A Personal Informatics Tool To Assist Digital Job Search

MS

C

I

NFORMATION

S

TUDIES

M

ASTER

T

HESIS

A Personal Informatics Tool

To Assist Digital Job Search

by

A

LESSANDRO

L

ATELLA

June 30, 2017

Supervisor:

Prof. D. B

Assessor:

Dr. F.M. N

A Personal Informatics Tool

To Assist Digital Job Search

Alessandro Latella

MSc Information Studies - University of Amsterdam

Amsterdam, The Netherlands

alessandro.latella@student.uva.nl

ABSTRACT

Considering and isolating digital job search as a specific digi-tal activity allows us to consider it as any other human activity in which the organisation of data and self-reporting is central. In this paper we formulate, build and evaluate a new Personal Informatics (PI) tool, meant to facilitate the self-organization, self-efficiency and motivational aspects of digital job search activities. More specifically, we allow for user (job-seeker) interactivity with the proposed prototype through the combi-nation of (a) a timeline-based view in which tasks and goals related to the recruitment process can be set, from a job-seeker perspective; (b) an advanced supporting interface inspired by project management tools with several features aimed at re-ducing workload in seeking and storing relevant information regarding vacancies, companies, and the process in itself; (c) an in-built job search browser frame rendering live job oppor-tunities from the web, and (d) a geographical search tool meant to increase the perception of existing opportunities. To justify our solution, we show the results of a survey made amongst academic students (n=29), which highlights the most impor-tant requirements in the organisation of tasks and activities aimed at getting employment through on-line sources.

Author Keywords

Personal Informatics, Digital Job Search

INTRODUCTION

Job seeking is a complex human activity that requires a range of disparate skills. This activity is usually recognized as being influenced by different factors linked to specific disciplines such as economics, psychology, sociology or cognitive stud-ies, among others [3]. Processes and factors influencing the outcomes of job seeking have been widely studied and repre-sent a vivid area of research. In this paper, we acknowledge the fact that multiple theoretical frameworks1underpinning traditional job search research coexist, as described in the in-troduction of the meta-analysis of job search interventions2 of Liu et al. (2014) [21]. These frameworks don’t adress the

1_{Behavioral learning theory, theory of planned behavior, social}

cog-nitive theory and coping theory are all theories that relate to job search.

2_{Job search interventions are training program designed to help job}

seekers look for employment or secure employment faster [21].

digital aspects of job search directly, but must be taken into account for their contributions on the critical components of this activity. Among the different theories, these components are often seen as: enhancing job search skills, improving self-presentation, boosting self-efficacy, encouraging proactivity, promoting goal-setting, managing stress and enlisting social support. Our paper is to a limited extent to be considered within the narrower theoretical framework of self-regulation theories [6] applied to job search [17], in which job search is seen as a dynamic, recursive self-regulated process [17] where the job seeker receives feedback from his job seeking behaviors and adapts, reduces or intensifies his activities. As Kanfer et al. mentions :

Job search [is] a purposive, volitional pattern of action that begins with the identification and commitment to pursuing an employment goal. The employment goal, in turn, activates search behavior designed to bring about the goal. [17, p. 838]

Usually, to support traditional job search, job search interven-tions have focused on : transmitting job search skills (making phone calls, acquiring necessary knowledge about business directories, job boards, etc.), improving résumés; training job seekers in interviewing techniques; allowing for better self-efficacy and proactivity by widening the variety of positions considered; promoting goal-setting by focusing on specific goals about the desired occupation (job type, salary) or the job search process in itself (amount of phone calls, applications) [21].

On the other hand, almost no research has been devoted to discover the potential of Personal Informatics (PI) tools in as-sisting job seeking activities. Usually adopted in health-related self-quantification, Personal Informatics systems are robustly seen as having a high influence on behavioral change by al-lowing reactivity, which is in turn caused by self-monitoring (the procedure by which individuals record the occurrences of their own target behaviors). Reactivity, or reactivity to self-monitoring, is defined as changes in (the targeted) be-havior rate[22]. In other words, counting or recording the occurrences of a desired or undesired behavior should help increase or decrease the targeted behavior. In our context, which is characterized by an increasingly digitally-driven job search environment and by the proliferation of digital spaces to locate employment, job seekers might fail to assess their self-efficacy, or performance, as they are not provided with

the specific recording and self-organisation tools they need. Alternatively, they might be provided with proprietary solu-tions linked to specific platforms, or they might even build their own ad-hoc analogic or digital workarounds. Another problem consists in the fact that job seekers might not consider job hunting as a "project" or a complex activity, or be aware of any particular tool meant to organize their searches. Lastly, the lack of a uniform and standardized way to collect data sur-rounding digital job search and its processes, does not serve the overarching purpose of understanding how to best advise job seekers confronted with the digital job search environment to maximise the depth and value of their experience3. Our paper is thus divided as follows: (a) we first introduce the common terminology, indicators and name the behaviors at play in traditional job search; in the related work section we also give account of previous work done regarding self-organisation of job search; (b) we provide the rationale of the proposed extension of these indicators to the specific context of digital job search and analyse the results of the survey based on these same indicators; (c) describe the model of our solution, which is based in assessing the results of our survey towards a comprehensive Personal Informatics tool; (d) lay down the functional components of our solution and (e) discuss the implications of our work. In the discussion section, we propose possible critiques and enhancements of the current state of our solution.

RELATED WORK

Two main lines of research inform our work. First, we give an overview of the models and indicators around job search that we want to serve and enhance with our solution (individual differences and job search behaviors). Second, we review the technological framework in which we attempt to encap-sulate these indicators (Personal Informatics). Additionally, we provide a review of existing attempts surrounding the self-organisation and tracking of job search behaviors and discuss if they can be considered PI tools, and their relative efficacy.

