Confirmation of what everybody already knows: iPhone retention 94% vs. Android 47%

[FrozenTomato]

Finds high pent up demand for iPhone 5 among Verizon subscribers and Android owners.

Among existing iPhone users, 94% expect to buy another iPhone (6% expect to switch to Android)

Among existing Android users 47% expect to buy another Android smartphone (42% expect to switch to iPhone).


http://tech.fortune.cnn.com/2011/08/01/survey-iphone-retention-94-vs-android-47/

 

[Rodimus Prime]

I read the article. It is a small sample group (216) in one city.

Translation means nothing.

 

[gkarris]

I think the question instead of "Phone", should be "Provider"...

 

[srf4real]

My droid was great for two months. Then it started crashing and randomly deleting my contact list entirely with nothing running on the phone. After the third time of this, I insisted on switching to an iPhone shortly after VZ began carrying them and have had no hiccups since then, six months ago. Now the iPhone4 is jailbroken and it still works flawlessly and is fully stable 100% of the time. You will never see me with a different phone again.

 

[yg17]

My droid was great for two months. Then it started crashing and randomly deleting my contact list entirely with nothing running on the phone. After the third time of this, I insisted on switching to an iPhone shortly after VZ began carrying them and have had no hiccups since then, six months ago. Now the iPhone4 is jailbroken and it still works flawlessly and is fully stable 100% of the time. You will never see me with a different phone again.

I woke up one day and my iPhone 3G was as dead as a doornail. I've also had an LCD in a PowerBook G4 crap out, the temperature sensor in the same PowerBook G4 quit working, and the keyboard in the PowerBook G4 die. Oh, and the video card in my parents' Intel iMac, the hard drive in their Mac Mini, and the logic board in my MacBook.

Products break. You got a bad phone. That is not representative of other Droids or Android itself.

 

[*LTD*]

Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 4_3_3 like Mac OS X; en) AppleWebKit/533.17.9 (KHTML, like Gecko) Version/5.0.2 Mobile/8J2 Safari/6533.18.5)

Apple customers stay Apple customers. There are a few minor exceptions, however. And they're all hanging out on MacRumors.

 

[ChazUK]

Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 4_3_3 like Mac OS X; en) AppleWebKit/533.17.9 (KHTML, like Gecko) Version/5.0.2 Mobile/8J2 Safari/6533.18.5)

Apple customers stay Apple customers. There are a few minor exceptions, however. And they're all hanging out on MacRumors.

That's because some of us own Macs.

 

[roadbloc]

Wirelessly posted (Mozilla/5.0 (iPhone; U; CPU iPhone OS 4_3_3 like Mac OS X; en) AppleWebKit/533.17.9 (KHTML, like Gecko) Version/5.0.2 Mobile/8J2 Safari/6533.18.5)

Apple customers stay Apple customers. There are a few minor exceptions, however. And they're all hanging out on MacRumors.

I'm unsure how you have come to such a conclusion. I buy more than just Apple products you know. I have two PCs, an IBM RS/6000 and a cheap brandless android based tablet.

How do I, or in fact, the majority of forum users fit into your supposed theory.

 

[rhett7660]

I'm unsure how you have come to such a conclusion. I buy more than just Apple products you know. I have two PCs, an IBM RS/6000 and a cheap brandless android based tablet.

How do I, or in fact, the majority of forum users fit into your supposed theory.

Looks like you are one of those exceptions that hang out at macrumors.

 

[CalBoy]

The writer didn't even mention if control metrics were taken (ie gender, age, income) to make sure that the sample was representative.

You can't conclude anything from this sample, let alone "confirm" something.

 

[imahawki]

Its an informal survey, that doesn't mean it's useless. In studying statistics, you'd be surprises how small a sample can be and still be meaningful.

Anecdotally I think what is happening is people are buying their first "2nd gen" smart phone (after a BlackBerry or Palm Treo type device) and they're looking at the the specs (processor, GPU, memory, screen size) and ticking off all the boxes and getting an Android and then realizing it just doesn't do what they expected it to and doesn't really integrate with anything that well. That's what happened with me personally.

 

[voonyx]

People who don't know how statistics work will dismiss this survey, but that's not what's really happening.

