SPEC Kit 337: Print Retention Decision Making · 135
GEORGIA INSTITUTE OF TECHNOLOGY
Georgia Tech Algorithm (ASERL Collaborative Journal Retention Program)
offerings to the project many libraries are offering titles that they have previously converted to
electronic, so this feels less critical. Currency, however, offers a boost for those titles that are currently
received or only recently cancelled. I chose to set a floor of 0 for currency, but you may also allow
negative values (e.g. a title cancelled in 1989 would have a currency of 0 in our model, or with negative
values have a currency of -‐2.2). Class, as a dummy variable, should be customized to your own needs,
and at Georgia Tech we chose to emphasize our science and engineering holdings equally, but we could
have been more selective (e.g. LC Class TA: Class =0.3 LC Class QL: Class =0.1). I would suggest limiting
the range of this variable to -‐0.3 to 0.3 to keep it in line with the rest of the formula.
Finally, multiplying the ASERL value times -‐2.25, reflects our decision process for these titles, and
reflects that we need no longer consider these titles another school has agreed to keep what we
previously had agreed to discard. This is a very aggressive approach that can be adjusted based on your
circumstances.
In our initial use of the algorithm, we had values between -‐2.24 and 2.25 with one outlier value of -‐5.1
(we did not receive 32 volumes of this title). We have two cutoff criteria +2 and +1. Values of 2 or more
will likely be offered by GT to the ASERL project (3%). Values of 1 or less are no longer being considered
for inclusion (68%). Values between 1 and 2, so far merit additional evaluation (29%). In our second
run, we updated the ASERL holdings and removed JSTOR titles from consideration narrowing our list to
21 likely titles (3.1%) and 155 review titles (23%).
It should be noted that the algorithm does not consider gaps in contributions by other schools. State
law prevents us from offering our holdings to other schools, but we are considering contributing to the
project (and retaining our copy) of large gaps where our holdings are complete (e.g. ISSN 0022-‐3093 on
the ASERL journals spreadsheet). A new variable ASERLGAP could be created using a method similar to
Missing and subtracted from the ASERL variable:
(FirstCopy)2 – Missing/10 +(LastCopy OR Currency) +Class +(ASERL – ASERLGAP) *-‐2.25
Additional Background
In 2010, a mold outbreak was discovered in our basement compact storage facility, which housed most
of our pre-‐1980 bound periodicals. A decision was made to clean the material and relocate the material
to an existing off-‐site warehouse. Prior to the outbreak we had been in the process of identifying
material to relocate to that facility. At the outset, we knew that the warehouse did not have enough
room to contain all of the materials from the warehouse, that we could not safely reuse the basement
facility, and that there was insufficient room in our stacks some material had to be discarded.
To determine materials to discard, we looked at our deep backfiles where we had both archival rights
and ILL lending rights. This list included titles from JSTOR, Wiley, Elsevier, American Chemical Society,
Royal Chemical Society, Institute of Physics, and Nature (one backfile that met the initial criterion was
retained -‐-‐ AIAA journals). We checked the holdings and discarded pre-‐1980 runs with electronic
GEORGIA INSTITUTE OF TECHNOLOGY
Georgia Tech Algorithm (ASERL Collaborative Journal Retention Program)
offerings to the project many libraries are offering titles that they have previously converted to
electronic, so this feels less critical. Currency, however, offers a boost for those titles that are currently
received or only recently cancelled. I chose to set a floor of 0 for currency, but you may also allow
negative values (e.g. a title cancelled in 1989 would have a currency of 0 in our model, or with negative
values have a currency of -‐2.2). Class, as a dummy variable, should be customized to your own needs,
and at Georgia Tech we chose to emphasize our science and engineering holdings equally, but we could
have been more selective (e.g. LC Class TA: Class =0.3 LC Class QL: Class =0.1). I would suggest limiting
the range of this variable to -‐0.3 to 0.3 to keep it in line with the rest of the formula.
Finally, multiplying the ASERL value times -‐2.25, reflects our decision process for these titles, and
reflects that we need no longer consider these titles another school has agreed to keep what we
previously had agreed to discard. This is a very aggressive approach that can be adjusted based on your
circumstances.
In our initial use of the algorithm, we had values between -‐2.24 and 2.25 with one outlier value of -‐5.1
(we did not receive 32 volumes of this title). We have two cutoff criteria +2 and +1. Values of 2 or more
will likely be offered by GT to the ASERL project (3%). Values of 1 or less are no longer being considered
for inclusion (68%). Values between 1 and 2, so far merit additional evaluation (29%). In our second
run, we updated the ASERL holdings and removed JSTOR titles from consideration narrowing our list to
21 likely titles (3.1%) and 155 review titles (23%).
It should be noted that the algorithm does not consider gaps in contributions by other schools. State
law prevents us from offering our holdings to other schools, but we are considering contributing to the
project (and retaining our copy) of large gaps where our holdings are complete (e.g. ISSN 0022-‐3093 on
the ASERL journals spreadsheet). A new variable ASERLGAP could be created using a method similar to
Missing and subtracted from the ASERL variable:
(FirstCopy)2 – Missing/10 +(LastCopy OR Currency) +Class +(ASERL – ASERLGAP) *-‐2.25
Additional Background
In 2010, a mold outbreak was discovered in our basement compact storage facility, which housed most
of our pre-‐1980 bound periodicals. A decision was made to clean the material and relocate the material
to an existing off-‐site warehouse. Prior to the outbreak we had been in the process of identifying
material to relocate to that facility. At the outset, we knew that the warehouse did not have enough
room to contain all of the materials from the warehouse, that we could not safely reuse the basement
facility, and that there was insufficient room in our stacks some material had to be discarded.
To determine materials to discard, we looked at our deep backfiles where we had both archival rights
and ILL lending rights. This list included titles from JSTOR, Wiley, Elsevier, American Chemical Society,
Royal Chemical Society, Institute of Physics, and Nature (one backfile that met the initial criterion was
retained -‐-‐ AIAA journals). We checked the holdings and discarded pre-‐1980 runs with electronic
