Dear Readers,
I would like to respond to some comments made today in an
article by Gareth Halfacree on the bit-tech.net website. You can view that article here. Specifically, a number
of claims were made by Ray Gorman of Lenovo about our Laptop Reliability study, and I would like
to clarify each of those claims, and give a rebuttal where necessary.
Ray Gorman: “total number claimed in this
report is not a statistically significant sample for a study where no attempt
is made to control key variables affecting repair rates, such as comparable
machine types, end users, geography, and applications”
SquareTrade’s study examined data from 30,000 laptop
owners, which we believe to be a very statistically significant sample. When a national publication polls Americans on
their attitudes towards political matters, they poll 1000-2000 people; a sufficient
sample size to represent the 230 million Americans.
We did not control for the variables Mr. Gorman
mentioned, but we also did not feel it necessary. With an average sample of over 3000 laptops
per manufacturer, our sample size is large enough to assume a random
distribution of users for each variable.
As an example culled from Lenovo's own
warranty repair data, Gorman points out that the company would "expect a
10X difference in repair rates between systems bought for [secondary school]
students and systems used only in a home office by adults," a distinction
which isn't made by SquareTrade's study.
Our data set assumes an equal distribution of the types
of consumers for each manufacturer. Thus,
we believe that our data sets for each manufacturer contain a similar
proportion of students, home office users, frequent travelers, and so on.
We analyzed our data to produce a general study of manufacturer
failure rates relative to one another, not to produce a study only about the
reliability rates of laptops used in a home-office setting.
To return to the polling analogy, pollsters choose random
survey participants in the phone book, and do not control for geography, race,
income, or any number of other variables that are may be very relevant in
determining which candidate the participant would vote for.
Arguing that SquareTrade "has a vested
interested in showing scary failure rates as they have done here [as] they are
in the business of selling after sale warranties," Gorman claims that
Lenovo's internal data shows that laptop failure rates are "at least
two-thirds lower than what is claimed in the Square Trade survey."
I won’t dispute the fact that SquareTrade is in the
business of selling warranties. However,
I do dispute the fact that we’re trying to scare consumers to buy warranties.
We find our overall results for laptop failure rates to
be very consistent and in line with other organizations that have reported research.
We direct our readers to the following
reports:
- In June 2006, Gartner
reported actual failure rates of 28% for systems purchased in 2003-2004. They projected a 22% failure rate, based on 1
year of data for systems bought in 2005-2006.
With our results showing a LOWER failure rate than reported
by these other sources, we are telling consumers they need warranties LESS
frequently than the numbers published by Consumer Reports. If anything we’re providing “less scary”
numbers than Consumer Reports and Gartner on laptops.
Furthermore, we have also published past research reports
that show device failure rates far below the point at which it makes sense for
most consumers to purchase warranties.
Most recently, our Game
Console failure rate study from August 2009 found that only 2.7% of Nintendo
Wii systems failed over 2 years. We
would not have published such data if our only interest was in selling
warranties. We provide the information
to consumers and make affordable warranty protection available to them. It is up to them to make the decision of whether
a warranty is worthwhile.
For those concerned by the high rate of
failure that SquareTrade quoted in their study, Gorman says not to worry:
"PC hardware is extremely reliable, and this study is full of holes[,
as the] method is flawed, the data is inaccurate, and the conclusion is wrong."
I strongly believe that our study of over 30,000 laptops is
one of the most comprehensive studies on laptop reliability. We have not surveyed customers on past
failures, which allows for subjective recollection and brand affinity bias their
memory. Instead, we have recorded device
failures as they have been reported to us for warranty service.
Readers, let me know what you think. I am happy to discuss any questions about the study in
further detail.
Respectfully yours,
Vince Tseng
VP of Marketing, SquareTrade
PS. I am typing this out on my Lenovo T61 ThinkPad, which
is the standard notebook distributed by SquareTrade’s corporate IT department. Before my current Lenovo ThinkPad, my prior personal laptop and corporate laptop were both IBM ThinkPads. For
what it’s worth.
