Research studies differ greatly, and depending on the topic under investigation, the target demographic, and many other factors, each researcher makes different choices during the study design. We have done our best to make sure Ethica is fully configurable and can match the specifics of the study design as closely as possible. In order to do so, we have worked closely with different research groups on topics including smoking habits, foodborne illness, psychology, and kinesiology to make sure the platform covers the needs for these studies.
In this section we will discus the assumptions we have made in design and implementation of Ethica, and what configurations are possible in Ethica. We are also planning to write down different use-cases of the studies already deployed with Ethica to demonstrate the possibilities. Meanwhile we still understand there are many studies which might not be totally feasible with the current capabilities of Ethica. If you think the materials presented here is not sufficient for your needs, please let us know and we can either improve the system to cover your scenario as well, or suggest alternatives.
Studies in Ethica
The main building block in Ethica is a study. An Ethica study captures many details of a research study as it's designed by the researchers and stores them in a structured way, so it can be understood by Ethica app on different platforms (such as iOS or Android). These details include:
- What type of data should be collected.
- How long each data collection should last.
- Who will have access to the raw data or analysis results.
You can define all the study specifications using Ethica Dashboard and subsequently deploy the study without the need to do any programming. At the same time, Ethica offers platform-specific API to extend the system for studies with more complex requirements.
In the following sections, we will review each of these in more details.
Any research related to human behaviour involves collecting data from survey responses or using sensors embedded in mobile devices. The variety and complexity of the mediums available to collect any of these data, and the technical skills required to implement a customized data collection tool makes it nearly impossible for researchers to take advantage of all the available technologies.
Ethica enables any research group to easily collect the data they need without any technical background. These data can come from different sources, such as:
- Common sensors available in most of the off-the-shelf smartphones such as GPS or Wi-Fi.
- Specialized sensors available as a peripheral, such as EEG headband.
- Electronically generated data such as records of applications used or calls made.
- Survey responses, whether triggered by user or triggered automatically based on a specific context.
Each type of data in Ethica is referred to as a streams. As part of the study definition, you should specify one or more streams you need to collect, the durations and conditions under which that stream of data should be collected, and Ethica will take care of the rest.
Ethica streams are of two types: automated or self-reported. Each of these types are described below. You can read more about available streams in Streams section.
Automated Data Streams
Automatic data streams are the ones that don't require active participant engagement. While the participant is part of the study, Ethica automatically collects data from these streams and stores it in the specified location. This all happens in the background without any interaction from participants. An example of this can be sensor-based data such as GPS or accelerometer, or electronically generated data such as reports on application usage or Internet content exposure.
Ethica provides many automated streams out of the box, which can be deployed with a few clicks. The exact type and number of streams available is different for different platforms. For example in Android Ethica supports 19 automated streams, including GPS, Screen-State, Accelerometer, and Bluetooth.
While the stock automated stream can cover majority of basic needs, there are many custom streams which Ethica does not support by default. For these cases, Ethica offers an API which allows research groups to design their custom stream. Using customized streams, researchers only have to implement the data collection logic for the new stream, and Ethica will handle other tasks such as data encryption, upload, and storage.
Additionally, the custom streams can be reused for different research studies, or it can be shared between groups who require the same type of data. Ethica currently offers three custom streams, for collection of custom data such as stress level using Empatica smart-watch.
Custom streams are only available on Android-based devices. Certain restrictions on iOS prevents implementation of this feature on iPhone.
Automated streams play an important role in abstracting the complexity of data collection, particularly for the studies leveraging participants' personal devices. In these scenarios, Ethica allows researchers not to be concerned about different devices used by different participants. Researchers can pick the stream of data they need, and Ethica collects the data from devices which support the that stream, and reports the devices which do not support it.
Self-report Data Stream
An important source of data for any study is the participants' self-reports in the form of surveys or EMAs. Each study in Ethica can include as many surveys as needed. There are three ways to trigger a survey in Ethica: through online dashboard, through the app, or automatically based on the user's context. Each survey can be triggered in one or more of these methods.
Triggering a survey on participant's device through the dashboard is available to any member of the research team. They can select a survey and a subset of participants, and release the survey to their device. As soon as the selected participants are online (or right away, if they are currently online), they receive the survey.
A survey can be configured so participants can answer it any time and as many times as they want. This is similar to the approach used in ecological momentary assessment (EMA). In this case, participant decides when he or she wants to fill the survey. In order to do so, they can open the app, tap on the related button, and respond to the questions. Examples can be asking participants to record their food consumption, record the quality of their conversation throughout the day, or to record their exposure to tobacco advertisements.
