Wave 2 completes

Geo-locatoin

Lat-Long: ex-US

ex-US response:
- 5 Nigreria

US map

Declared state of residence and lat-long position

Response farms will many repeated lat-long positions with varrying declared state of residence

Repeat Lat-Long.

  • 21 response from the same Lat-Long in rural Kansas (free text is short & similar) in WAVE 1

Sample Demographics:

Gender:

Gender Freq % target
Woman 2028 59.0 53-54
Man 1378 40.1 46-47
Other 29 0.8 < 5

Overview:

When creating diagnostic variables, we focused on a few qualities of the data.
- The length of free-text responses.
- Inconsistent or impossible combinations of responses.
- Location provided by latitude and longitude Some suspicious responses we observed and can consider when flagging but cannot serve as diagnostic of fraudulent data: C-suite jobs

Types of bad respsonse:

  • poor quality humans: too fast; inconsistent answers to pairs of questions; weird text from external sources, speach to text
  • fraudulent
    • mass responses from the same source: same lat-long or IP host & similar text repsonses.
    • machine generated responses: odd text; inconsistent answers to pairs of questions.

Text responses received:

  • R_5JOVVlUisllqQO0: ai language model, designed to assist with various tasks and inquiries.

Possible AI gen text:

RW generated AI text:

biggest_influence prompt: write a response to this survey question: Overall, what would you say was the biggest influence on your initial career choice?

Determination of bad response:

Age @ college degree (age - graduate)

  • age: What is your current age?
  • graduate: In what year did you complete your Bachelor’s degree?

Advance degree at too young an age: (PhD OR MD) AND age<28.

doogie howser check

Work_exp - job_tenure: work experience beyond current job

  • work_exp: How many years of post-college work experience do you have?
  • job_tenure: For how many years have you been employed by your current employer? Please provide the total number of years since you first started working for this employer, even if you have been promoted or changed jobs with this same employer since then.

fix work experience for users that entered calendar year
mutate(work_exp_fix = case_when(work_exp>1000 ~ 2024-work_exp, work_exp<1000 ~ work_exp) )
fix job tenure for users that entered calendar year
mutate(job_tenure_fix = case_when(job_tenure == 99 ~ 0, job_tenure >= 66 ~ NA, job_tenure < 66 ~ job_tenure) )

total work_exp must be >= job_tenure

Age-work_exp: age at first work experience

work_exp: How many years of post-college work experience do you have?

Negative values are impossible
Minimum should be around 20 years

College begin and graduation and age

Duration - too fast: 40% of median time of good completes

Jobs count & Employer ct