multiple operating systems that still have access to hardware capabilities.

2.1.5 Requirement 5: Focus on ease of use.

Focusing on ease of use means that a sophisticated User EXperience (UX) has the same priority as the features.Adding the latter should never be at the expense of the former. However, sophistication does not mean complexin the context of UX, but rather it means making complex systems easy to use, and there are many ways toachieve this. For example, the rst law of usability as dened by Krug [2006 ] is "Dont make me think!",which calls for software products that are simple to use and instantaneously understandable. While he focuseson web pages in his book, the same principle is applicable to any UI. More specically, he denes his rstprinciple as, "It means that as far as humanly possible, when I look at a Web page it should be self-evident.Obvious. Self-explanatory. I should be able to get it-what it is and how to use it-without expending any effortthinking about it." Furthermore, [ Krug, 2006 , Page 34] positively emphasizes the use of conventions as theyare developed over time, and universally understood by users. Staying close to tested conventions makes usingsoftware products easier to users, opposed to reinventing the wheel every time. Mobile operating systems havedeveloped their own conventions, constituting the next requirement.

2.1.6 Requirement 6: Keep close to the paradigms of the operating system.

Mobile operating systems have unique ways for navigation between content and interaction with certain UIelements. Those conventions are unique to the respective OS. When developing an application for it, theyshould be used in most instances. The reason being, that users are used to them if they are familiar with theplatform. Hence, even if they are new to a platform, those interface paradigms are a useful guideline for craftinguser interfaces, and should therefore not be ignored [Ebner, Stickel, and Kolbitsch, 2010 ]. The software vendorsof both Android and iOS operating systems have created guidelines for their platforms. For Apples iOS, thereare the Human Interface Guidelines (HIG) [ Human Interface Guidelines 2013 ]. Goole offers a similar guidelinefor their operating system on the developer homepage [ Android Design 2013 ].

2.2 Ecosystem

To shorten development time, two open source software products were used. The rst is queXML, which is aXML schema that denes questionnaires [ queXML: Open Source XML Questionnaire Schema 2013 ]. The sec-ond is LimeSurvey, which is a software tool for creating, analyzing, and exporting surveys [ LimeSurvey: OpenSource Survey Creation and Distribution 2013 ]. Although it would be greatly useful to create surveys directlyon a mobile device, it was not deemed feasible for a prototype application. So the creation of questionnaireshad to come from an external source, and LimeSurvey is chosen for this task. The ability to create complexquestionnaires and export them in queXML, which is easy to understand, makes it a tting solution for theproblem at hand. The whole ecosystem, together with the export features of the TCAPI prototype application,can be seen in gure 2.1. Questionnaires are created with LimeSurvey and exported as queXML les. Theyare then imported in the application, either through a tethered connection, or through a Dropbox download[ Dropbox 2013 ]. Survey and interview results are exported in CSV les, and the audio recordings are exportedin the .CAF format [The Comma Separated Value (CSV) File Format 2013 ]. Both can be exported through localstorage and a tethered connection or Dropbox [ Dropbox 2013 ].

2.3 Development Process

During prototype development, two approaches, which go hand in hand, have been incorporated to cope withthe difculties of software development. One is an agile development process, where software is continuously


9/19

8 CHAPTER 2. METHOD

Figure 2.1: The ecosystem that was selected for the prototype.

iterated upon in short circles, and the other is Test Driven Development (TDD) [Shore and Warden, 2009 ].In the latter, tests are written rst, and only then comes implementation code. By following both, softwaredevelopment becomes more dynamic, in the sense that unpredicted changes are easier react upon, fosteredby the short iterations and the security that tests provide. Especially in prototype development, where largechanges are the norm, these approaches are indispensable [ Shore and Warden, 2009 ; Martin, 2009 ].

2.4 Features

2.4.1 Import Questionnaires

The central aspect of the whole application is the survey or questionnaire. This requires a way to describe such aquestionnaire. At rst, using EXtensible Markup Language (XML) for the document description is considered.XML is in a human readable form, making it easy to understand. However, creating a description of the wholedocument structure for questionnaires with sections, titles, comments, questions, as well as numerous questiontypes, seemed unreasonable for the scope of this project, especially in contrast to all the other features that werealso relevant. Even if a new data description is created, the need to create questionnaires would still be there.