Additionally, a survey can be triggered automatically based on participant's context. As we discussed in Automated Data Streams section, a study can continuously record data on participant's context. If the survey requires to analyze these data to decide the triggering time, it can provide a context analyzer. A survey's context analyzer will have access to the recorded data immediately after the collection, and it can decide the following:
- Whether the survey has to be triggered or not?
- If the survey has to be triggered, what set of questions should be asked?
- If there is any context-dependent placeholder in the question content, how they have been filled.
For example, a survey can have a context analyzer which triggers the survey only when participant visits another study participant. The context analyzer can continuously process the collected data, and when the data indicate two participants are in close proximity, it can trigger a survey such as
Have you met _____ recently?
It also can detect who is the other participant, and fill the placeholder accordingly. Therefore the participant will receive a question similar to the following::
Have you met Alice recently?
Ethica offers a few common context analyzers which can be easily used for a survey, without any custom programming. These context analyzers leverage the data on time and location to trigger a survey. Using time-based context analyzer, for example you can design a survey to be triggered once every other day at 8am, or to be triggered with 30% chance between 8am to 9pm. Using location-based context analyzer, you can design a survey to be triggered when participant changes their location and goes from point A to point B, or when they enter a specific geo-fence.
If the default context analyzer is not sufficient for your scenario, you can implement your custom context analyzers. Ethica offers an API for this purpose, through which you can implement a custom context analyzer which has access to all study data, and can process them continuously and instruct Ethica on when the survey should be triggered and what questions should be asked.
Similar to implementing a custom stream as was described in Automated Data Streams section, a custom context analyzer can also be used in multiple studies and shared between different groups. We are currently collaborating with different groups on designing different context analyzers including:
- Detecting when two study members have met in person.
- Detecting when a conversation is taken place
- Detecting when the participant has surfed the Internet on their device.
We hope to share these context analyzers with the research community as soon as they are available, and we are looking forward to work with other teams to implement and share more of such context analyzers.
An important part in designing each study is to specifying the study period. Ethica defines three variables for this purpose:
- Study period is defined per study and specifies the period for which the study will be conducted.
- Participation period is defined per participant and specifies the period each participant is part of the study.
- Data collection period is defined per type of data being collected and specifies the period which Ethica should collect the requested data.
The study period defines the time and date when the study starts and the time and date when the study ends. During this period the study is marked as active. Ethica dashboard enables or disables certain features depending on the study period. For example, the data analysis is only applicable for ongoing or past studies, as there is no point to perform data analysis on a study which has not started yet.
Each study should also define what is the participation period for each participant, i.e. from what time and date the participant's device should start data collection and at what time and date it should stop the data collection. As it's not readily possible to define this per participant, particularly for larger-scale studies, you can specify participation duration instead, as the number of days that the data collection continues for a given participant. This can be any number from 1 day up to the study duration (number of days between start and end of study period). The participation period for a given participant starts from the time they join the study, and ends after the participation duration in days, or the end of study period, whichever is earlier.
Additionally, for each stream of data you collect (automated or self-report), you can specify how long that data collection should last. This time is relative to the participation time, and is specified in the form of
x days after until
y days before start study participation.
You will always have the option to modify the durations while the study has not started yet, or is in progress. Specifics on how these durations can be modified are explained in the following sections.
As an example, assume you are planning for a study which requires collecting GPS, Wi-Fi, Accelerometer, and Survey Responses. You can set the study period from June 1st 2016 to June 1st 2017, and participation duration to 60 days. You further specify the GPS, Wi-Fi, and Survey Responses data are needed for the whole participation period, while Accelerometer data is only needed for the 3rd to 6th week of the participation, inclusive.
Now if a participant joins this study on July 1st 2016, she will be part of the study from July 1st 2016 to August 29th 2016, which is refered to as participation period. During this period, the app continuously collects GPS, Wi-Fi, and Survey Responses, based on the surveys defined for the study. The Accelerometer data will be collected only from July 15th 2016 (beginning of the third week) to August 12th 2016 (the end of the sixth week).
While anyone can create an account in Ethica and participate in a study, defining a new study and accessing the collected data from participants is only available to researchers. Researchers are individuals who are members of a research group in Ethica. You can register your research group by simply send us an email with the name of your group and its members.
Research groups allows Ethica to control the access to the data collected for a specific research study. You can define multiple research groups, and share it with different researchers you work with. Each researcher, including yourself, can be part of multiple research groups at the same time.
When creating research groups it's important to note:
- Each researcher part of a given group will have full access to the data and analysis from all the studies defined in that group.
- Any study created in a research group, and any data collected for that study, is automatically shared with all other researchers in that group.
In practice, members of a research groups are usually the individuals who are mentioned in the institutional review board (IRB) application for that study, and hence have been granted access to the study data, rather than members of the same research lab.