You dont need 1,000,000 responders to get a representative sample. 216 is more than enough, there'll just be a slightly higher margin of error. In this case, that doesn't mean that instead of 94%, it actually means 2% will buy an iPhone, or instead of 42% expecting to switch that 0% are going to switch.

If the people who took this survey answered truthfully, there is no reason to think that this is not going to be the case. Not sure why people think half the galaxy has to be surveyed to get a representative sample.

 

[Rodimus Prime]

Its an informal survey, that doesn't mean it's useless. In studying statistics, you'd be surprises how small a sample can be and still be meaningful.

Anecdotally I think what is happening is people are buying their first "2nd gen" smart phone (after a BlackBerry or Palm Treo type device) and they're looking at the the specs (processor, GPU, memory, screen size) and ticking off all the boxes and getting an Android and then realizing it just doesn't do what they expected it to and doesn't really integrate with anything that well. That's what happened with me personally.

If memory serves me you can get some ok results with a sample size of around 30 depending on the standard deviations. I know under 30 you need to have a tau factor put into it to help form the results but larger than that it oddly enough can give you some what of an idea but standard deviation come into play plus your margin of error is an issue.
To get down to around a 1% error you need something like 10k. compared to 3% of around 1k.

People who don't know how statistics work will dismiss this survey, but that's not what's really happening.

You dont need 1,000,000 responders to get a representative sample. 216 is more than enough, there'll just be a slightly higher margin of error. In this case, that doesn't mean that instead of 94%, it actually means 2% will buy an iPhone, or instead of 42% expecting to switch that 0% are going to switch.

If the people who took this survey answered truthfully, there is no reason to think that this is not going to be the case. Not sure why people think half the galaxy has to be surveyed to get a representative sample.

I know full well how statics works.
I also know can see huge errors in how this one was gathered.
A sample size of 216 is going to give you a large margen of error. 1000 gives you around +/- 3% and that is if you do everything right.

This case they choose a northern city and only a northern city and from the state it is located in I am going to call it a small city by my standards.

It has huge errors which is why it is worthless. Tossing out the sample size as a problem there are still huge problems with the survey

 

[voonyx]

If memory serves me you can get some ok results with a sample size of around 30 depending on the standard deviations. I know under 30 you need to have a tau factor put into it to help form the results but larger than that it oddly enough can give you some what of an idea but standard deviation come into play plus your margin of error is an issue.
To get down to around a 1% error you need something like 10k. compared to 3% of around 1k.



I know full well how statics works.
I also know can see huge errors in how this one was gathered.
A sample size of 216 is going to give you a large margen of error. 1000 gives you around +/- 3% and that is if you do everything right.

This case they choose a northern city and only a northern city and from the state it is located in I am going to call it a small city by my standards.

It has huge errors which is why it is worthless. Tossing out the sample size as a problem there are still huge problems with the survey

216 would give you probably +/- 6 or 7%...like i said its not a foolproof survey and that margin of error isn't small, but to dismiss it as baseless makes no sense. 94% might not stick with the iphone but a sample size of 216 doesn't mean that only 4% actually will or something...

 

[Rodimus Prime]

216 would give you probably +/- 6 or 7%...like i said its not a foolproof survey and that margin of error isn't small, but to dismiss it as baseless makes no sense. 94% might not stick with the iphone but a sample size of 216 doesn't mean that only 4% actually will or something...

That part I can go with on the 216 but you failed to address the larger issue on why I call this survey crap.

It has a very poor sample group no matter how you cut it. It could have 10k people and it still would be crap. There sampling methods are not random, poor location grouping and hell located only in one small area. That makes makes it crap.

The fact that they are pushing it like it is the norm and the end all result is piss poor journalism at best. No getting around that fact.

Garage in= Garage out. Since the the data in was garage the results are garage.

 

[imahawki]

This case they choose a northern city and only a northern city and from the state it is located in I am going to call it a small city by my standards.

What are your standards? MSP is the in the top 15 metropolitan areas in the US by population. Its not NY or LA but calling it a small city is dumb. Des Moines is a "small city".

As far as garage [sic] in, there is a very large continuum between perfect statistical sampling and garbage. If I try to draw a conclusion by asking the 5 people on my team at work that's garbage. A limited geographical sample in a large metropolitan area is imperfect but far from garbage (or a garage).

