having access to dataset of communication in crowdsourcing domain

Dataset for data analysis

Hello everybody, I am doing a research on crowdsourcing with the focus on the existing problems of communication between workers and requesters.

I need a dataset containing the history of their communications and interactions (e.g., emails, messages, etc.)

Do you think that I have any chance to have access to such data for research purposes? if yes, would you please tell me how I can gain such dataset?

Thank you very much in advance.

 

There is no direct contact portal between requesters/workers on MTurk so anything you get is going to be some really funky like email across clients/etc. I'm not sure you can just purchase a standardized dataset for this kind of thing (unlike say communication w/ an at-home assistant like Alexa which is generally cleanly logged & available for upload).

You can ask workers to sell you their communications I imagine, but I have zero insight into what kind of laws that might be violating given you're unlikely to get consent from both parties to the conversation or if those kinds of things even apply to emails/etc.

Only really straightforward way I can see of this working is interview format w/ workers & requesters separately, I cannot fathom a simple way of collecting hard data of communication between two parties that you aren't privy to but to be completely honest I am wholly unqualified to really speak to any of that with authority.

 

Thank you so much for your reply.
would you please explain a bit more what are the possible or common ways that workers get aware of the result of the evaluation process and how they understand the reason for e.g. rejection of their work?
or if they have any questions regarding the tasks or the requirements of the task, how do they contact the requesters?
or do you know any specific problems regarding their communication style or quality?
I appreciate your time and consideration.

 

On the actual platform itself, when a requester rejects, or approves a hit, there is a feedback box which the requester can type in a reason for a rejection or any other pertinent feedback:

Turk offers a native way to contact requesters once you've submitted a hit for them in the past 45 days:



But if a worker hasn't submitted a hit for them, you need to find the contact link and input the requester ID of the requester you're trying to contact. Links for this are often found on sites like TV and TO. (on the worker end, it comes up an error when you send that message, but the message does go through)



Emailing is the primary way for a worker to contact a requester with questions.

 

Thank you very much for the very detailed explanation.

Could you estimate or give a rough percentage of usage of the inner portal for communication between both parties?

or have you ever heard workers complaining about the requesters that they do not respond to emails or they do not give reasons for the rejections?

I know that I am asking many questions, but I would appreciate it if you ave time and answer these questions. I would like to understand current state in general.

 

Almost all communication is initiated through the portal at first. Sometimes requesters include their personal / business emails into HITs but its rare. Sometimes workers bypass portal communication for IRBs. Both are probably a small portion of overall communication.

Sure, people complain about lots of things.

One thing to keep in mind is this place isn't really representative of the average MTurk experience. Even the least experienced users here trend toward the most active on the platform because thats what forums tend to specialize in.

You'll need to create actual HITs on the platform if you want a "general" experience poll.