Now that you have worked with a participant to understand their perspective, it's time to structure that information into useful data. 

 

Video TRanscription

STRUCTURING THE TRANSCRIPTION

OPTION 1: IN-HOUSE ANALYSIS OF BIOMETRIC DATA

Option 2: ASSISTANCE WITH BIOMETRIC DATA ANALYSIS

Exporting GPS DATA to Json

 

VIDEO TRANSCRIPTION

Before you can begin to dive into qualitative analysis, you’ll want to have a transcript of your interview and walkabout. For this process, we recommend using Mechanical Turks for a first pass. 

Start by creating an account through Amazon. 

 

Select a New Project by highlighting "Transcription from A/V".

Name and describe the project as accurately as possible. This will help you with organization in the future. 

Set a price for how much you would like to a pay a worker per the time allotted to the project. We recommend not going below $15 and allowing around two hours for each 15 minutes of video you have.

 

Mechanical Turk has a great template for video transcription, and you'll like find that you do not need to add any additional instructions. 

 

Add in your video link to the field that asks for a link. 

 

Lastly, you'll need to create a CSV file that the workers will upload their data to. If you have experience with Mechanical Turks, create a custom file that includes fields for timestamps and text. Otherwise, feel free to use the sample CSV file they provide. You'll extract the data later. 

 

STRUCTURING THE TRANSCRIPTION

Once you've received your transcription, either through Mechanical Turks or another source, you will want to structure the data so it can be entered into the Civic Viewer page. For this purpose, we recommend using Excel. 

  • Set up five columns, including:
    • Timestamp
    • Absolute Time
    • Interviewer
    • Participant
    • Subject Keywords
  • Copy and paste Timestamp, Interviewer, and Participant from your Mechanical Turks transcription file. The Interviewer and Participant information might be listed as Person 1 and Person 2, Check to make sure you're pulling the correct text. 
  • Now you will create the Absolute Timestamp, by adding the Timestamp time to the start time of the interview. For example, if you started the interview at exactly 12:25pm then when the Timestamp reads 1:06 minutes, the Absolute Timestamp will read 12:26:06pm. 

The Absolute Timestamp is a pivotal piece of data. It connects the qualitative data to the quantitative data points, like GPS location. By creating the Absolute Timestamp, we can read what the participant said when they were in a specific location, or feeling anxiety. 

 

Video Upload

YouTube is a great option for video analysis. It’s a free service that will host large video files. But, more importantly, we can access specific moments of time in the video by using a bit of code and YouTube’s sharing capabilities. 

To get your video on YouTube, either log in or create an account through Google.

Click the "Upload" button in the top right corner. From here, simply drag your GoPro footage file from the GoPro folder where it was saved, onto the screen. 

Depending on the length of the video, the file will need a few minutes to both upload to YouTube's servers and then be processed for display on the site. 

Once your video is ready to view,all you need is the shareable URL. You can find this in the dropdown menu located under the video player. 

(Optional Privacy Considerations)

  • To keep your video private on the Civic Viewer site, you'll need to configure the settings on YouTube. Click on Pencil Icon, click on privacy options on the right and select unlisted. You may also choose private, though you’ll need to add the email of anyone working with the data to the video on YouTube. 

 


OPTION 1: IN-HOUSE ANALYSIS OF BIOMETRIC DATA

The most experimental element of Civic View is the use of biometric data. There are many points in the process where the data can become skewed, including the analysis of that data. 

Keep in mind, that this analysis method is for estimation only. 

  • To start, create a spreadsheet (we recommend Excel) with the following columns: 
    • Absolute Time
    • Time Difference
    • BIT_Time
    • EDA Reading (Or EKG Reading, depending on the type of data you're using)
    • Icon
  • Open up your biometric data file in OpenSignals. The file name will have the exact date and time the device began recording data, and should look something like this: 
opensignals_file_2015-03-29_12-39-03
  • Load the file and isolate the channel that shows only the data stream you collected. 
You can hide streams of data, by clicking on the downward-facing arrow on the right side of each channel.

You can hide streams of data, by clicking on the downward-facing arrow on the right side of each channel.

Your screen should look something like this:

From here you can start the process of estimating the biometric readings. You're looking for three shapes: 

  • Flat lines, or areas with only relatively slight changes. This is your baseline. 
  • Peaks, or areas where the data spikes above the baseline.
  • Valleys, or areas where the data dips below the baseline. 

Start scanning the data slowly from left to right. Whenever the data shifts up or down, or returns to baseline, mark the time and either Mid (baseline), High (peaks), or Low (valleys), in your spreadsheet. Your data will begin to look like this:

Lastly, you will assign an icon to each category. While not necessary for the data, the icons are used in the viewer to show changes to the data with a simple illustration. 

In the column marked Icon, you can assign icons by entering in URL addresses. The Viewer will read these URLS and display the images. 

  • Low = http://brandondoman.com/sarah/Icons/Biometric_low.png
  • Mid = http://brandondoman.com/sarah/Icons/Biometric_mid.png
  • High = http://brandondoman.com/sarah/Icons/Biometric_high.png

 


OPTION 2: ASSISTANCE WITH BIOMETRIC DATA ANALYSIS

If you chose not to run analysis, or don’t have the resources on staff to do so, this data can still be useful. However, you’ll want to make one adjustment to the way the data is recorded by the BITalino. 

  • Switch the data output from .H5 to a .txt file before you go out into the field with a participant. 
opensignalsfiletype

As a .txt file the data is far easier to read, and whomever you recruit to help you analyze the data will be able to view the readings without downloading the OpenSignals software. Instead they can use a program like MatLab, or their analysis software of choice. 

 


EXPORTING GPS DATA TO JSON

Moves uses a system called Connected Apps to help process the GPS data it collects. All of these apps process the data for particular reasons, like integrating activity data with a nutritional diary such as My Fitness Pal. We want to use the simplest form of data, so we recommend using the Moves raw JSON export app.

Once you have landed on the page, click the “Authenticate at Moves” button. You’ll be asked to authenticate your Moves program.

authenticatemoves

Follow the instructions and you’ll be taken to the main Export screen. 

  • You can either use the API-Key provided at the top (our system bypasses this step for now), or scroll down to the Settings section
  • Using the format the app has suggested, enter in the date of your field work and click Start Export. Once the data has been generated, click Download JSON file at the bottom of the screen. 
  • You may also copy and paste the generated data into a new file, but we avoid this step in case of mistakes.