Job Search Behaviors, Intensity And Individual Differ-ences Within Sequential Models Of Job Search

Job search has been widely studied and a certain consensus has been established around how to model this activity. A good example is seen in the four-phases, sequential job search and choice model of Soelberg (1967) [26] that has been further operationalized by Blau (1993) [8]. For both Soelberg and Blau, job search is conceptually broken into phases.

Soelberg considers job search as a process that follows these steps:

1. Identifying an ideal occupation 2. Planning job search

3. Carrying out explicit job search and choice

4. Deciding, confirming and committing to the offered position

3_{We will give account of this in the analysis of our survey.}

Figure 1: The structural model of determinants and consequences of job search behaviors [8] in which active job-search behaviors follow preparatory searches. Note that Blau’s research do not include all Saks and Ashfort’s individual differences, and especially Job Search Self-Efficacy.

In his model, Soelberg agreed that these steps could occur si-multaneously, although could not be separated [23]. In reality, Soelberg was more concerned with implementing an individ-ual decision-making model and used job search as a testing field. Nevertheless, his assumptions became very useful in that strand of research as they allowed a useful distinction -highlighted by Blau - between a two-dimensional measure of job search behaviors (preparatory and active).

In Blau’s work, which focused on the job search phases 2 and 3 of the four-phase model, a preparatory job search is defined as the gathering of potential job leads through various sources and can behaviorally measure the effort to gather job search information [8]. An active job search is instead defined as a formal activity, such as sending out resumes to specific prospects, telephoning prospects, and hopefully interviewing with prospective employers [8]. We will see in the next sub-section the reasons why we believe an additional effort was needed in order to allow for a better alignment of those two kinds of job search behaviors within a self-monitoring tool, and why current attemps did not yet completely filled that gap. Another important predictor of employment outcomes, usually mentioned in job search literature is job search intensity. Job search intensity is defined in various ways depending on the authors of reference. Sometimes it is defined as the frequency and scope of different job search activities [28], others as the degree of job search effort by a job seeker [8] or even as a

function of motivation [2]4_{. Perhaps, job search intensity is} better understood by its effects: it is often agreed that an in-crease in search intensity should inin-crease the average number of job offers [9]. More importantly, Gali´c (2011) has stressed the importance of the time-varying nature of job search inten-sity, by demonstrating the dynamic nature of this variable and its importance with regards to its mediating effect with other biographic and financial characteristics. As Gali´c points out, because of the self-regulatory nature of the search process

[...] job search is a behavior whose intensity changes depending on current circumstances of an unemployed individual [13, p. 3].

Examples illustrating this point also concern last-year graduat-ing students, typically increasgraduat-ing their job-search efforts after graduation [3, 7]. We will discuss possible integrations of our solution with existing attempts of visualizing the time-bound student journey, in order to potentially tackle the school-to-work sensible transition period5.

The prominence of job search self-efficacy among individual differences predicting job search behaviors and outcomes

In their seminal paper, Saks and Ashforth (1999) have ex-tended the existing research surrounding job search literature by considering the role of psychological variables (individ-ual differences) within existing sequential job search models. Then, they have analyzed how individual differences affect job search behaviors, which are known to predict employment outcomes [25].

For these authors - at a psychological level - self-esteem, job search self-efficacyand perceived control represent individual variables affecting the aforementioned job search behaviours (preparatory and active). In their results, Saks and Ashforth have discussed the prominence of self-efficacy over other vari-ables. In accordance with Bandura’s theory, and adapted to our context, job search self-efficacy represent "one’s confidence in performing tasks that are important in the job search process6" [6]. In their study, job search self-efficacy was a significant predictor of preparatory and active job search behaviors along with job search intensity. Additionally, perceived control, de-fined as "the perceived control over the job search outcomes", has also been linked to correctly predicting employment status of participants at a later observation stage.

Several authors agree that these individual differences and others variables (such as job insecurity, financial need) have a significant impact on the quality and quantity of job search

4_{It has to be mentioned that a relative important amount of studies are}

related to the motivational aspects related to unemployment benefits and their impact of job search intensity, which is not the focus of our research given the fact we do not focus per se on unemployed individuals. For a review, see Welch (1977) [27].

5_{See Buzzo and Phelps (2016) for their most recent attempt in}

strong-framingthe passage of time in the lived experience of the students’ educational journey [10].

6_{A more general definition is provided by Ramachaudran (1994):}

Perceived self-efficacy is defined as people’s beliefs about their ca-pabilities to produce designated levels of performance that exercise influence over events that affect their lives.[24].

behaviors (see Figure 1), which in turn are good indicators of employment outcomes for the simple fact that they are needed to trigger a job offer.

As said, these concepts became additional parts of existing sequential models of job search; our solution is an attempt to adapt some indicators of these original studies to the specific digital environment, in which most job seekers are supposed to gain employment nowadays.

Personal Informatics As A Self-Monitoring Tool For Task-Specific Goal-Setting For Job Search

Personal Informatics (PI) tools have been evolving exponen-tially, and with them the attempt to better the lives of their users. In our specific context, and to avoid an in-depth review of models and practical applications of PI which is out of the scope of this paper, we will limit ourselves to the contact points that might exist between job search as a playing field and the models of goal-setting through self-monitoring.