Therefore, existing approaches needed to be considered. For this purpose, LimeSurvey is examined. Notonly is it possible to create questionnaires with over thirty question types, they can also be exported in twole formats. One of which is queXMl and the other is the survey structure le format (.lss) which is used byLimeSurvey internally. The difference between those two is that the survey structure format supports all of theLimeSurvey features and can also be used to import surveys into LimeSurvey while queXML does not supportas many features and question types. Nonetheless, queXML is much easier to understand and a comprehensivedocumentation is available, making it the preferred format for the prototype application.

Even with this approach, this is still a complex feature because the questionnaire schema needs to be

validated, mapped to the applications internal database and rendered correctly on the screen. However, bychoosing open source software, the ramp-up for development is much faster than developing custom solutionsand therefore, many other features can be considered.

2.4.2 Start, Abort, Continue Survey or Interview

After a questionnaire is imported, it should be possible to perform a survey or interview with the device. Itshould not only be possible to start and end surveys and interviews, but also abort and continue them at anytime. To incorporate this, the interviews that are saved in the internal database have to have a status. With thisapproach, they can be listed by this status. Open interviews at the top, and already closed interviews or surveys

beneath it. Furthermore, the user can freely navigate within the application as well switch to other ones withoutloosing state. This can be used, for example, to show participants a website during surveys. Also, continuingan already started interview is possible at any time.


10/19

2.4. FEATURES 9

2.4.3 Questionnaires and Results on Device

Since one requirement of the application is to perform interviews and surveys ofine, the data has to be savedlocally. Building on this premise, the only thing left to do, is to design a user interface which lists interviewsand the questionnaire. Also, it should be possible to view the questionnaire without starting an interview.

2.4.4 Many Question Types

Important for the standardized part of the application, is a wide variety of question types. This is commonamongst applications for surveys and forms. Prominent examples are iSurvey [ DataField 2013 ] and [Fluid-Surveys 2013 ]. The schema queXML also supports a rather large selection of question types, making it feasiblefor a prototype application. However, the smaller screen size in comparison to web surveys has to be consideredwhen implementing these questions, and the interface elements also need to be kept consistent with the overallinterface of the selected operating system, as described in the requirements.

2.4.5 Record and Play AudioIn face-to-face interviews, conversations are often recorded for later transcription. This fact makes audio record-ing one of the most important features for qualitative approaches. The central thing to consider is that the con-versation is clearly understandable. Also, the recording has to continue in the background, even if the screenof the device is locked, or if the interviewer or participant switches to other applications. Another thing toconsider is when audio or video is played inside the application that the recording remains of such a qualitythat the conversation is still understandable.

2.4.6 Export Results in CSV

One of the major points of the project concept, is the export of data through CSV. This feature has a high priority

and is also straightforward in its implementation. It offers users the ability to analyze their data independent of the application and operating system.

2.4.7 Import and Export of Files

When considering how data is copied to and from mobile devices, their unique capabilities need to be accountedfor. They offer, as their name suggests, portability and also comprehensive communication capabilities throughwireless connections. These were used for le exchange in the prototype application. Since only les needto be exchanged, and no API to a database server is needed, existing cloud services can be used for this task.Dropbox looked like a tting candidate for the task, due to its popularity as described by Hunsinger and Corley[2012 ]. Also, it offers a developer facing API so that the service can be integrated into third party applications.

In addition to a cloud service for le exchanges through the internet, a tethered connection with a PC has to beoffered in the case that there is no internet connection.

2.4.8 Question Status

Catering to the requirement that the application needs to be usable as an assistive tool for open interviewsas well as for standardized surveys, the answer status of questions and sections should be indicated visually.The reason being that, in weakly structured interview situations, interviewers tend to jump between questionsand sections. Therefore, they need a visual indicator for answered and unanswered questions and sections.Since questionnaires can also be divided into sections, those sections should also indicate if all of the containedquestions were answered, or only a subset of them. For this purpose, the metaphor with read and unread

statuses that are used in email applications was adapted accordingly. Unanswered questions and sections shouldbe indicated through a circle, partly answered sections through a half circle, and fully answered sections andquestions with no circle.


