Previous month:
March 2008
Next month:
May 2008

Posts from April 2008

Watching SurveyUSA

The Survey Geek has a thing about these automated RDD interviews that are the staple of SurveyUSA's business.  Last week I noted that they were reporting an 18 percent lead for Clinton in the PA primary while most other pollsters were in the single digits.  As of yesterday they were showing Clinton at 50 percent, Obama at 44 percent.   So Clinton has lost 12 points in just over a week.  Or at least according to Survey USA. Imagine.

More tomorrow.

Interesting Telephone Numbers

The recent issue of Directions, a newsletter put out by Marketing Systems Group, has a feature article on cellular telephone sampling.  They are worth listening to because they and Survey Sampling have long been the manor players in the telephone sampling space.  They still supply us with the data and software that we use to draw RDD samples.  The highlights of the article:

  • There currently are 110 million residential landline telephone numbers in the US.
  • The total number of possible telephone landline numbers in the US is 1.1 billion.
  • The total number of possible cellular numbers is 400 million.
  • There are 250 million cellular subscribers, accounting for about 65 percent of the available numbers.
  • The current estimate is that 2.3 million landline numbers have been transferred (in technical terms, ported) to a cellular phone. That's about 5 percent of total assigned landline numbers.
  • The working estimate for cell only households continues to be 15 percent.

The issue also has an interesting note on calling cell phones with a predictive dialer. This is specifically prohibited by the FCC. But according to MSG, if you ask someone for their number and they give you a cell phone number you can call it with a predictive dialer.  In other words, if the cell phone number comes to us from a client in a customer sat study we can call it.  As a practical matter, this happens with some regularity.  Nice to know we probably are not breaking the law.

Automated Telephone Surveys Redux

2Way back in June of 2006 I did a post on automated telephone surveys.  These are telephone surveys that use RDD samples to initiate calls to households and then try to administer an IVR survey.  Not humans involved.  The leading practitioner of this approach is Survey USA, a company that spends a good deal of time claiming that what they do is just as scientific and as reliable as traditional phone surveys with those pesky interviewers.

Well, one of the Newsweek bloggers has looked at the recent Pennsylvania Democratic Primary polls and made some pretty interesting comparisons.  The various polls show the Clinton lead over Obama like this:

    Survey USA                            18%
    American Research Group    15%
    Quinnipiac                               6%
    Rasmussen                               5%
    Public Policy Polling                 3%

This is worth watching for a whole lot of reasons.

Great Minds!

No sooner had I finished my last post suggesting that perhaps the industry was focusing on the wrong things relative to panel data quality than into my inbox came a call for papers for the next ESOMAR Panels Conference.  Of particular relevance is this section of the call:

The research industry as a whole is at the edge of a new era where clients and providers are looking for partnerships to invest in the future of business. At the same time, facts are showing us that a decreasing number of clients view online research as a valuable decision-making tool... Has the online panel industry been too absorbed by its own issues and lost perspective? Time to zoom out and look at the bigger business spectrum whilst pushing for internal excellence!

What can I say other than, "Well said!"