Models Of Personal Informatics

Two models so far have been proposed to defend the use of PI: the stage-based model by Li et al. (2010) [20] and the Epstein et al. (2015) model of lived informatics [12]. To be brief, we will only mention in details the stage-based model of Li et al. The stage-based model is composed of five different stages and relative barriers. (1) Preparation : in which people’s motiva-tion to collect personal informamotiva-tion, how they determine what information they will record and how they will record it is as-sessed. Barriers : are usually determined by what information to collect and what collection tool to use. (2) Collection : in which people collect information about themselves. Barriers :usually determined by lack of synchronicity between data collection and availability of the tool, lack of time, finding data, accuracy, motivation. (3) Integration : in which people prepare, combine, and transform information for reflection. Integration can be long or short depending on the nature of the tool. Barriers : are usually determined by different data source or types, from multiple inputs that combine in a cum-bersome way. (4) Reflection : may involve looking at lists of collected personal information or exploring or interacting with information visualizations. Barriers : hindrances at this stage prevent the user from exploring and understanding in-formation about themselves, they can be of different nature (visualization, interpretation, contextual, non-usefulness). (5) Action : in which people choose what they are going to do with their newfound understanding of themselves. Some peo-ple reflect on the information to track their progress towards goals.

A very important classification within these stages is given by the distinction of a stage being user-driven, system-driven or both. In a user-driven stage, the user is the main responsible for collecting data, while the opposite is true for automated (system-driven) stages. Additionally, PI systems can be uni-facetedor multi-faceted, that is, showing only one or more aspects of a monitored human activity.

The Epstein model of lived informatics is a refinement of the previous model that considers the more complex reality of interruptions and resumptions linked to the use of PI tools. Epstein et al. add concepts such as lapsing (forgetting, upkeep-ing, skipping or suspending), but also resumption, migrating between tools, and goal tracking adjustment [12].

The Link Between Personal Informatics and (Job Search) Self-Efficacy

As posited by one of the most-cited models of behavioral change based on insight, such as the Trans-Theoretical Model (TTM) of behavioral change - on which the rationale of PI tools also rely - self-efficacy is a central component. The TTM is based on five stages: pre-contemplation, contemplation, preparation, action, and maintenance7. Self-efficacy supports the preparation phase (Figure 2), in which people are mak-ing specific plans to implement their behavior change. More specifically, PI are supposed to show smaller accomplishments to the user in order to confirm their beliefs of their own com-petences in achieving tasks. Supporting the self-monitoring aspects of generating leads (preparatory) or giving a visual feedback about applying to a vacancy, calling an employer, sending an e-mail (active) is therefore what we could foresee as supporting job search self-efficacy beliefs in job seekers. As we have previously seen in the description of job search mod-els, these PI tools could therefore embody a better alignment between both preparatory and active job search behaviors in the way they are visualized.

Very few authors seems to have explored how Personal Infor-matics tools should be designed to better serve self-efficacy, although Andrew et al. (2011), seem to have produced some design principles in that sense [1]. Some factors that might impact self-efficacy are hinted as follows : 1. Usability, that is - the well-designed aspect and professional aspect of the tool; 2. Alignment between the tool and its user’s goals; and 3. Understanding of the underlying technology.

Additionally, the same authors have pointed out that re-searchers would eventually face the challenge of evaluating a PI tool without being able to run long-term studies (i.e., on evidence of engagement) [1]. In this sense, preferring self-efficacy measurements over evidence of engagement might prove a viable and temporary solution for reasearchers facing publication deadlines or budget issues.

Instances of Job Search Self-Organisation Tools

Some tools exist on the web that are close or related to what a full-fledged PI tool could feature. Often, those features are linked to existing online job-boards, but also as external and indipendent tools (or both, with job-boards integrations). Here is a small review of their characteristics.

Indeed Archiving. The aggregator Indeed, one of the most famous vacancies aggregators worldwide, exhibits some func-tionalities aimed at assisting the self-organisation of digital

7_{These stages mark the evolution of behavioral change, from no}

intention to change the behavior to action and maintenance.

Figure 2: The stages of change according to the TTM: we sustain self-efficacy by tailoring a PI solution in the preparation and active phase of digital job search, as reviewed by Kersten et al. (2017) [18]

Figure 3: Indeed’s "My Jobs" area

job searches. Among them, the My Job area [14], in which a record of visited jobs can be further organised depending on their statuses (Saved - Applied - Interviewing - Offered). The My Job area of one own’s personal account on the job aggregator portal, is also part of the job search browsing expe-rience, because every following visit to an already-visited job description is shown as such (or as the corresponding status). This two-way labelling between the dedicated area and brows-ing locations reduces the aforementioned barriers and frictions in data gathering, as a combination of system-driven (visited job’s data collection) and user-driven (self-monitoring of ap-plications, interviews and deletions status) PI-like collection styles.

Also, Indeed’s My Searches area allows to review past searches, and counts the amount of specific searches linked to used keywords corresponding to searches done in the past. The list automatically re-order itself to give the highest rank to the latest search, and correspondingly updates its counter if a search has already been done.

Jobhero. JobHero has closer resemblences to a native PI tool compared to our previous example, in that sense that the self-monitoring aspects are much more evident [16]. As an ex-ample, JobHero allows to save jobs across the web and track the application progress, set reminders for follow-ups, dead-lines and due dates but also serves the user in documentation management (curriculums, annotations, contact informations, etc.).

In Figure 3, the dedicated Dashboard allows for a visualiza-tion of pending job vacancies and their current self-monitored status: small icons also remind the user of upcoming activi-ties and allow for external calendar integrations. Simply put, JobHero allows for a more detailed organisation of tasks and

Figure 4: JobHero Dashboard with color-coding status labelling and inte-grated due date icons (left from color-coded labels).

activities, showing characteristics that might be closer to the ones that define PI Tools. As Indeed’s features, JobHero’s ones exhibit both user- and system-driven data collection styles, as browsing is facilitated through a web-browser extension which reduces the friction of switching tabs and context (jobboards -PI tool and vice-versa). However, this do not exclude a higher chance of lapsing, as the tool is separated from the vacancies’ browsing experience8.

