05/06 - Turkerhub Masters Sunday

Discussion in 'Daily mTurk HITs Threads' started by Randomacts, May 6, 2018.

?

Do you accept your new king?

  1. Yes

    28.4%
  2. No

    71.6%
Thread Status:
Not open for further replies.
  1. Buntastic

    Buntastic Survey Slinger

    Messages:
    12,819
    Gender:
    Female
    Ratings:
    +23,740
    Elizabeth Friedman [A4W6F7C8EKV44] Tell us your shopping opinions! - $0.35 | PANDA
    [3HLRSARVHFHC0331JBCU2Z9WQ4M55Q]


    Fair

    Unrated

    Unrated
    $7.46 / hour
    00:02:49 / completion time
    Pros:

    Cons:

    To read Elizabeth Friedman's full profile check out TurkerView!
     
  2. Kayashi

    Kayashi Titler

    Messages:
    7,285
    Gender:
    Male
    Ratings:
    +14,928
    Title: Write a simple multiple choice question about science | AcceptA
    Requester: Peter Clark [A3D5Y7P08H10BD] Contact
    TV: No Reviews
    TO: [Pay: 3.90] [Fast: 4.80] [Comm: 5.00] [Fair: 5.00] [Reviews: 12] [ToS: 0]
    TO2: [Hrly: ---] [Pen: ---] [Res: ---] [Rec: ---] [Rej: 0] [ToS: 0] [Brk: 0]
    Reward:
    $1.30
    Duration: 8 hours
    Available: 1157
    Description: Write a multiple-choice science question to help us test whether a computer can combine two facts together!
    Requirements: Masters Exists; Total approved HITs GreaterThanOrEqualTo 1000; HIT approval rate (%) GreaterThanOrEqualTo 95; Location EqualTo US
    he bumped that pay a nickel and got rid of the extra work
     
    • Nom Nom Nom! Nom Nom Nom! x 2
  3. treespeaker

    treespeaker Well-Known Turker

    Messages:
    312
    Ratings:
    +1,831
    I thought he changed his requester name to Aristo Datasets?
     
  4. Kayashi

    Kayashi Titler

    Messages:
    7,285
    Gender:
    Male
    Ratings:
    +14,928
    he did and added 50% more work to those hits.
     
  5. marcusdavvid

    marcusdavvid Well-Known Turker

    Messages:
    2,926
    Gender:
    Male
    Ratings:
    +5,187
    You've received a bonus from ofir yakobi for work related to 3R5LWXWHR08UXX3IKVYZZ2Z1TM2GX0.
    The value of your bonus is: $1.62 USD

    The Requester included this note:
    Thank you
     
    • Like Like x 1
  6. rater9

    rater9 Survey Slinger

    Messages:
    2,541
    Gender:
    Female
    Ratings:
    +3,920
    • Nom Nom Nom! Nom Nom Nom! x 2
  7. Rocknrolla

    Rocknrolla Survey Slinger

    Messages:
    10,014
    Gender:
    Male
    Ratings:
    +7,585
    I'm still sitting here thinking of all the people I want to send sugar-free Gummi Bears to
     
    • Like Like x 1
    • LOL LOL x 1
  8. Mothra

    Mothra Survey Slinger

    Messages:
    6,468
    Ratings:
    +11,952
    The previous version, which indicated the angle wasn't parallel, rather than acute, was even more open to interpretation.

    But:

    What part of the book case do you consider to be jutting? The part running roughly perpendicular to the wall, upon which the laptop is resting? Do you mean the taller section positioned closest to the wall?

    To which angle are you referring? As the object is oriented in a euclidean space the sum of any two angles compared to the plane of the wall, which we can consider to be a straight line, must be 180. Unless both angles presented in the image are exactly equal to 90, then one of them must, by definition be acute. Since the sentence mentions an angle, we must assume that an angle exists, otherwise this clearly wouldn't have been selected as an example sentence because there wouldn't be any good reason to insert an ambiguous question designed to cause workers to fail your test.

    Since there appears to be some small amount of space between the tallest portion of the book case and the wall, which permits the casting of the noticeable shadow, there must be some angle there, however acute. If we extend this to consider the longer lower portion, based on the angle of the photo, it's impossible to determine if the book case is 100% perfectly flush against the wall.

    Again, this necessarily indicates that there is the possibility of a non 90 degree angle, which implies one of the angles is acute. Presumably, when this image is compared to the other image pairs, there wouldn't be another grouping with the same two colors of book-cases, with the same color of laptop, permitting this description, even if somewhat incorrect, accurate enough to determine the truth value of the sentence.

    Because the sentence isn't obviously and grossly incorrect, which would seem to be the reasonable standard for a training/qualification test, it seems reasonable to treat it as accurate enough to generate truth value.

    TLDR: Perhaps it would be best if you included several examples in a training document, with yes/no ratings, and explanations as to why you have selected the chosen rating. If an image is ambiguous, as this image appears to be, don't penalize workers for making a judgement call.

