Ina Fried, reporting on a survey of iPhone users which apparently shows the iPhone 5c doing worse than the equivalent model in the range did last year, the iPhone 4S:
The iPhone 5c, meanwhile, represented 27 percent of sales, less than the 32 percent that the iPhone 4s had a year ago when it was the mid-range model. CIRP’s findings are based on a survey of 500 buyers of Apple gear during the survey period.
How accurate a survey is depends on the sample size. And while 500 is a valid sample size considering the overall “population” of around 50m iPhone buyers, it isn’t enough to reduce the margin of error to the kind of point which would justify the claims being made.
With a sample this size, unless they’d taken a lot of precautions, the margin of error would be around +/–4–5%. This means, potentially, the models could have the reversed share: 32% for the 5c (27% + margin of error upwards of 5%) and 27% for the 4S (32% – margin of error down of 5%).
All that you can say based on these numbers is the iPhone 5c is performing about as well as the iPhone 4S did in the same role. However, a conclusion like that isn’t enough for CIRP. According to CIRP analyst Josh Lowitz:
“If the old iPhone 5 had been the mid-priced phone, we expect that it would have sold a higher percentage of iPhones than the 5c did, as previous mid-priced legacy iPhones have,” Lowitz said. “The 5c seems to have been designed to force certain buyers to the 5s.”
I’m not sure how you can “force” buyers into buying something more expensive if they haven’t got the money. Sure, it probably does well at “upselling” customers: those who would never consider buying the top-end phone and so walk into an Apple Store to get the 5c, but then play around with a 5S and get a case of profound gadget-lust. But I can’t imagine that more of those kinds of customers took at a look at last year’s top of the line phone and didn’t feel equally lustful.
A “population” in survey terms is the number of people who fall into the group you’re trying to sample: in this case, the roughly 50m people in the US who bought iPhones that year. The bigger the population, the bigger the sample size you need. ↩