11/19


12/19

2.6. USER INTERFACE 11

Figure 2.2: A part of the mapping of queXML to the CoreData database.

2.6 User Interface

In this section, the UI of the TCAPI prototype application is described. Guidelines for the iOS operating systemwere used and the team communicated the creation process of the UI through mockups.

2.6.1 Home

When the application starts users are greeted with the home menu. The interface is kept simple, and all theimported questionnaires are listed as can be seen in gure 2.3a . Questionnaires can be imported by pressing

the "+" button in the upper right corner, which reveals a popover menu (Figure 2.3b ). Users can select eitherlocal les that are copied to the application with a tethered connection, or by importing them through Dropbox[ Dropbox 2013 ; Dropbox 2013 ]. In order to load les from the latter, a user has to link the application tothe service by tapping it under the "Services" label. After granting permissions, browsing through the onlinelesystem is possible.

For easier navigation, only folders and XML les are listed. When tapping a XML le, the questionnairegets imported if its schema matches the queXML schema. Otherwise, a warning is displayed. No false datacan be imported as every import is validated. By selecting a questionnaire from the list, the user is taken to thenext screen which shows the associated data.

2.6.2 Questionnaire

All previous surveys and interviews are listed in the questionnaire screen. They are separated in active inter-views and saved interviews as can be seen in gure 2.3c . Directly under the navigation bar are two selectableoptions. One being "Questionnaire", which shows the whole questionnaire without starting an interview, andthe other being "New Interview", which, as the name suggests, creates a new active interview. In the top right isa "Share" button, from which the data export menu for this questionnaire and its associated data entries can bereached. Tapping an interview navigates to its content and the current GPS location is persisted to the databasefor newly created entries.

2.6.3 Interview

The interview screen is busier than the previous ones, offering a large variety of options. The list of sections,which are part of the questionnaire, are located at the left side of the interface. Selecting one of those itemsshows the contained questions, as can be seen in gure 2.3d .


13/19

12 CHAPTER 2. METHOD

(a) The home menu of the TCAPI prototype. (b) The menu where questionnaires can be imported.

(c) Questionnaire menu with active and saved inter-views. Also, the option to show the questionnaireand sharing data is possible from here.

(d) The interview screen with sections, questions, au-dio, FSM and question answer status indicators.

(e) The Full Screen Mode (FSM) with one question at atime.

(f) Exporting CSV data and audio recordings to Drop-box.

Figure 2.3: The UI of the TCAPI prototype. (a) shows the home screen, (b) the overlay with importoptions, (c) the questionnaire with associated interviews and (d) shows the interview screen.gure (e) shows the Full Screen Mode and (f) the export options.


14/19

2.6. USER INTERFACE 13

Questions consist of the question text, additional informal text, and UI elements for answering. In 2.3d ,there are only single choice selections. The selected item is clearly highlighted, and every answer can alsobe deselected, giving users full control over the answer status. This status is indicated by a blue circle forunanswered question, a half-lled circle for answered questions, and no circle if the question was answered.

This is not only implemented for single questions, but also for questions with multiple sub-questions and alsofor sections, giving the interviewer or self administered user a quick glance over how much of the questionnairewas accomplished.

At the top right (navigation bar), is a button to save and continue interviews. If this interview will be saved,changing the answers in questions is disabled. This status is indicated by graying out the UI elements thatdo not offer any interaction anymore. This is also the mode in which already saved interviews or surveys areshown. With this approach, users get presented all the selected answers, and all the other answer options aswell. Furthermore, audio can be played for saved interviews. Users can leave the application at any time andconveniently get back to it without loosing data, as any change to any question is instantly persisted in theinternal database. With this approach, users can switch to any application they want to and never lose progressor data.

