5. Viewing Data



From the website

You can directly view participant data, both survey and sensor, from the Data section of a study, on the ODIN Researcher UI:
Every dataset you retrieve from this page can be downloaded as a CSV, using the button at the top-right corner of the table.

Downloads

Data.

You can download the entirety of your study's data as a collection of CSV files. The zipped files will contain everything stored on our database.

Codebook.

It's also possible to download the study's codebook from this page, which includes a PDF summary of the survey, as well as a human-friendly CSV version of the actual data. For more info, take a look at  this guide⁠ .


Data visualizer

The ODIN website has a visualization module, from where you can access a quick overview of your participants' data:

Axis.

  • The x axis represents time (adjustable with the controls at the bottom).
  • The left-y axis indicates which coupon's data is being represented (adjustable with the controls at the left).
  • The right-y axis tells you which data type each line of segments (rectangles) corresponds to. You can toggle data types with the checkboxes.


Segments.

A segment represents the intersection of all three axis; i.e., how many data points there are for the intersecting coupon, data type and timespan.
  • A grey segment without a border means there's no data at all.
  • A segment with a colored border means there's at least 1 data point for that coupon and data type, within the given time frame.
  • The intensity of the segment's internal color signals how much data there is, relative to the whole dataset for that data type.

Here's a closer look:
On the above example, you can see that coupon vzugwowmga had no data on Monday 8th, but did have some on the next day.
Coupon ugtjkcilmu, on the other hand, had only 1 GPS data point on Tuesday 9th.

Once again, you can see that the intensity of the colors is relative to other analogous segments. For instance, look at the difference in intensity between ugtjkcilmu's single GPS reading and vzugwowmga's single answer.

You can always get a hint of what you're seeing by hovering over a segment, and also click on it to get a more detailed overview of the data.

Answer segments.

Answer segments are somewhat special. For starters, not all answers are created equal. When getting a question, a participant may reply directly, by opening the ODIN app, writing a reply and submitting the answer. But they may also leave their phone on the side until the answer expires. Or they may submit an empty response; or even turn their phone off altogether. For this cases, ODIN still records the answer as a special case, namely expired, skipped or unset.
Since the visualization is about giving you a quick glance of the health of your coupons, showing this kind of answer would be misleading. And therefore the visualization module ignores them.


Health Metrics

Health metrics are per-coupon measurements aimed at making anomaly detection and data monitoring easier. At the time of writing we have support two types of metrics: expected vs. actual GPS readings and expected vs. actual answers.

Health metrics are only computed for studies that have the "Enable health metrics" option enabled, which can be set upon study creation:
Or edited on the About page of an existing study:

After having at least one day worth of data, and as long as it has the "Enable health metrics" option enabled, the study will automatically generate these measurements, which can be seen on the Health Metrics section:

Each measurement is calculated using a time interval relative to the Computation Date. For instance, the "Expected GPS readings" and "Actual GPS readings" measurements always go 24 hours into the past. The interval that each measurement considers can be seen on its info tooltip.

Let's look at a closer example by filtering the table to only show gurevshucy's metrics:
The last row's computation date is 2022-09-15 00:00:01.0, so that means that the expected and actual GPS readings are considering what we expected to receive from that coupon between 2022-09-14 00:00:01.0 and 2022-09-15 00:00:01.0.
We how what to expect because the GPS sensor has a sensor interval, which in this case is 5 minutes:
The actual GPS readings are just how many data points we got between those two dates. So this number would be equivalent to going to the Data page and retrieving the GPS readings for this coupon between 2022-09-14 and 2022-09-15.