online: is it the next lily pad? - tns...
TRANSCRIPT
This is not the kind of presentation some of you are used to seeing delivered by a research supplier
Sit back, relax and please hold your questions
There is a belief that TNS and all private sector suppliers have a vested interest that influences our position and design recommendations
Our reputation is influenced by the quality of decisions that are made with the data we deliver
We have all had experience with shoot the messenger and
we should not not allow ourselves to be gunned down
Results matterAt a minimum, the data we release have to be internally consistent and
capable of gauging relative change odifference
Your suppliers must have an expressed STRATEGYto stay ahead of the curve.
They must commit to the heavy investment needed to deliver the best data to the task at hand …and in the context of shrinking client budgets and time frames
The number of text messages sent and received in a single day
exceeds the population of the entire planet
A week’s worth of New York Times has more information than a person was likely to come
across in a lifetime in the 18th century
That’s why marketing research companies have been merging and agglomerating at
break-neck speed over the past several years
Virtually all major marketing research firms have a global strategy to move
as much work as quickly as possible to the Internet
Today, that is the best way that these publically traded goliaths can deliver the biggest returnto their investors, and guarantee share-price appreciation
Investments in technology and Internet access panels are substantial
And it is clear that the maintenance and development costs needed to remain leading-edge are escalating
First:
broad-brush research that is executed almost routinely for manufacturers and service providers
Broad-brush research is done to gauge consumer reactions and to get a sense of opinion and relative difference
There could be wide variance on the data gathered,
but the downside is minimal and the results do not have to
be projectable
It includes certain program-evaluation studies, media audience measurement, and precise calibration studies
In all corners of theglobe, sophisticatedtechnical-mindedresearch professionalsare fixated on how toobtain accurate measurement on these types of studies –how to get all of those angels on the head of the pin
The fundamental difference between how these two general types of research are executed is HUGELYimportant
Precise calibration studies, such as NADbank and PMB audience measurement, use random samples, with known probabilities of selection
They are representative and projectable, and the results tend to be stable over time
Since these studies are replicable – we expect to get the same results if they are repeated
Other forms of research are typically conducted using quota samples, be it in malls, on the phone, by panel, (mail or internet)
These samples are self-selectingconvenience samples with no known probability of selection for participants
Perhaps as a consequence, even within the broad-brush arena panel samples or quota samples were always looked upon as cheap alternatives in Canada
Since they were so common, it was not a big leap to move from panel based
research to interactive panels
Quota samples are quota samples, after all
Therefore, the buzz is tremendous about interactive panels in the U.S. and it has permeated Canadian airwaves
And the mediais in on the hoax
Send us your views
Ten minutes later –here are the results of the CBS poll
Internet research solves all survey problems and can be used for all kinds of research in all circumstances
All that buzz is countered byless audible voices from professionals,
who work in major research corporations, and who repeatedly SAY:
With convenience samples
and Internet panels are convenience samples
the bias is simplynot knowable or understandable
Yes, cell phones are increasing in popularity and there is some migration away from land lines,especially among youth
Many of these difficulties can be overcome while retaining the advantages of probability sampling
…but at a cost$$$$$
For some research purposes, panel response rates, really, cooperation rates
are largely meaningless, since there is no statistical probability of measurement
involved at the outset
That is, the differences observed are not just expected differences between those with and without Internet access
This may not matter for many research purposes, but it is important to recognize that
sometimes it does
And it is not because they are not representative of the population in terms of their
age or sex or education or income
Even when we make them look the same demographically,
their consumption patterns, lifestylesand attitudes may be different