SURVEYING ACADEMIC STUDENTS ON DIGITAL JOB SEARCH

One of the challenge of a project surrounding the ideation and building of our solution, was the virtual absence of any research focusing on specific aspects surrounding the experi-ence and awareness of the digital job hunting behaviors. For this reason, we have decided to run a survey among academic students. One of the most striking conclusions of our survey is to be seen in the relative absence of awareness shown by participants about self-organisation tools. As we have seen reviewing them, these can be used (a) either to maximise job search behaviors or (b) to enhance the the self-efficacy during processes of recruitment, from a job-seeker perspective. Fur-thermore, the results showed that a relative weak self-efficacy related to digital job search was shown among participants, giving some credit to the idea that a support could be needed.

Graduating University Student Sample and Distribution

We asked a student population of 29 individuals (mean years = 22.5, 51.6 % males and 48.4 % females) to follow a web-based survey, in which several aspects of the aforementioned indicators have been measured. Our population was either looking for or browsing for jobs (61.29 % at the moment of the survey, of which 38.71 % not actively); overall, 61.29 % of respondents said they were expecting to start an employment within max. 5 months, of which 41.94 % within 3 months. A solid 67.74 % of participants were already working, where a 22.58 % was not, and 74.19 % had already 1 year or more of working experience. Not gaining employment in the next 3 months would have caused financial hardship for 58.06 % of our sample population (of which 41.94 % from a moderate to a great amount).

8_{See Model Structure, p.7}

Graduating Population Sample, n = 29

Indicator Value Percentage %

Sex M 51.6

F 48.4

Currently working Yes 67.74 Yes, no compensation 9.68

No 22.58

Max. start more than 5 months 38.71 within 5 months 19.35 within 3 months 22.58 within 1 months 9.68 as soon as possible 9.68

Distribution of our sample of student population

We collected the results by means of an online survey mainly because of the relative prior well-established literature avail-able regarding the processes of job search (i.e., sequential models, two-dimensional job search behaviors), which were not the focus of our survey. On the contrary, we have focused on investigating how individual differences previously studied in relation to traditional job search were to be characterized if contextualized to a specific digital job search scenario. In this sense we have divided our survey in different sub-categories, for a total of 66 questions (mainly Likert scales and open-ended questions). The nine categories were : 1. Demograph-ical Information, Global Self-Esteem and Financial Need; 2. Job Search Clarity; 3. Task-Specific Self-Esteem; 4. Digital Job Search Self-Efficacy; 5. Preparatory Digital Job Search Behaviors; 6. Active Job Search Behaviors; 7. Job Offers (<3 months); 8. Awareness and Suggested Use Of PI-Tools Assisting Digital Job Searches and 9. Perceived Usefulness of PI-Tools Assisting Digital Job Searches.

Measures

Global Self-Esteem was measured using a shortlist of 5 of the 10-item scale of the Global Rosenberg’s (1965) Self Esteem Inventory [19].

Financial Hardship was captured by applying Vinokur and Schul (1997) 5-point Likert scale of financial strain (such as "How difficult it is for you to live on your total household income right now?").

Task-Specific Self-Esteem was surveyed thanks to an inven-tory based on a 10-item scale related to job search (such as a reversed scale for the question "Overall, I don’t expect to be very good at job search").

Job Search Self-Efficacy (JSSE) was in turn measured by a 8-item scale used for the first time by Ellis et al. (1983) [11] for which we purposedly modified the setting of questions to adapt to the digital context of our research (giving prominence to those preparatory and organisational behaviors), and in which the beliefs of one’s efficacy were strictly confined to the intermediary tasks and goals, and not to the overarching

objective of gaining employment9_{. As this indicator is the} main source of inspiration of our solution, we include the items in Appendix I.

Preparatory Digital Job Search Behaviors have also been adapted and measured with a focus on gaining knowledge on activated digitally-led efforts in preparing the ground for active job searches.

Active Job Search Behaviors was measured by asking the re-spondents about their active job search behaviors. As these behaviors cover different modality (phone, real presence dur-ing interviews) they were not specifically modified.

Job Offers in the past 3 months were also recorded to observe if they might be influenced by other indicators.

Awareness and Suggested Use of PI-Tools targeted a re-spondent’s knowledge of tailor-made monitoring or self-organisation tools for the management and self-organisation of digital job searches and their existing or suggested applica-tions.

Perceived Usefulness of PI-Tools, for which we faced respon-dents with the "My Searches" and "My Jobs" screens of the Indeed aggregator’s features, was also quantified.

Results As Requirements

We focused on analysing the aspects of Job Search Self-Efficacy that were immediate concerns, and subsequently in-cluded them in the logic of our prototype (See Figure 5). Requirement 1: Self-organizing & Follow-up

Among the scale items of this particular indicator, 62.07% of respondents said they somewhat disagreed or had a neutral stance on being able to self-organize and follow-up with as many jobs openings they thought necessary to receive at least one job offer. Requirement rationale: An area of the PI tool should allow the user to have an overview of the job openings he/she is interested in, and allow for a quick understanding of where follow-ups are needed.

Requirement 2: Single Information Location

72.41 % of the respondents had either a neutral stance or a disagreement (with 37.93 % openly disagreeing) with the state-ment that they were always able to have all the information about the opportunity in a single location. Requirement ra-tionale: The PI-tool should not trigger unnecessary external browsing behaviors, and information should be visualized in a single place.