 

[voonyx]

That part I can go with on the 216 but you failed to address the larger issue on why I call this survey crap.

It has a very poor sample group no matter how you cut it. It could have 10k people and it still would be crap. There sampling methods are not random, poor location grouping and hell located only in one small area. That makes makes it crap.

The fact that they are pushing it like it is the norm and the end all result is piss poor journalism at best. No getting around that fact.

Garage in= Garage out. Since the the data in was garage the results are garage.

I agree with you that they shouldn't publish it as "94% will switch" like it's fact, but I disagree on the other stuff.

You don't need a nationwide, sweeping survey of major cities to get a representative sample. Obviously, the bigger the sample, and the more diverse the sample the more accurate. But again, the study of statistics exists for a reason. You would be surprised at how accurate a very small sample can be (provided of course that all respondents told the truth).

 

[BaldiMac]

The fact that they are pushing it like it is the norm and the end all result is piss poor journalism at best. No getting around that fact.

Here is the first sentence of the article:
"It's only a tiny sample, so large conclusions must not be drawn."

 

[Rodimus Prime]

Here is the first sentence of the article:
"It's only a tiny sample, so large conclusions must not be drawn."

Cop out is what i would call that.
Head lines, and the rest of the article try to glaze over that fact and try to act as it is the norm.
As this thread proves people here are acting like it is a fact and that is just the way it is. The company publishing that knew it. Also I have seen this article pop up in multiple sources which tells me they care more about the headline than reporting.

 

[voonyx]

Cop out is what i would call that.
Head lines, and the rest of the article try to glaze over that fact and try to act as it is the norm.
As this thread proves people here are acting like it is a fact and that is just the way it is. The company publishing that knew it. Also I have seen this article pop up in multiple sources which tells me they care more about the headline than reporting.

the reporting was poor, sure and the article was designed to get hits, but the survey is the survey and it cant just be dismissed as baseless because of poor or shady journalism

 

[Rodimus Prime]

the reporting was poor, sure and the article was designed to get hits, but the survey is the survey and it cant just be dismissed as baseless because of poor or shady journalism

no it can be dismissed on being a poor survey with many many problems with it.

 

[BaldiMac]

Cop out is what i would call that.
Head lines, and the rest of the article try to glaze over that fact and try to act as it is the norm.
As this thread proves people here are acting like it is a fact and that is just the way it is. The company publishing that knew it. Also I have seen this article pop up in multiple sources which tells me they care more about the headline than reporting.

It sure is convenient to just throw out and ignore any facts that disagree with you.

 

[voonyx]

no it can be dismissed on being a poor survey with many many problems with it.

the problems you're naming have no basis...small sample size doesn't discredit a survey, you say its a small city, someone said it's not, not having diversity doesn't discredit a survey...you're just saying things without looking at the big picture...i think you're more upset with the results of the survey then the techniques used.

 

[Rodimus Prime]

the problems you're naming have no basis...small sample size doesn't discredit a survey, you say its a small city, someone said it's not, not having diversity doesn't discredit a survey...you're just saying things without looking at the big picture...i think you're more upset with the results of the survey then the techniques used.

Small city is realtive and to me it is a small city but then again I lived in Houston and Dallas so that should show you my base line for a large city.

From quick look over it I question if the sample group is random at all. In they went up to people to talk with them at places like Food courts, baseball game which leave some pretty big problems in how it was done. It smells of bad survays and those locations have certain demographics that tend to show up. So again leaving out a lot of other demographics.

And even if it was random Geographically it is a small area no matter how you cut it. Hence the reason I call it a bad survey over all.

 

[BaldiMac]

Small city is realtive and to me it is a small city but then again I lived in Houston and Dallas so that should show you my base line for a large city.

From quick look over it I question if the sample group is random at all. In they went up to people to talk with them at places like Food courts, baseball game which leave some pretty big problems in how it was done. It smells of bad survays and those locations have certain demographics that tend to show up. So again leaving out a lot of other demographics.

And even if it was random Geographically it is a small area no matter how you cut it. Hence the reason I call it a bad survey over all.

And yet it was good enough for a professional analyst to provide a quick analysis suggesting certain trends to his customers.