Changelist: Pictopercept Jobs Pilot

[saurabh-khanna]

Here is a list of changes to be made in Pictopercept Jobs before we launch.

Sample country: US (representative sample on Prolific)
Sample size: 300
Survey duration: Roughly 5 minutes (2-minute regular questions + 3-minute image interaction)

Test URL (we can keep updating this version based on feedback): https://pictopercept.theinvisiblelab.src.surf-hosted.nl/survey/occupations

[saurabh-khanna]

What images to use

• We will use randomly selected images from 597 images in the Chicago Face Database. The images to use are in Images > CFD
• Understand the image naming convention. CFD-AF-200-228-N.jpg refers to the image of an Asian Female, with model ID 200, image ID 228, and a Neutral expression.
• Within each folder, let us use the face with a Neutral expression because every image folder has one such image.

[saurabh-khanna]

List of Jobs to use

• Eight jobs to use. Since we have 64 questions, each job will appear 8 times.

• A Chief Executive Officer (CEO)

  • High‐status, business leadership (often male‐coded)

• A Computer Programmer

  • STEM/tech (often male‐coded)

• A Doctor

  • Medical/STEM (historically male‐coded)

• A Nurse

  • Caregiving/medical (strongly female‐coded)

• A Primary School Teacher

  • Education/caregiver (often female‐coded)

• A Police Officer

  • Protective/service role (often male‐coded)

• A Housekeeper

  • Domestic/custodial role (often female‐coded)

• A Construction Worker

  • Manual labor (strongly male‐coded)

---

Resoning:

• STEM vs. Non‐STEM: Computer Programmer/Doctor (STEM) vs. Housekeeper/Teacher (non‐STEM)
• Gender stereotypes: Nurse/Housekeeper/Teacher (commonly female‐coded) vs. CEO/Police Officer/Construction Worker (commonly male‐coded)
• Prestige & manual vs. intellectual: CEO/Doctor (high prestige, white‐collar) vs. Construction Worker/Housekeeper (manual/service roles)
• Breadth in a short set: Eight roles are enough to pick up on a wide variety of biases (e.g., around gender, authority, “caring” vs. “technical” work) without making the survey too long

@dnjm02 @pffeelgood @JiayiYYY thoughts?

[saurabh-khanna]

Order of images

• Gender in CFD takes 2 values (M or F)
• Ethnicity in CFD takes 4 values (A or B or L or W)
• Each participant will see a total of 70 image pairs:
  • 8 * 8 = 64 images pairs, so that each possible combinations of images is seen by each participant
  • 6 image pairs are attention checks.
    • These will a randomly chosen 6 pairs (out of the above 64)
    • These 6 pairs are shown a second time. We can show
    • On the second showing, the left and right images are interchanged.

[saurabh-khanna]

Timer

• Show timer to everyone
• Timer duration is 3 seconds

[saurabh-khanna]

Regular Questions Randomized Before/After Images

• We will have a small section with a few regular survey questions. Questions being developed here: Traditional survey questions #2
• This section comes before the image selection part for 50% randomly chosen participants, and after the image selection part for the other half.

[saurabh-khanna]

Saving Survey ID

• We will recruit participants from prolific.com. They will be given a URL to come to our survey platform.
• Prolific ID of participants from prolific.com can be saved using URL parameters. See here.

[saurabh-khanna]

Script to fetch results

• We need a simple Python jupyter notebook/script to fetch results and view them all in a single pandas dataframe

[pffeelgood]

List of Jobs to use

• • [ ]
    [ ] Eight jobs to use. Since we have 64 questions, each job will appear 8 times.* A Chief Executive Officer (CEO)

    • High‐status, business leadership (often male‐coded)* A Computer Programmer
    • STEM/tech (often male‐coded)* A Doctor
    • Medical/STEM (historically male‐coded)* A Nurse
    • Caregiving/medical (strongly female‐coded)* A Primary School Teacher
    • Education/caregiver (often female‐coded)* A Police Officer
    • Protective/service role (often male‐coded)* A Housekeeper
    • Domestic/custodial role (often female‐coded)* A Construction Worker
    • Manual labor (strongly male‐coded)

@saurabh-khanna @dnjm02 perhaps we change "police officer" to "security agent"? I feel like a police officer actually conveys a sense of authority for me, instead of a service role (in the Western world), and especially so if we are doing the pilot for the US populations.
Or alternatively, change the description of "protective/service" for police officer to "authority/ high-status" (for that we then hypothesize this occupation is more white-dominated), if we prefer to include police officer as a choice in the pilot.