Analyze data that was collected by a group of people wearing bodycams

All participants wear body cams, to record their field of view and activities. This allows for detailed post analysis after your tests are done.

Challenge & Goal Summary

Only such recordings allow you to understand activities and decisions made, because you have a recorded reality which can be played back repeatedly. That way you can understand the overall context for each participant and see every single detail which might be very important for your final research results.

This setup can be applied to any research that requires similar ways of data collection. Possible application areas are flight attendants, police officers, nurses etc.

The biggest challenge we face here is:

  • How can the independently recorded videos be synchronized for combined analysis?

The goal is:

  • Analyze the collected information per camera view, delivering frequencies and durations per activity.
  • Perform a sequence analysis to identify typical reactions.
  • Identify latencies between two specific activities.
  • Find co-occurrences of activities captured by different cameras.
That’s how you can do it

Study Design

Equip each participant with a body cam, so that you reliably record the necessary and interesting scenes. Make sure all participants wear a digital watch (that includes seconds) that can be used for synchronization. For synchronization purposes, it is important that the body cams record throughout the whole time. Synchronizing video snippets that were recorded in different locations is too complicated for a general explanation.

Data Collection

Ask your participants to synchronize their watches exactly before the test starts. Once all watches are in sync, ask everyone to start his/her recording device and hold the watch into the camera’s field of view for at least 5 seconds. This information allows you to synchronize all recordings afterwards, before you start with your observations.

Prepare Data Logging

When your test is finished and all recordings were stopped, copy all video files onto your computer’s hard disk – best all within the same folder. Make sure each video file is named individually, so it reflects the camera it was recorded with.

Open all videos simultaneously in INTERACT and synchronize them, based on the time information of the watch at the start of your recordings. To store this information, create a DataSet named 'Synched Videos' or something and link all videos, including their synchronized time information, to this DataSet.

To individually log all activities and make transcriptions per camera perspective, it is best to use a separate DataSet for each video during your observations. That is why you should create as many DataSets as you have video files.
To spread the workload over multiple coders, create a single data file for each video. To make sure everybody works with the same time information, predefine the file references per Set and video:

Simply copy the corresponding reference from your 'Synched Videos' DataSet and paste it into the transcription area of the DataSet for one video. Only then, your observations will be synchronized across all recordings during analysis!

Specify at least one Independent variable per Set that holds the camera number and maybe add variables for gender, name or function of the wearer of the body cam, if this information might be important for statistics and data analysis later.

Collect Observations

Open the prepared data file and double click the time value of your Set to open the video that was linked to that Set. This ensures that the required synchronized time information is known by INTERACT. Do NOT open your video manually!

If your focus is the conversation structures of the various participants, you can start by setting-up at least two sets of Codes that build a lexical chain, to identify 'Speaker' and 'Receiver' and turn those into so-called 'Comment'-Codes.

This allows you to easily log the duration of each speaker turn, just with your SPACEBAR!

A separate transcription dialog pops up, enabling you to transcribe whatever is said and while this dialog is open, you can re-play the just captured sequence again and again, to make sure you got it all right.

In a second run through the same video, you can collect information like 'Location' (for example 'Pantry', 'Gangway', 'Cockpit', 'Stairs', etc.), 'Flight Phase' (for example ‘Take-off', 'Airborne', 'Approach', 'Landing', 'Taxiing', etc., 'Activities’ (like 'welcoming guests', 'safety instructions', 'serving food’, 'serving beverages', 'Individual assistance', etc.) and anything else you may think of. Using mutually exclusive codes for each of those situations, simplifies and speeds up the coding process tremendously, because you will only need to log the start of a behavior.

Analysis

Collecting all information mentioned, in one or multiple passes, using one or more different Code Definitions, provides you with a lot of information to be analyzed. If you collected your observations in a separate document per video, make sure you label each Set in a way that enables you to identify the data inside without the file name.

Merge all documents into one big compilation file before you continue. Keep the original files at all times.

No matter in what order you collected the data, at the end the INTERACT Time Line Chart will put everything in a chronological perspective and provides you with all descriptive statistics like frequency, duration, latency, percentage over time and many more, with a single click. All split per Set = per perspective (video). But this is only where the fun part starts!

With the Text Analysis routine, you get a detailed word count over all transcriptions, split per DataSet or for just the selected Events from a specific person. This allows you to compare the vocabulary used by different flight attendants or other participants.

Cleverly combining the various routines for finding co-occurrences and contingencies, enables you to answer questions like "What happened during the Flight Phase 'Prepare for landing?" or "Identify all instances where a flight attendant was speaking with a college while serving food" as well as "Measure the time between Take-off and serving drinks".

These numerous possibilities are what truly helps you with your study.

 

Your advantage with INTERACT

The biggest advantage of INTERACT is its ability to re-organize, rename and shuffle collected data, should you notice that your initial structure is not perfect to find the answers you are looking for.

INTERACT