4

I am doing my graduation project to become a Software Engineer. The project aims to study the behavior of freelance developers to subsequently design a process that allows more quality and efficiency.

To know the freelancer behavior I have made a survey, but I don't know what is the best way to publish the survey and get responses; after which I'll post an article.

I have published the survey in at least 7 forums, but in most of them my post has been closed for spamming.

Can any one tell me some advice for doing the survey? (on internet)

  • 3
    I don't have any experience with this, but you might look to your institution's social sciences departments (psychology, sociology, economics, education, etc). A lot of research in those areas involves surveys, so faculty and grad students in those areas may have helpful experience for you. Don't forget that since your survey participants are "human subjects", this whole process probably requires approval from your institution's IRB or a similar ethics panel. – Nate Eldredge May 7 '14 at 20:46
  • This question appears to be off-topic because it is not about academia. Cross Validated might be able to help, but the question may be too low quality for them. – 410 gone May 8 '14 at 6:05
  • 1
    I would image questions on "how to get people to take my survey" would be completely off topic on a statistics site. CogSci does appear to have a survey tag, though, it might fit there. – ff524 May 8 '14 at 6:30
  • 5
    @EnergyNumbers I would argue that this is in fact a rather general question about academia. I certainly fits better here than at CrossValidated. I'm voting for leave open. – xLeitix May 8 '14 at 7:06
  • Your target are freelance developers, I think you can use your existing network and I guess you must know at least one of them. Start from there. – bingung May 8 '14 at 8:11
6

If you're using forums, don't just make a thread and then write:

Class project, please fill out this survey. Link: www.xxxxxx984asdfac. Thx!!

Most of the time these annoying threads can trigger shutting down or be simply ignored as spam.

I'd recommend picking a few major forums and actually start a communication with the administrators. In no more than a few hundred words explain who you are, what questions you're trying to answer, and how your results may help both parties. If the outcome is positive, here are a bunch of wishes you can usually ask for:

  1. Pin your thread at the top for a fixed amount of time to enhance visibility.
  2. Tap into their membership e-mail or online messaging network for an announcement and about 2 waves of weekly reminders.
  3. Provide you with important statistics including open rate, click rate, bounce rate, initiation rate, and completion rate. These are all necessary if you are to publish your results in an academic journal.

Try to enhance your credibility by, in the mail to the administrator and the message to the users:

  1. Detailing the motive of your survey, what kind of questions respondents should anticipate, and how long would it take.
  2. Getting an institutional review board approval, and report so in the announcement.
  3. Explicitly mentioning the measures to ensure respondents' confidentiality and/or anonymity.
  4. Explaining how the results will be used, and even be fed back to the communities that have helped you filling in the survey.
  5. Being real and present by providing your official contact information (names and school e-mail are sufficient). Invite respondents to contact you if they have questions about the study. Check with your supervisor and department for policy of using the school's name.
  6. Clearly stating the opening period of the survey.

You can also try contacting companies that hire freelancers, or talk to websites that organize freelance jobs.

I'd also suggest to make duplicates of your survey so that each forum will have their own referral link. You can pool the data later and compare the differences between forum. If you just make one survey you may not be able to figure that out at the end. Putting a "where do you learn about this survey?" question may also help, though probably not as objective.

Now, if you don't care about knowing all those rates, and just want a quick turn around of data, you may use your network to contact some high profile people in the field and see if they are willing to broadcast that for you. Most of the time, a tweet from them would be quite effective.

I also agree with the comments asking you to pay a visit to some social science faculty. At the very least, make sure your questions are examined by some professional for validity. Most researchers also do not mind share with you some templates of the survey introduction.

Some other technical strayed thoughts:

  1. Do make a copy of online survey for each major source. Aka forum A will have its own link, and forum B will have its own link, etc. That way later you can stratify them in the analysis.
  2. Invest in a good online survey service. Make sure not to enroll in those with limited responses. It'd be very annoying for the 501st respondent to get a "quota full" warning if your free service account only allows 500 responses.
  3. Write good questions and include only relevant questions. Again, talking to some researchers will help a lot.
  4. The Checklist for Reporting Results of Internet E-Surveys (CHERRIES) provides a good list of components you should include in manuscript. Read them before you start data collection and make sure you have thought about all these factors.

Lastly, check if someone has already invented the wheel for you. Secondary data sets may have already covered information you wish to collect. For example National Longitudinal Surveys maintained by the US Department of Labor (I just went ahead and assume that you're interested in the US market) does ask question about freelancing. They even once released a report about this.

| improve this answer | |
  • I've never heard of a (reasonable) online survey being conducted through a list-serve like you are proposing. I've known people to be able to get email lists from professional organizations and then send emails directly. Posting a link publicly online is fraught with difficulty, at least sending a unique link through email potentially filters the results. (Either are both highly suspect to selection bias in who answers - especially since response rates tend to be quite low.) – Andy W May 8 '14 at 14:05
  • Hi @AndyW! It's been a while and thanks for the comments! At this moment I'm actually analyzing one that was done through a patient-support group website. As you mentioned the response rates are indeed saddening (11-15%). As for selection bias, I really have no better way to tackle this other than suggesting reporting all the rates and anticipate how the biases would have changed the estimates. As usual, please improve my answer as you see fit. Thanks. – Penguin_Knight May 8 '14 at 14:17
  • Over 10% is good! For email ones I am familiar with the norm was closer to 5%. My response was mainly in regards to I see (on regular occasion) people send out surveys willy-nilly across the internet to whomever will listen (post on personal blogs, to list-serves, etc.) These are not reasonable ways to conduct research, and seem often the hasty attempt to gather some data - even if the data is meaningless. This answer demonstrates the further thought that needs to go into the endeavor - I'm concerned though about how many people will make that effort or realize the bias in their findings. – Andy W May 8 '14 at 14:31
  • I agree. Freelancers who frequent the forum or looking for jobs on a freelance websites are probably different kinds of people. The really successful ones could have been working and have no time to mingle online. Non-sequitur: I also had another occasion in which we set up a gift card lottery... 25% of the responses were all empty (they flew through with the skip button) except with an e-mail address to enter the lottery. Online surveys are full of amusements. – Penguin_Knight May 8 '14 at 14:47
  • For in person surveys of deviant behavior I've seen instruments place fake questions that will potentially trawl out people just going through the motions and filling out the survey. (One was asking about recreational drug use, and they made up a fake drug - as some teenagers would say they used everything all the time!) The pure empty responses are easy to spot - but people may just pick random answers to the same end goal (if you have measures of the time it took to complete the survey this would be a good check as well). – Andy W May 8 '14 at 14:54
2

To borrow from Willie Sutton:

Why do I rob banks? Because that's where the money is.

You need to put your surveys where the people are. I've seen a few studies that used Facebook, which lets you specify exactly which demographic should see your survey. SurveyMonkey can also help put your survey in front of the people you'd like to see. I know that I had a lot of success putting up advertisements across campus offering to pay for participation, but that was for a study that involved more than just a single survey.

I would also suggest what was said in the comments... talk to the local economics, psychology, and sociology departments to see if they can give you any assistance with this. Researchers in all those areas will face this problem all the time.

| improve this answer | |
  • @JuanCamiloCerón Sure! If this answer helped you, don't forget to vote it up! – eykanal May 9 '14 at 21:52

Not the answer you're looking for? Browse other questions tagged or ask your own question.