Lifelogger

Fostering College Students’ Healthy Behavior Through Self-monitoring Data Reflection

Authors: Haining Zhu, Sarah Almoaiqel, Frank T. Materia, Jiawei Chen, Natalie T. Cope, John M CarrollAffiliation: Penn State UniversityYear: 2019

My Role: Project lead, lead designer, developer

Summary

Background: College-life presents a plethora of possibilities for new social experiences, activities, and other prospects, it can be difficult for college-aged young adults to decide how to spend their time in an efficient, healthy, and personally beneficial fashion. With so many experiences, opportunities, and possibilities vying for the college student's time, learning to manage time for optimum individual benefit is of the utmost importance.

Goal: Our research goal is to design, deploy, and evaluate a system that harnesses commercially available sensor-based technology and leverages a semi-automated tracking approach to help students become more aware and reflect on their time, while reducing the burden of fully manual tracking.

Method: We recruited 14 participants and assigned them into control and experimental conditions. We collected pre-study survey data and let both groups interact with the prototype daily for seven days. A post-study survey was collected and followed by an in-person semi-structured interview.

Result: We find that LifeLogger is able to facilitate self-awareness, reflection, and the potential for time management behavior changes.

LifeLogger Design

College students are constantly struggling to maintain work-life balance. A common saying is that of the three pillars: sleep, social life, and grades, students can acquire only two. However, previous research has shown that this might resulted from inefficient use of time.

Inspired by the three pillars of a balanced life for college students, and prior research on semi-automated self-tracking. We attempted to address lifestyle balance problems by utilizing tracking of time use, and visualization to encourage student self-awareness and reflection of how they spend their time. We design and develop a web-based prototype application, LifeLogger. The idea is to use a semi-automated tracking (combine both manual logging data & automatically captured data) to monitoring three important aspects of students' life: work, personal maintenance, and leisure activities.

Term definition:

Work: studying, jobs, interviews, classes, homework, group projects/work, etc

Personal maintenance: hygiene, cooking, eating, sleeping, exercising, personal transportation, etc.

Leisure: social life, shopping, watching movies/TV, playing games, attending parties, club activities, and group sports.

These are the first 2 version of the LifeLogger design. We did not implement the 1st version (left), because it only utilizes automatically captured data as a feedback mechanism. We need to utilize manual logging to reinforce students' intrapersonal reflection, therefore we revised the design and implemented the 2nd version of LifeLogger (right).

The system consists of two components: a dashboard and a daily survey. Using APIs of three commercial services (RescueTime, Fitbit, and Moves), we incorporated automatically captured data on the left-hand side, and collecting daily survey from students' manual logging on the right-hand side.

Implementation

We implemented LifeLogger as a web-based service. The back-end is implemented with the Python Flask framework, which takes charge of: 1) user authentication, 2) access management, and 3) data retrieval (left).

Our frontend relies upon the JQuery library, CanvasJS, and Google Maps APIs for visualization. On the dashboard, various data from the three services asynchronously load through AJAX. Once data is available, Fitbit and RescueTime data are visualized with CanvasJS as a time series, and Moves data are visualized via Google Maps.

Pilot

We conducted a small pilot study (N=8), using convenience sampling method. From these 8 participants we randomly assigned them into two conditions: n=3 for the control group (manual tracking), n=5 for the experimental group (semi-automated tracking). The study procedure include three main components: a pre-study session, deployment of LifeLogger, and a post-study survey and interview.

Pre-study session: Participants were then given an introduction to the study and were informed of the study procedures. The session was split into 3 parts: Introduction of study, procedures, and consent (10 mins); Pre-study survey, same for both groups (10 minutes); and Q&A (5 mins). Individuals in the experimental group were asked to install the apps and given a tutorial on how to use the system (15 mins).

Deployment: Participation lasted one week. All participants received email notifications every night at 10 pm, which requested that they fill out the daily web survey on the study website for the week. The survey took 10 - 15 minutes to fill out.

