|
Field
Intervention (Hidden)
|
Before
We are fielding a choice-based survey of individuals on Mechanical Turk. Each survey taker will be given 3 hypothetical choice scenarios, each consisting of two vacancies (Blank Co and Brown Co). For each comparison, the participant is asked ‘which company do you value applying to most’?
The options differ in several characteristics:
A) The name of the company (Blank Co, Brown Co). The first option is always Blank Co.
B) The hourly wage. For the first comparison, this is $18 for Blank Co and $20 for Brown Co. For the second comparison it is $19 for Blank Co and $21 for Brown Co. For the third comparison it is $22 for Blank Co and $24 for Brown Co.
C) The current number of applicants. For each comparison, there are two preset values and these are allocated with 50/50 probability to Blank Co and Brown Co or vice versa. For the first comparison the values are 5 - 20 and 200+, for the second comparison 200+ and 0-4, and for the third comparison 0-4 and 5-20.
D) Whether an additional line in green is shown. Whether this line is shown is randomize at the participant level with a 50% probability. This line states that ‘AI probability you get an offer: X%’ and is identical across options within a choice, but differs across choices. For choice one, this probability is 25%, for choice two it is 13%, and for choice three it is 30%. If participants see this line the instructions also state:
“Our job search engine uses an artificial intelligence (AI) algorithm that uses all information to predict whether you would receive an offer if you applied. This probability is also included in the job description.”
We also ask an open ended question about each choice after all choices are made. The prompt displays the information about the choice and asks the following, in comparison (1, 2, 3), you picked (Blank Co, No Preference, Brown Co). Please explain why you made this choice?
Lastly, we ask participants whether they responded randomly and if they have any comments about the survey.
|
After
We are fielding a choice-based survey of individuals on Mechanical Turk. Each survey taker will be given 3 hypothetical choice scenarios, each consisting of two vacancies (Blank Co and Brown Co). For each comparison, the participant is asked ‘which company do you value applying to most’?
The options differ in several characteristics:
A) The name of the company (Blank Co, Brown Co). The first option is always Blank Co.
B) The hourly wage. For the first comparison, this is $18 for Blank Co and $20 for Brown Co. For the second comparison it is $19 for Blank Co and $21 for Brown Co. For the third comparison it is $22 for Blank Co and $24 for Brown Co.
C) The current number of applicants. For each comparison, there are two preset values and these are allocated with 50/50 probability to Blank Co and Brown Co or vice versa. For the first comparison the values are 5 - 20 and 200+, for the second comparison 200+ and 0-4, and for the third comparison 0-4 and 5-20.
D) Whether an additional line in green is shown. Whether this line is shown is randomize at the participant level with a 50% probability. This line states that ‘AI probability you get an offer: X%’ and is identical across options within a choice, but differs across choices. For choice one, this probability is 25%, for choice two it is 13%, and for choice three it is 30%. If participants see this line the instructions also state:
“Our job search engine uses an artificial intelligence (AI) algorithm that uses all information to predict whether you would receive an offer if you applied. This probability is also included in the job description.”
We also ask an open ended question about each choice after all choices are made. The prompt displays the information about the choice and asks the following, in comparison (1, 2, 3), you picked (Blank Co, No Preference, Brown Co). Please explain why you made this choice?
Lastly, we ask participants whether they responded randomly and if they have any comments about the survey.
An additional study we will run will in December of 2022 will exclude the AI component and will also slightly modify the question asked of the participants. This question will state the following about the comparison:
Imagine that you are currently unemployed and are searching for a job:
- Each job requires you to work for 40 hours a week.
- Each job was posted at the same time.
- You would take any of the jobs if it was the only offer you got.
For each job, please state whether you would value the ability to apply for Blank Co or Brown Co more.
Note that in the summer version of the study, we had an additional line:
'Each job is identical except for the information presented to you'
|