Requirement 3: Preparedness

Around 55 % of the respondents had either a neutral stance or a disagreement with the statement that they were always feeling prepared in case they would receive a call or an interview proposal. Requirement rationale: In case an event such as incoming phone call, or request of information regarding the recruitment process arises, the user should be able to locate

9_{As Saks et al. (2015) have noted in a paper aimed at}

reconceptu-alizing job search self-efficacy, this is a common mismatch, that is to misinterpret the measured self-efficacy as a direct measure of job search outcomes.

immediately all the relevant information in the fewest amount of clicks.

Requirement 4: Recall

A percentage of 62.06 % of the respondents had either a neutral stance or a disagreement with the statement that they were always able to apply for the jobs they were most interested in at browsing time. Requirement rationale: As users might postpone the application process, browsing for opportunities and applying must be considered as two separate events in time, therefore, the visualization of the self-monitored tracking of vacancies should include quick retrieval of the possibility to apply.

Requirement 5: Re-organising & Discarding

A percentage of 48.28 % of the respondents had either a neu-tral stance or a disagreement with the statement that they were always able to save and re-organize their preferred vacancies for further use in the application process. Requirement ratio-nale: being able to save or discard vacancies based on users’ appraisal of the current status (or utility) should be included. Accordingly, we did not include in our rationale the items in which the respondents showed higher level of self-efficacy (Quantity of Information; Strategy).

Figure 5: Self-efficacy is the belief that one can perform effectively the behavior required in a given situation, or the overall confidence in the job-search context. In this metric (top bars = strongly agree; lowest bar = strongly disagree). See Appendix I, for an overview of the scale.

Perceived Usefulness and Awareness of self-organisation tools

Perceived Usefulness

To shield our proposal against possible early drop-out or dis-engagement, we also investigated the perceived usefulness of existing solutions present on the web. For this matter, we asked our respondents to rate the Indeed Archive components against their perceived usefulness. Asked if they considered

the My Jobs and the My searches useful features in organiz-ing digital job searches, and if those features would increase efficiency in applying for more positions and receive more job offers, 75 % stated that the My Jobs features (allowing for a self-organisation and tracking of statuses of different vacancies) would help them organise their searches; and 53.57 % believed this feature could increase efficiency in the job search process and allow for more applications. On the other hand, only 44.85 % of the population trusted the My Searches option to be related to a more organised search, whereas 68.97 % thought it could increase efficiency and allow for more ap-plications. If considered as Personal Informatics features, the analysis of our data provides a good ground to believe that engagement with tools dedicated to online job search could be high. On the other hand, the relative lack of knowledge about them made it difficult for our sample to distinguish the tools from their proprietary platforms.

Awareness

In a specific open-ended question related to their knowledge of any dedicated tool meant to facilitate the organization of job searches, it appeared clearly that respondents were either (a) confusing specific self-organisation tools with the job boards themselves10; (b) mentioning other tools for very specific aspects of job search11_{or c) proposing a wide range of possible} workarounds12.

A specific respondent seemed to go deeper in analysing the shortcomings of looking for jobs in a digital context by high-lighting the lack of integration between platforms and pro-cesses13.

MODELLING A PERSONAL INFORMATICS TOOLS TO ASSIST DIGITAL JOB SEARCH

Combining the found requirements with previously reviewed elements of models at play both in job search and in Personal Informatics was the challenge of the modelling part of our

10_{A respondent wrote: "I am not currently looking for a job, but I}

would consider Monsterboard simply because that is the name that pops to mind.".

11_{"Latex for my résumé"}

12_{A respondent highlighted a possible full preparatory job search}

cycle, using 3 different tools: "Bookmarking and saving websites with job applications I’m interested in online in a browser. Using Sketch to update [my] résumé, and maybe Word or any other word processor to keep track of jobs and companies you’re interested in."; another was more focused on a combination of job boards and communication tools: "Even though I didn’t use it for this purpose, the first software that comes to mind is Excel and Outlook (Microsoft Office). I feel comfortable with using both of them and I consider them to be the most relevant of which I know of. From another point of view, LinkedIn is a good way of being updated with relevant job offers or professional changes in potential hiring companies, based on my preferences.".

13_{"I need to have an overview in one place. Now all my applications}

are scattered around and on every jobsite you come you have to use a new account. Sometimes they are not even integrated with Linkeding (SO ANNOYING!). I want a place where I can keep all relevant things. - Information about the job market [...]. Also that you can see the progress of the application process, like a progress bar in which part of the process you are and whatever is happening.

solution. We propose a simple way to consider our solution as an attempt to address frictions and barriers seen in the stage-based model of PI and by enhancing one particular individual difference at play in the job search process : self-efficacy. Considering the Self-Efficacy aspect of Job Search as con-textual to the proposed Personal Informatics tools allows us to maintain our indicators for further testing (we could rea-sonably re-test our solution against the same indicators to understand if there are improvements over time). A possible design research could see two distinct groups of job-seekers being compared on both job search outcomes and self-efficacy indicators.

Model Structure

Our solutions is modelled around the combination of the previ-ously mentioned requirements (taken from the revisited scale for Job Search Self-Efficacy adapted to Digital Job Search) and proposed components related to common components of self-monitoring, self-tracking, goal and task setting of PI Tools.

Figure 6: Our proposed model of requirements integration within the Job Search Self-Efficacy context of job search.

