A friend of mine works for a retailer that is heavily invested in the surveys that they get back from customers. Anything less than 10's in every category is a complete fail. Some of the employees are sending the surveys to emails that they created and they answer the surveys on their iphones. They reset the iphone every time to create a new IP. Is it possible or likely that they (the survey company) are also able to see the device ID on an iphone? They claim that they (survey company) track devices/computers, but I know for a fact that some of these people have taken dozens of them with no negative result. Any insight would be greatly appreciated.
Can a survey company track an apple device ID?
- Thread starter eyerish
- Start date
Since you're looking for an Apple device ID this mysterious retailer is probably...Apple?
There are a lot of companies in AM that can pick up on devices, so I'd say a monster company like Apple would be able to.
There are a lot of companies in AM that can pick up on devices, so I'd say a monster company like Apple would be able to.
Obviously apple can track anything they want on their own devices....
No...it is a third party company that provides customer feedback for retailers in order to track customer service metrics, such as facility appearance, attentiveness of the salesman, etc. The customer contact information is collected by the retailer and then forwarded on to the company that sends the survey out. JD Powers uses these survey results to award the best retailers, etc.
To ask the question another way: If I collect data (leads, etc) through various marketing methods, can I track a unique device ID (specifically Jim's iphone for example) on my website to combat fraud? Does apple block my ability to do so or is there a way to do it?
No...it is a third party company that provides customer feedback for retailers in order to track customer service metrics, such as facility appearance, attentiveness of the salesman, etc. The customer contact information is collected by the retailer and then forwarded on to the company that sends the survey out. JD Powers uses these survey results to award the best retailers, etc.
To ask the question another way: If I collect data (leads, etc) through various marketing methods, can I track a unique device ID (specifically Jim's iphone for example) on my website to combat fraud? Does apple block my ability to do so or is there a way to do it?
So.. the surveys are done on a web page?
As for a cursory research, it is not possible to track a unique device ID, it is simply not open to the browser (Javascript/HTML)
Most work done now is on persistent cookies (which would not change when you change the IP), or probabilistic device recognition (basically, trying to GUESS which device it is by usage patterns), and good old session IDs.
But to be blunt:
This sounds like you (or the employees) are trying to game the system.
A slightly better approach would be to school the client on some basic statistics (Bell curve, statistical improbability of all 10s, survey answer bias male/female, etc..)
Some links for further reading:
Very good Probabilistic fingerprinting:
https://github.com/Valve/fingerprintJS
And the UUID being OPEN to apps:
http://docs.phonegap.com/en/3.0.0/cordova_device_device.md.html#device.uuid
This is applications developed through cordova, not open to the browser.
And a tricky way to get the UUID through Safari (the user still has to click to a fake install)
http://www.joshwright.com/tips/Getting-an-iPhone-UDID-from-Mobile-Safari
::emp::
As for a cursory research, it is not possible to track a unique device ID, it is simply not open to the browser (Javascript/HTML)
Most work done now is on persistent cookies (which would not change when you change the IP), or probabilistic device recognition (basically, trying to GUESS which device it is by usage patterns), and good old session IDs.
But to be blunt:
This sounds like you (or the employees) are trying to game the system.
A slightly better approach would be to school the client on some basic statistics (Bell curve, statistical improbability of all 10s, survey answer bias male/female, etc..)
Some links for further reading:
Very good Probabilistic fingerprinting:
https://github.com/Valve/fingerprintJS
And the UUID being OPEN to apps:
http://docs.phonegap.com/en/3.0.0/cordova_device_device.md.html#device.uuid
This is applications developed through cordova, not open to the browser.
And a tricky way to get the UUID through Safari (the user still has to click to a fake install)
http://www.joshwright.com/tips/Getting-an-iPhone-UDID-from-Mobile-Safari
::emp::
hey, emp, do you know a lot about surveying? Can you post an article on that blog of yours? Actually, simply a resource list would suffice.
ps: I'm interested in the basics as well, but I think with a little bit of digging I can find that myself. What I'm particularly interested is methodologies in designing surveys with data manipulation in mind, and the process of data manipulation itself. Pandas, scikit preferred, but stata, even R would suffice too.
ps: I'm interested in the basics as well, but I think with a little bit of digging I can find that myself. What I'm particularly interested is methodologies in designing surveys with data manipulation in mind, and the process of data manipulation itself. Pandas, scikit preferred, but stata, even R would suffice too.
No. But I see how you can jump to that conclusion. Another way to look at it would be a reader voicing his (common) painpoint, giving you an idea for a blog post.
I can't reallistically sit and read all of the answers in the survey. 2k with 3 or more open ended questions and another 4 or 5 closed ones with 4 to 5 options, to help me segment what I'm really after (the open ended ones).
The only way I can imagine this happening is with statistical packages and machine learning (for processing the open ended questions).
But as with anything of this scale (and precision), I am sure there are many best practices and common errors/mistakes to avoid (in the coding, particularly). <- this is what I asked for. Anyway, I'm sure amazon will have plenty of stuff to help me out, I was just hoping to have an easier time sifting through the bullshit.
Ah.. well, bt your question is faaaar away from the main topics of my blog.
Anyhow.
I used to work in market research.
A LOT of manpower (cheap students) goes into open ended questions.
Here is the breakdown:
Say you got 10K surveys back
- Students read the first 1K
- Answers get grouped into categories
"Tastes like poo" "Tastes like old woman under the feet" --> 1.1 Bad taste
"Like an angel singing" --> 1.2 Good taste
..
etc..
This step is called "pre code"
Then all 10K surveys get read and coded --> "Code"
Lists are kept with answers that don't fit.
If enough "samesies" are collected, they get their own code.
If not, they get collected for presenting as outliers/amusing / getting dropped.
Then you can do your stats as usual.
::emp::
Anyhow.
I used to work in market research.
A LOT of manpower (cheap students) goes into open ended questions.
Here is the breakdown:
Say you got 10K surveys back
- Students read the first 1K
- Answers get grouped into categories
"Tastes like poo" "Tastes like old woman under the feet" --> 1.1 Bad taste
"Like an angel singing" --> 1.2 Good taste
..
etc..
This step is called "pre code"
Then all 10K surveys get read and coded --> "Code"
Lists are kept with answers that don't fit.
If enough "samesies" are collected, they get their own code.
If not, they get collected for presenting as outliers/amusing / getting dropped.
Then you can do your stats as usual.
::emp::
Thanks for your time.
The process you describe is exactly how I imagined it being done... 10 years ago. I was hoping I can replace the interns with ML, given how far packages like scikit-learn have come.
Anyway, I'm gonna order some books now and will report back if I find something valuable on the topic.
Well, you could do a simple "sentiment analysis" and maybe combine it with a summarizer /keyword extraction.
AlchemyAPI | Transforming text into knowledge
AlchemyAPI has everything you need (and with 3K free API calls per day, probably enough to work with)
::emp::
AlchemyAPI | Transforming text into knowledge
AlchemyAPI has everything you need (and with 3K free API calls per day, probably enough to work with)
::emp::