From: David Prosser <[log in to unmask]>
Date: Mon, 11 Jun 2012 22:04:18 +0100

At the risk of disturbing the wonderful harmony that has developed
between Phil and I, I think I was being a little less disingenuous
than he suggests.

Phil thinks that the statement "There is absolutely no evidence to
show that OA causes cancelation" is a little strong and compares it to
those who say: "Cigarettes don't cause cancer."  But what we do know
is that people who smoke on average tend to develop lung cancer more
often than those that do not.  Even if one cannot prove causation
there is a massively strong correlation.

In our case we don't even have the correlation.  We cannot say that on
average journals that provide OA after a six month embargo loose
subscriptions more rapidly (or even less rapidly) than those that do
not.  We don't have the data.  (and like Rick I would love to see the
data.)  What we can say is that a number of journals have made their
content available after 6 months without suffering the haemorrhaging
of subscribers that the surveys predict.  My point is that the level
of cancellation reported in the surveys would surely be noticeable -
we are not talking about tiny nudge factors here, but 44%
cancellations!

I do agree with Phil that cancellation decisions are much more complex
than the headlines from the surveys suggest - and are definitely more
complex than the press releases from the PA and ALPSP would indicate.

David Prosser


On 10 Jun 2012, at 23:59, LIBLICENSE wrote:

> From: Philip Davis <[log in to unmask]>
> Date: Fri, 8 Jun 2012 13:18:52 -0400
>
> I don't usually agree with David, but I think his analysis of this
> report is accurate and perceptive. There are some situations where
> surveys accurately predict real behavior: this is not one of them.
>
> Personally, I believe that the attempt to make a causal statement
> linking open access and cancelation decisions is futile simply because
> cancelation decisions are complex, involve many different factors, and
> are based on retrospective and longitudinal data.
>
> There *is* an argument to be made that freely accessible articles
> located on a large portal (viz. PMC) would lead to decreased
> readership on publisher websites. Librarians base their cancelation
> decisions, in part, on publisher-provided usage data. Hence, we can
> imagine an indirect link between OA policy and cancelation working
> through usage as an intermediary. The best way to state this is "OA is
> a contributing cause to subscription cancelation". The wrong way to
> state it is that "OA causes cancelation". Similarly, it is also
> disingenuous to make a negative causal statement, i.e. "There is
> absolutely no evidence to show that OA causes cancelation".
> Technically, this is a correct statement, but it assumes a direct,
> deterministic causal connection between OA and cancellation. It would
> be akin to claiming -- as special industry groups have made:
>
> Cigarettes don't cause cancer.
> Guns don't kill. People kill.
> There is absolutely no evidence that consuming soft drinks
> (soda/pop/Coke) cause obesity
>
> David is right that these survey studies do not accurately reflect
> actual behavior and that the smoking gun is just not found in the
> data. He is correct. On the other hand, drawing eyeballs from a
> journal to a public repository poses many risks that we need to
> acknowledge. Publishers miss the opportunity to point readers to other
> related articles on their site. They miss an opportunity to convey
> public service announcements and advertisements. They miss the
> opportunity to draw readers to participation in conferences and
> society membership. You may be cynical and say this is nothing but
> marketing and profit-making. I call it building a community.
>
> --Phil Davis
>
>
> On Jun 7, 2012, at 8:57 PM, LIBLICENSE wrote:
>
>> From: David Prosser <[log in to unmask]>
>> Date: Thu, 7 Jun 2012 16:31:56 +0100
>>
>> This survey reminds me of one carried out by Chris Beckett and Simon
>> Inger back in 2007.  Both were surveys of libraries carried out on
>> behalf of publisher associations, both appeared to suggest that
>> six-month embargoes would trigger massive rounds of cancellations, and
>> both took no account of what actually happens in reality.
>>
>> There are a growing number of journals that already make their content
>> freely available after six months.  These journals make the content
>> available ether through their own websites, through others (such as
>> PubMedCentral), or both.  Some of them have been doing so since well
>> before the Beckett and Inger study so we potentially have six years of
>> data to test the hypothesis that a six-month embargo could lead to
>> subscription losses in the order of 44%.  We also have extensive
>> experience in some fields of what happens to subscriptions if
>> peer-reviewed copies are made available immediately on publication.
>>
>> As far as I know, there is no evidence to suggest that six months
>> embargoes have led to 44% reductions in subscriptions as predicted by
>> this latest survey.  The journals that make their content available
>> after six months appear to be thriving and are sustainable.  If I have
>> missed the evidence please let me know.
>>
>> So, what do you do if you discover that your survey results do not
>> necessarily reflect reality?  Well, I guess that if the results match
>> your ideological bent you ignore the discrepancy and issue press
>> releases.  If you are interested in actually looking at behaviour you
>> try to dig a little deeper.  One can begin to think of a number of
>> possible reasons for this discrepancy, including:
>>
>> 1. Librarians don't realise that the material is available six months
>> after publication (unlikely, surely)
>> 2. The journals that make material available tend to be in biomedical
>> fields - perhaps these areas are less susceptible to cancelations.
>> But it doesn't explain high-energy physics and economics where green
>> OA copies of all papers in some journals can be found without embargo.
>> 3. People don't always do what they say they do in surveys.
>>
>> The trouble is that in the cases where six month embargoes have been
>> tried they give results that completely fail to match the results of
>> the survey.  Until we come to terms with why this is the case I find
>> it hard to see how we can take the survey at face value.
>>
>> Best wishes
>>
>> David