Components

We maintain our 5 requirements (Self-organisation and Follow-up; Single Information Location; Preparedness; Recall; Re-organising & Discarding ) as informants and propose three main components and an additional geographical search tool (see live prototype) :

1. A Timeline View with Tasks/Goals Setting and Tracking 2. An In-Built Job Search Engine Relying on a Third-Party

Job Board Platform (Monsterboard)

3. A Leads Visualization Pane (with a vacancy pane integrating a quick Company Search and Vacancy Url functions) 4. An Additional Geographical Search Tool

We gave a paradigmatic importance to the Single Information Locationof the PI Tool by integrating an In-Built Job Search, in order to reduce frictions in job search exploration and data collection. However, for prototyping reasons, we acknowledge

that other forms of integration exist for vacancies’ or compa-nies’ data collection (plugins, etc.) : the reflection around these very specific aspects fall out of the scope of our research, but could represent very interesting aspects in user-centered design studies surrounding Digital Job Search and efficiency. Also central in our proposed solution is the Timeline View, which does not appear as central in the previously reviewed solutions. Including the time-varying nature of job search intensity and allowing for a temporal visualization of the job search behaviors could echo Gali´c call [13] for a better under-standing of the temporal aspects of job search intensity. As mentioned by this author, data related to time-varying aspects of job search intensity aren’t yet fully understood. Job Search might wrongly be understood as the time in which a person is unenmployed, we know instead that it is a diffuse behavior that has less to do with the different employment status and more about the circumstances of each individual. Given the existing walled gardensnature of Job Boards and their relative data unaccessibility, it might be very difficult for policy-makers or analysts to measure or understand what are the real efforts made by job seekers online to gain employment, as opposed to their employment status.

PROTOTYPE

The fully-functional clickable prototype is available athttp: //qqzttp.axshare.com/timeline_leads_at_t1_.html. Although this is a prototype, the functionalities mentioned in this paper are fully available, meaning that it could be used as close to a production version. This prototype has been built using Axuresuite of prototyping tools [5], including AxShare for live rendering [4] and availability and it is best viewed in large displays.

Main Interface

In Figure 7 we show the main intention of the User Interface, which encompasses a central frame dedicated to the usual browsing of vacancies. Below, a Leads Visualization allows for the self-monitoring of the preferred vacancies, which de-tails can be saved on the leftmost Vacancy Panel.

The main interface is mainly informed by the Single Infor-mation Locationrequirement, that imposed a limitation in unnecessary external browsing behaviors14.

Leads Visualization and Vacancy Pane

The Leads Visualization allows the user to have a compara-tive view of his current preferred/saved vacancies (Figure 8). In the same way, the user is able to re-organise or discard his vacancies by triggering the corresponding actions in the left-most Vacancy Panel answering this way to the Re-organising and Discardand Recall requirements.

14_{It has to be said that some calls to external vacancy}

por-tals/aggregators might be prevented by iframe rendering limitations and thus inevitably require opening more tabs. Nevertheless, partner-ing with job aggregators could help circumvent this problem, or even be prevented by a more advanced coding.

Figure 7: The main user interface during a typical browsing stage (In-Built Job Search). The central pane can be switched to 2 other modalities (Timeline View and Geographical Search).

Vacancy URL

As users might postpone the application process while brows-ing, we have included an option that directly talks to the Recall requirement, by allowing the user to re-open the vacancy in the frame view (vacancy URL component). Considering the previ-ously mentioned models for PI, this function should reduce or even prevent lapsing, as applying to the vacancy would/could still happen within the context of our solution. Lastly, we have limited the amount of vacancies to 7 in this prototype, but do not prescribe any further limit upwards in case a full-fledged PI would be developed.

Company Information

We allow the user to further gain knowledge about the com-pany he is interested in by triggering a query in a random search engine. Information collected this way could poten-tially nurture the motivation to apply for a specific vacancy. As these kind of information might not pertain to the usual flow of data collection during digital job search, we implemented this search, which results can be easily stored within the same Vacancy Pane.

Figure 8: The Leads Visualization includes vacancy, company and salary information that can be self-tracked either at browsing time, or during a more advanced moment in the recruitment process.

It has to be said that this pane is in some ways independent from the In-Built Job Search, meaning that users could rea-sonably still use components of the PI tool and ignoring others. Additionally, although being conceived as a support for digital job search, there are no limitations of use if leads were to come from other sources than the web.

In-Built Job Search

The prototype allows the user to trigger searches that rely on third-party job aggregators that accept iframe rendering (See Figure 9), by implicitely modifying the URL query that

usually is triggered by the search engine of the aggregator. We coded the prototype in a way that the results shown could respect the intention of a potential dutch job seekers.

Figure 9: A third-party aggregator reads the query launched within the PI Tool and triggers automatically the corresponding job browsing modality

Timeline View

The Timeline View is the central part of the prototype (See Figure 10) and is responding to the Self-organization and Follow-uprequirement. These requirements made clear that our surveyed individuals weren’t as confident in their ability to self-organize, but also following-up with given vacancies. The functionalities we have implemented in this specific area of the prototype include and tasks-setting, but also goals-and tasks self-monitoring.

Coupled with the description box in the Vacancy Panel, it is easy for the user to collect intermediary data about communi-cations, further information regarding deadlines, interviews, etc, without messing with the overarching perception of the time issue. We believe visualising the temporality of tasks and goals related to processes pertaining to each vacancy, could give an additional incentive to job seekers in increasing their leads, as they have a clear visualisation of their relative current processes.

Figure 10: A snapshot of the Timeline view, allowing the user to track ad-hoc and customized tasks and goals related to Job Search.