By tapping the "Audio" button on the top right, a popover is displayed. This popover shows differentoptions according to the interview status. For an active interview, recording can be started and paused, andfor a saved interview, the recorded audio can be played, if any is available. Like saving in the background,audio recording continues if the user leaves the application. Even if the device is locked, the recording is notsuspended. Audio is saved in the CAF le format. This is natively supported through hardware decoding andencoding, saving as much battery life as possible. Also, the standard prole that was chosen is congured forbest clarity of voice recordings in conversations.

Finally, the button titled "FSM" activates the Full Screen Mode.

2.6.4 Full Screen Mode

In the FSM, only one question at a time is displayed to the user. This mode is supposed to be used in situationswhere the interviewer hands over the tablet to the interviewee, enabling autonomous answering to standardizedquestions. When pressing the "FSM" buttons, all the questions of a section are added, giving an interviewer theability to structure and separate interviews accordingly.

Question navigation is possible with "Next" and "Previous" buttons at the bottom. Also, the current questionnumber with the number of all questions next to it is displayed as seen in gure 2.3e . In addition to pressing thebuttons for navigation, user can swipe left or right to change questions. If the navigation reaches an end pointin either direction, the buttons get disabled and greyed out accordingly. Users can leave the FSM any time bypressing the "Done" button in the top right corner.

Other features like automatic question proceeding and having to provide an answer for proceeding have notbeen implemented for the prototype. Mainly because a interview is seen as a situation where both parties work

together, reducing the need for additional control of the answer process. However, those features are relevantfor self administered questionnaires as they increase data quality [DeBree et al., 2008 ; Fann et al., 2009 ].

2.6.5 Export

Interviews can share their results and audio through the questionnaire screen, like mentioned earlier. Afterpressing the share button, an overlay is displayed where les can be selected for export. This menu can be seenin gure 2.3f . This functionality is for Dropbox export only. Additionally, all the data is saved through localstorage, and can be exported using a tethered connection to a PC.


15/19

Chapter 3

Discussion

The previous chapter showed the development process of the prototype. The result is an application that sup-ports both qualitative and quantitative methods, as well as mixed method approaches. Special capabilitiesof mobile devices were used to offer new possibilities, one of which is implicit data collection through GPSrecordings for every survey or interview.

This approach dramatically reduces the overhead for collecting reliable location data for surveys, offeringa unique usage scenario for surveys on distributed campuses. While not implemented in the prototype, datavisualizations overlaid on a map become possible with little to no effort on the researchers or data collectorspart. For example, campus buildings of the Graz University of Technology are distributed over the whole city.With such a tool, visualizing data of surveys becomes possible, for example with a heat-map approach as seen ingure 3.1 . With a library like HeatmapJS for JavaScript created by Wied [ 2013 ], the technical implementationof such a feature is straight-forward. Building this into the tool would give interviewers the possibility to drawconclusions from the connection of location and survey results, making it useful for surveys regarding thequality of campus life on distributed campuses and other surveys where location data is relevant.

14


16/19

15

Figure 3.1: Data visualization possibility for campus surveys. Created with HeatmapJS [Wied, 2013 ].


17/19

Chapter 4

Conclusion

This paper presented a prototype application for surveys and interviews. As a tool for researchers and surveys inthe eld it supports qualitative as well as quantitative methods. A central part of it is to use special capabilitiesof mobile devices to benet surveys and offer new approaches. The implemented features support this and thefollowing are in the current implementation:

Import and export of questionnaires and results

Start, abort, continue survey or interview

Questionnaires and results on device

Large selection of question types

Record and play audio

Export results in CSV

Full Screen Mode

Location data

Finally, this work prosed a usage scenario where implicitly collected location data can be used to visualizeresults of campus surveys. Data collected over long periods of time from the sensors which mobile devices offercan harvest new research domains. There are opportunities to study mobility, communication and interactionpatterns, which will lead to new insights in computing, social science and education [ Laurila et al., 2012 ].However, to be useful for research, mobile devices and applications have to offer reliable and easily accessible

sensor data.

16


18/19

Bibliography

Andrews, Keith [2012]. Writing a Survey Paper . Graz University of Technology, Austria. Oct. 22, 2012. http://ftp.iicm.edu/pub/keith/survey/ (cited on page 5).