Control group: Participants recorded time expenditure via the daily web survey. The control group did not receive LifeLogger feedback from Moves, Fitbit, and RescueTime (right).
Experimental group: This group utilized LifeLogger feedback and visualizations. We designed a Semi-automated Day Reconstruction Method (S-DRM) for the experimental group. Various data, such as location (Moves), steps (Fitbit), and computer usage (RescueTime) were automatically collected, and participants received detailed time series feedback, which they could use for the day’s deconstruction.

Post-study survey & Interview: After completion of the one week study, we held a post-study session, in which participants first completed the post-study survey. We then conducted semi-structured interviews.

Running a pilot study for the initial prototype of LifeLogger enabled us to get feedback on system design. We have identified several design improvements to make before running a full study.

Revision of the LifeLogger Design

Adjusting the ‘Moves’ Location Bar Graph: The experimental group in the pilot study stressed that the location feedback was not useful to them. This issue lead us to design a better visualization for location in the full study. We changed map view to color coded block views (see below, same color indicates it is the same location).

Adding Weekly Summary Report: Although both groups found manual tracking and the pie chart visualization helped to increase awareness and reflection, the experimental group saw room for improvement. We added weekly view (line graphs) from student input and RescueTime, in general, summarized how college students spent their time over a week and provided students a more accurate perception about their time management.

The 3rd version LifeLogger user interface. Participants from both groups had access to Parts A (daily survey entries) and B (result of A in forms of pie chart). The experimental group was asked to import data streams from RescueTime, Fitbit and Moves, which would allow them access to part C (feedback from automatic data). The 1st line graph represents weekly data summary, with data points from Part B. The 2nd line graph represents student weekly RescueTime usage for productivity monitoring.

*Note: Data shown in this example is fictitious and does not represent data from a participant in our study.

Full study

we sought volunteers from the university website to randomly sample 14 students. Then they were randomly assigned into two conditions: n=6 for the control group (manual tracking), n=8 for the experimental group (semi-automated tracking). Participants’ ages ranged from 18 to 22 (mean = 20.75 for experimental group and mean = 19.83 for control group). There were 1-2 participants from each year of college in both groups. The study procedure remain the same with the pilot study.

Data Analysis

We find that features of data visualization in LifeLogger help to improve the granularity that participants reflected on their daily activities. We make a comparison on the number of log entries by participants from the experimental group and control group, and find that participants in experimental group entered more entries everyday compared with participants in control group.

Welch Two Sample t-test showed that the number of entries in experimental group (M=17.25, SD=10.74) were significantly higher than the control group (M=14.01, SD=5.03), t=1.96, p=0.05. However, due to inadequate sample size, we can not detect any significant difference between groups regarding time spend on the system.

Perceived the usefulness: We were interested in how participants perceived the usefulness of LifeLogger. The majority of participants in the experimental group were positive about the usefulness of LifeLogger: 5/8 participants considered the visualization feedback “extremely useful” or “pretty useful”, and 6/8 considered Lifelogger to be “extremely useful” or “pretty useful” in terms of helping with time management. In the control group, however, only 2/6 participants were positive about the usefulness of using a daily survey to aid time management.

Satisfaction: We found participants in the experimental group (2 pick “extremely satisfied”, 4 pick “pretty satisfied”, 2 pick “a little satisfied”) reported more satisfaction with LifeLogger when compared to the control group (1 pick “extremely satisfied”, 2 pick “pretty satisfied”, 3 pick “a little satisfied”).

Contribution

In our paper, we discuss how students react to different data visualizations, and suggest potential directions for future work which supporting enhance student self-awareness and self-reflection.

This research offers the following contributions:

1) It elicits innovation for evaluating and enhancing life balance for college students,

2) The pillars of a balanced lifestyle have not been assessed in the past using technology to this degree,

3) The tool has the potential to decrease user burden and support awareness using widely-accessible technologies,

4) Awareness and reflection will be supported by instantaneous data visualization,

5) We offer an alternative to real-time momentary sampling intended for assessing how college students spend their time.

Our study builds upon existing literature by providing a set of requirements and insight for designing technology which better accommodates college students’ awareness and reflection of time management. This presents opportunities for further research into student time expenditure and life balancing.

Page updated

Google Sites

Report abuse