We gave the user several ad-hoc tasks and goals, that would respect the possible interactions that a usual recruiting process might follow. To avoid disengagement due to the limited op-tions given, we have allowed the user to customize the content of the events he/she would want to self-monitor in the timeline. As the modalities of communication might be as varied as

the devices on which job search is carried out, giving a high flexibility to the user seemed central. Additionally, we do not promote a typical process structure for the obtainment of job offers: considering that our theoretical framework informs us about a certain structure (preparatory job search - active job search behaviors) we consider these ad-hoc tasks and goals as active job search behaviors but agree that they might dif-fer from the traditional behaviors cited in literature, as our paradigm is around digital job search.

Another very important indicator such as self-esteem could be measured by studying the usefulness of this prototype. Self-esteemcould potentially be affected by not being hired, and thus maintaining it high during the process is a known problem during Job Search. As discarded vacancies would be put in perspective with other current visualized processes (Figure 11) - in which the user has still chances to gain employment, - they

might loose their negative impact on global self-esteem.

Figure 11: A possible Job Hunting scenario, with two solid leads, a partially filled lead, and an unsuccessful application

Geographical Search

An additional functionality has been built within the proto-type, by linking a geographical visualization made in Tableau (www.tableau.com/) to real vacancy data from Indeed’s Job API (https://www.indeed.com/publisher). We link this component to a possible development of the prototype towards a play-ground for further comparative analysis of labour market data visualisation, that would complement possible other existing components of our solution (i.e. Leads Visualizations), or represent a field for testing components linked to PI tools to other psychological variables at play in Job Search, such as perceived control over employment outcomes. This kind of visualizations could give a boost to opportunity perception, and thus increase global self-esteem in a radical manner.

Figure 12: The rendered geographical visualization of opportunities: this view could increase global self-esteem at play in Job Search by increasing opportunities perception

IMPLICATIONS AND DISCUSSION

The implications of our proposed solution go beyond what is actually to be experienced with the actual prototype.

First, we believe that this is one of the few initial attempt to give more insight and control to job seekers over their experience of digital job search.

Second, this study is an attempt to operationalize job search self-efficacy around digital job search. From the data that we have collected, it seems digital job search is a highly frag-mented activity - in which job seekers might find difficult to exert self-monitoring -, which subsequently might drive job seekers towards non-informed or worse, wrong career de-cisions. Being able to compare existing opportunities in an elegant, appealing time-bound view seems to be a first step in that direction. Usability and alignment between the tool and the user’s goals have been proposed as influencers of self-efficacyas design principles for Personal Informatics Tools [1]: in this sense we have given high relevancy to the actual contents (goals) of Job Search Self-Efficacy as proposed in a digital version of that indicator items.

Third, Job Boards and job aggregators might make use of collected job seeker information and substitute policy-makers in important societal issues (i.e. unemployment), this resulting in an increased fragmentation regarding the monitoring of data surrounding Digital Job Search Behaviors. At a time where big players of the internet such as Google are planning to enter the lucrative market of job search aggregation [15], we should begin to think about the best ways to maintain job search and career planning topics in which the job seeker stays central in the process.

Lastly we make ours the call of authors such as Gali´c [13], re-garding the time-varying nature of job search intensity. In fact, we think it is central, and acknowledge that certain aspects that could mitigate the consequences of a drop in job search inten-sity could be added to a fully-functional Personal Informatics Tool for Digital Job Search. For example, a combination of offline job search interventions and online self-monitoring tools could be proposed as a platform, where the focus would be targeted at improving or training job seekers in specific job search behaviors, and - especially - when they need it. Also,

we think that for specific populations (such as students), who are by definition under some pressure during the end of their studies, a solution that could integrate previous work around the time-bound journey of students, such as in Buzzo and Phelps (2016) [10] and a Personal Informatics Tool such as this one, are all possible paths that could be taken in order to mitigate the effects of extended school-to-work transitions and youth unemployment.

REFERENCES

1. J. Fogarty A. Andrew, G. Borriello. 2011. Understanding Self-Efficacy and the Design of Personal Informatics Tools.

http://www.personalinformatics.org/docs/chi2012/andrew.pdf (2011), 1–4.

2. Ramon Aldag, Donald P. Schwab, and Sara L. Rynes. 1987. Theories and Research on Job Search and Choice. (1987). DOI:

http://dx.doi.org/10.13140/rg.2.1.1004.4323

3. Burke K. Ashforth and Alan M. Saks. 2000. Change in Job Search Behaviors and Employment Outcomes. Journal of Vocational Behavior56, 2 (April 2000), 277–87.

4. Inc. Axure Software Solutions. 2017a. Axure Share | Host and Share Axure RP Prototypes.https://share.axure.com. (2017). [Online; accessed 29-Jun-2017].

5. Inc. Axure Software Solutions. 2017b. Axure’s website. https://www.axure.com/. (2017). [Online; accessed 29-Jun-2017].

6. Albert Bandura. 1988. Self-Regulation of Motivation and Action Through Goal Systems. Springer Netherlands, Dordrecht, 37–61. DOI:

http://dx.doi.org/10.1007/978-94-009-2792-6_2

7. Alison E. Barber, Christina L. Daly, Cristina M. Giannantonio, and Jean M. Phillips. 1994. Job Search Activities: An examination of changes over time. Personnel Psychology47, 4 (December 1994), 739–766. 8. Gary Blau. 1994. Testing a Two-Dimensional Measure of

Job Search Behavior. Organizational Behavior and Human Decision Processes59, 2 (1994), 288 – 312. DOI: http://dx.doi.org/10.1006/obhd.1994.1061

9. Hans G. Bloemen. 2005. Job Search, Search Intensity, and Labor Market Transitions an Empirical Analysis. The Journal of Human Resources40, 1 (2005), 231–269. http://www.jstor.org/stable/4129572

10. Daniel Buzzo and Philip Phelps. 2016. JourneyMap: Visualising the time-bound student journey. In Electronic workshops in computing (Online). BCS, The Chartered Institute for IT, Bristol.

http://ewic.bcs.org/content/conWebDoc/56277 11. Rebecca A. Ellis and M. Susan Taylor. 1983. Role of

self-esteem within the job search process. Journal of Applied Psychology68, 4 (November 1983), 632–640.

