Study Identifies Factors That Could Lead to Cancelled Subscriptions

[Heather Morrison <heatherm@eln.bc.ca>]

On 20-Nov-06, at 5:29 AM, Chris Beckett wrote:

  Just briefly, in response to Heather's points,

1. "Removing a key factor, at best, makes the validy [validity] 
of a study questionable."

   - No it doesn't. The survey doesn't pretend to have measured 
all the factors in content selection preference - but most of 
them. The factors measured were validly measured. (Bad brakes and 
worn tyres increase the likelihood of car crashes - I don't need 
to know how bad the brakes are to know that worn tyres increase 
the likelihood of car crashes.)

Interesting, Chris, however I disagree.

Since we're using car comparison, let's say we remove a key 
variable when investigating the likelihood that cars will work 
well for us - for example, the engine.

We might do a study of consumer preferences to determine car 
purchase decisions that looks at a number of admittedly important 
variables - such as cost, safety, colour - but if we overlook 
something that is a key variable to the consumer - does the car 
have an engine?, then our study is likely to be a poor predictor 
of consumer behavior, isn't it?

"Importance to the collection" is to library purchasing decisions 
what "does it have an engine?" is to the person purchasing a car. 
If this key variable is missing, all the others are irrelevant.

best,

Heather Morrison
http://poeticeconomics.blogspot.com