    Or, if a single image in your test seems to be tripping up a large number of respondents, send an email to your workers indicating why the previous response was incorrect, to help them learn and provide you with better quality work.
     
    • Like Like x 1
    • LOL LOL x 1
    • Love Love x 1
  9. rater9

    rater9 Survey Slinger

    Messages:
    2,541
    Gender:
    Female
    Ratings:
    +3,920
    • Nom Nom Nom! Nom Nom Nom! x 1
  10. Rocknrolla

    Rocknrolla Survey Slinger

    Messages:
    10,014
    Gender:
    Male
    Ratings:
    +7,585
    She is talking about the other bookcase
     
  11. rater9

    rater9 Survey Slinger

    Messages:
    2,541
    Gender:
    Female
    Ratings:
    +3,920
    Title: Respond to a social science survey [ $ | $ ] PANDA
    Requester: J Reich [A2O1DG4NUD3EOR]
    TurkerView: [ $ / hour ]
    Description: Respond to questions about your perceptions and attitudes towards those with autism
    Duration: 00:15:00
    Available: 1
    Reward: $1.00
    Qualifications:
    • TookStigmaSurvey DoesNotExist [ Request | How to Use ]
    • Masters Exists [ Request | How to Use ]
    • Total approved HITs GreaterThanOrEqualTo 1000,
    • HIT approval rate (%) GreaterThanOrEqualTo 99,
    • Location EqualTo US,
    [39VHG1VU5PT87SPMH9MO7I2J8B67R5]
     
    • Nom Nom Nom! Nom Nom Nom! x 2
  12. LLL

    LLL Survey Slinger

    Messages:
    3,092
    Gender:
    Female
    Ratings:
    +5,994
    Peter Clark
    2:14 PM (2 hours ago)
    [​IMG]
    [​IMG]
    [​IMG]
    to me
    [​IMG]





    Thanks, that’s really helpful, LLL – yes, I think I made a mistake including the validation questions with the authoring questions. My goal was to help people think about what a good question would look like by seeing other good/bad questions that other Turkers had written, but it seems like the constant context switch from validation to authoring and back is a pain for the Turkers. Next time I’ll follow your advice and split it back into two tasks again.



    I really appreciate your feedback, thank you so much – that’s really helpful!



    Best wishes, Pete
     
    • Like Like x 3
    • Nom Nom Nom! Nom Nom Nom! x 1
  13. Girl Polar Bear

    Girl Polar Bear Queen of the North

    Messages:
    29,273
    Gender:
    Female
    Ratings:
    +45,742
    did not read

    [​IMG]
     
  14. LLL

    LLL Survey Slinger

    Messages:
    3,092
    Gender:
    Female
    Ratings:
    +5,994
    l2p.JPG

    Your welcome
     
  15. Girl Polar Bear

    Girl Polar Bear Queen of the North

    Messages:
    29,273
    Gender:
    Female
    Ratings:
    +45,742
    • LOL LOL x 1
  16. Lil Kitchen

    Lil Kitchen Well-Known Turker

    Messages:
    5,192
    Gender:
    Male
    Ratings:
    +12,419
    Greetings from Amazon Mechanical Turk,

    The following is a summary of activity for your Mechanical Turk account for the week ending May 05, 2018.

    Your HIT activity for this week:
    - Number of HITs submitted: 1,320

    Approvals and payments that occurred this week:
    - Number of HITs approved: 1,352
    - Number of HITs rejected: 0
    - Bonus reward earned: $5.35
    - HIT reward earned: $256.08
    - Total Amount earned this week: $261.43
     
    • Like Like x 5
    • 5/5 Pay 5/5 Pay x 4
    • Nom Nom Nom! Nom Nom Nom! x 1
  17. GiMiK

    GiMiK Well-Known Turker

    Messages:
    911
    Gender:
    Male
    Ratings:
    +1,284
    This Hazing HIT

    [​IMG]
     
    • LOL LOL x 1
  18. Buntastic

    Buntastic Survey Slinger

    Messages:
    12,819
    Gender:
    Female
    Ratings:
    +23,740
    Nope, correct one.
     
  19. LLL

    LLL Survey Slinger

    Messages:
    3,092
    Gender:
    Female
    Ratings:
    +5,994

    He's not rude just Scottish. Offer him a pint and he'll be your best friend!
     
    • LOL LOL x 3
  20. Sunlite

    Sunlite Survey Slinger

    Messages:
    5,028
    Gender:
    Female
    Ratings:
    +7,755

    This^^^
    Plus, show a pic for your rules and why this fits or doesn't. If the object is to get people to qualify by understanding what you want them to do or how you want them to look at something, show them. Isn't that what an example is for?
    I went to the PDF which was no help. Where there are 2 pics of the same animal- is that considered 2 animals (1 in each pic) or 1 animal (because it is obviously the same)? Why use the same animal- use a pic of 2 different animals.
    This has taken up too much time and I think maybe you should go with the people who understood what you were trying to say.

    :dunno:
     
Thread Status:
Not open for further replies.