12. Daniel Epstein, An Ping, James Fogarty, and Sean Munson. 2015. A lived informatics model of personal informatics. In Proceedings of the 2015 ACM International Joint Conference on pervasive and ubiquitous computing (UbiComp ’15). ACM, Osaka, Japan, 731–742.

13. Zvonimir Gali´c. 2011. Job Search and (Re)employment: Taking the Time-varying Nature of Job-search Intensity into Consideration. Revija za socijalnu politiku 1, 1 (April 2011), 1–23.

14. Indeed.com. 2017. Indeed My Jobs Section. https://www.indeed.nl/myjobs/. (2017). [Online; accessed 29-Jun-2017].

15. Áine Cain. 2017. Google launched its own job search engine — here’s how it works. Business Insider (2017), 1–4.https://goo.gl/b9kheh

16. jobhero.com. 2017. Jobhero.https://www.jobhero.com. (2017). [Online; accessed 29-Jun-2017].

17. Ruth Kanfer, Connie R. Wanberg, and Tracy M. Kantrowitz. 2001. Job Search and Employment: A Personality–Motivational Analysis and Meta-Analytic Review. Journal of Applied Psychology 86, 5 (October 2001), 837–855.

18. Elisabeth T. Kersten-van Dijk, Joyce H.D.M. Westerink, Femke Beute, and Wijnand A. Ijsselsteijn. 2017. Personal Informatics, Self-Insight, and Behavior Change: A Critical Review of Current Literature. Human–Computer Interaction0, 0 (January 2017), 1–29.

19. D. J Lee. 1969. Society and the Adolescent Self-Image. Sociology3, 2 (May 1969), 280–280.

20. Ian Li, Anind Dey, and Jodi Forlizzi. 2010. A stage-based model of personal informatics systems. In Proceedings of the SIGCHI Conference on human factors in computing

systems (CHI ’10). ACM, Atlanta, Georgia, USA, 557–566.

21. Songqi Liu, Jason L. Huang, and Mo Wang. 2014. Effectiveness of Job Search Interventions: A

Meta-Analytic Review. Psychological Bulletin 140, 4 (July 2014), 1009–1041.

22. Rosemery O. Nelson and Steven C. Hayes. 1981. Theoretical Explanations for Reactivity in

Self-Monitoring. Behavior Modification 5, 1 (1981), 3–14. DOI:http://dx.doi.org/10.1177/014544558151001 23. Daniel Power and Ramon Aldag. 1985. Soelberg’s Job Search and Choice Model: A Clarification, Review, and Critique. The Academy of Management Review 10, 1 (January 1985), 48–58.

http://search.proquest.com/docview/1305653119/ 24. V. S. Ramachaudran. 1994. Self-efficacy. Encyclopedia of

human behaviour(1994), 71–81.

25. Alan M. Saks and Blake E. Ashforth. 1999. Effects of Individual Differences and Job Search Behaviors on the Employment Status of Recent University Graduates. Journal of Vocational Behavior54, 2 (1999), 335–349. 26. P. Soelberg. 1967. A Study of Decision Making: Job

Choice. Carnegie-Mellon University, Pittsburgh. 27. Finis Welch. 1977. What have we learned from empirical

studies of unemployment insurance? Industrial amp; labor relations review : a publication of the New York State School of Industrial and Labor Relations, a statutory college of the State University, Cornell University, Ithaca30, 4 (1977), 451–461.

28. Hannes Zacher. 2013. Older job seekers’ job search intensity: the interplay of proactive personality, age and occupational future time perspective. Ageing and Society 33, 07 (October 2013), 1139–1166.

Appendix I

As mentioned, this appendix contains all the questions relative to JSSE (Job Search Self-Efficacy), that have been modified to represent a consistent possible overview of behaviors supporting the overall objective of gaining employment.

Note how questions marked with (*) are complementary questions that would inform basic components of our solution, and the remaining 8 have been weighted to be reduced at 5 requirements informing our prototype.

Usual Likert scale were used for the scoring of the scale : (Strongly agree, Somewhat agree, Neither agree nor disagree, Somewhat disagree, Strongly disagree)

1. If I am really interested in a job I found online, I feel I am able to plan and execute a good strategy to move further in the application process.

2. I feel I am able to self-organize and follow-up with as many job openings that I think are necessary to get at least one job proposal.

3. Based on experience, or your own appraisal, how many job applications do you think are necessary to get at least one job proposal, based on your curriculum, experiences and/or professional profile. (*)

4. I feel I am always able to gather and store enough information for further use about companies I might be interested to work in.

5. I feel I am always able to apply for the job(s) I was most interested in at browsing time.

6. I feel I am always able to select, save and re-organize my preferred vacancies for further use in the application process. 7. Once applied for job opening(s) online I feel I am always in control of the application process.

8. Once applied for job opening(s) online I feel I am always prepared in case I would receive a call or an interview proposal. 9. Once applied for job opening(s) online I am always able to have all the information about the opportunity in a single location. 10. I think having more information at hand, regardless of my skill match with the current opening, would have an important