Android Design [2013]. Mar. 28, 2013. http://developer.android.com/design/index.html (citedon page 7).

Atteslander, Peter [2010]. Methoden der Empirischen Sozialforschung . ESV Basics. Erich Schmidt Verlag,2010. ISBN 3503126187 (cited on page 5).

Caviglia-Harris, Jill et al. [2012]. Improving Household Surveys Through Computer-Assisted Data CollectionUse of Touch-Screen Laptops in Challenging Environments. Field Methods 24.1 (2012), pages 7494(cited on page 3).

DataField [2013]. Sept. 17, 2013. https://www.data-field.com (cited on page 9).

Dawson, Alexander et al. [2012]. Essentials of Mobile Design . 1st Edition. Smashing Media GmbH, 2012.ISBN 9783943075427 (cited on page 2).

De Bruijne Arnaud, Marika and Wijnant [2013]. Comparing Survey Results Obtained via Mobile Devices

and Computers: An Experiment With a Mobile Web Survey on a Heterogeneous Group of Mobile DevicesVersus a Computer-Assisted Web Survey. Social Science Computer Review (2013) (cited on page 5).

DeBree, R et al. [2008]. Touch Screen Computer-Assisted Health-Related Quality of Life and Distress DataCollection in Head and Neck Cancer Patients. Clinical Otolaryngology 33.2 (2008), pages 138142 (citedon pages 3, 4, 13).

Denny, Simon J et al. [2008]. Hand-Held Internet Tablets for School-Based Data Collection. BMC Research Notes 1.1 (2008), page 52 (cited on page 4).

Dropbox [2013]. Sept. 22, 2013. http://www.dropbox.com (cited on pages 7, 11 ).

Dropbox [2013]. Sept. 22, 2013. http://www.apple.com/at/itunes/download/ (cited on page 11 ).

Dutzler, Roland, Martin Ebner, and Robert Brandner [2013]. Indoor Navigation by WLAN Location Fingerprinting-Reducing Training-Efforts with Interpolated Radio Map. In: UBICOMM 2013, The Seventh InternationalConference on Mobile Ubiquitous Computing, Systems, Services and Technologies . 2013, pages 16 (citedon page 3).

Ebner, Martin, Christian Stickel, and Josef Kolbitsch [2010]. iPhone/iPad Human Interface Design. In: HCI in Work and Learning, Life and Leisure . Springer, 2010, pages 489492 (cited on pages 2, 7).

Fann, Jesse R et al. [2009]. Depression Screening Using the Patient Health Questionnaire-9 Administered ona Touch Screen Computer. Psycho-Oncology 18.1 (2009), pages 1422 (cited on pages 3, 4, 13).

Flick, U, E von Kardorff, and I Steinke. Qualitative Forschung - Ein Handbuch . 7th Edition. Hamburg, Ger-

many: Rowohlt Taschenbuch Verlag. ISBN 3499556289 (cited on page 5).

FluidSurveys [2013]. Sept. 17, 2013. http://fluidsurveys.com (cited on page 9).

17
http://ftp.iicm.edu/pub/keith/survey/http://ftp.iicm.edu/pub/keith/survey/http://developer.android.com/design/index.htmlhttp://www.amazon.com/exec/obidos/ASIN/3503126187/https://www.data-field.com/http://www.amazon.com/exec/obidos/ASIN/9783943075427/http://www.dropbox.com/http://www.apple.com/at/itunes/download/http://www.amazon.com/exec/obidos/ASIN/3499556289/http://fluidsurveys.com/http://fluidsurveys.com/http://www.amazon.com/exec/obidos/ASIN/3499556289/http://www.apple.com/at/itunes/download/http://www.dropbox.com/http://www.amazon.com/exec/obidos/ASIN/9783943075427/https://www.data-field.com/http://www.amazon.com/exec/obidos/ASIN/3503126187/http://developer.android.com/design/index.htmlhttp://ftp.iicm.edu/pub/keith/survey/http://ftp.iicm.edu/pub/keith/survey/


19/19

Documents

Polling with Mobile Devices on University Campus