Rdap13 Dharma Akmon The Role of Value in Data Practices

D H A R M A A K M O N

R D A P 1 3 A P R I L 5 , 2 0 1 3

THE ROLE OF VALUE IN DATA PRACTICES

MOTIVATION

• Increasing attention to data as a valuable

product of science

• Scientists’ actions throughout data life cycle

impacts significantly on what is available for

preservation and reuse

PREVIOUS WORK

• Scientists withhold or inadequately manage

data because:

• Documentation is labor intensive and

unrewarded (Birnholtz & Bietz, 2003; Campbell et al., 2002;

Louis, Jones, & Campbell, 2002)

• They are more concerned with publications (Borgman, Wallis, Mayernik, & Pepe, 2007)

• They fear data contributions will not be

recognized (Louis et al., 2002)

PREVIOUS WORK CONT.

• Social scientists reported they’d be more likely

to document and deposit data if they thought

data "would be used and have a broader

public benefit" (Hedstrom & Niu, 2008)

• “Shareable” data are those that are expected

to have the greatest potential for generating

new results (Cragin, Palmer, Carlson, & Witt, 2010)

• Less likely to share high value or hard-won data (Tucker, 2009; Borgman, Wallis, & Enyedy, 2007)

RESEARCH QUESTION

How do scientists conceive of the value of their

data, and how is this reflected in their data practices?

• What uses for data are salient to scientists?

• What time spans do scientists use to think about

data's value?

• How do scientists create data that are valuable and what do they do to make data accessible over

time?

SITE & METHODS

• 3 small teams of scientists at an ecological field

station

• Teams differed across:

• PI career stage

• Methodological approach to research

• Length of study

• Funding source

“ECOLOGIST” FASHION

NUTRIENT UPTAKE IN STREAMS (NUS) TEAM

Name* Career Stage Discipline Project Role

Elizabeth Assistant prof. Biogeochemistry PI

Jessica Assistant prof. Stream ecology PI

Tina Graduate student Hydrogeology Graduate

researcher

Carolyn Undergraduate student Environmental studies Undergraduate

researcher

Janet Undergraduate student Chemistry Undergraduate

researcher

*pseudonyms are used to protect identities

AN EXPERIMENTAL STREAM CHANNEL

TAKING WATER SAMPLES

CONCEPTIONS OF DATA’S VALUE

• Data exhibited primarily an instrumental value

• Value conceptions made up of:

• Assumptions about purposes for specific data

at hand

• Characteristics data needed to exhibit to

meet those ends

• Beneficiaries of data’s value

• Timespan over which data would be valuable

[. . .] if you think about it short-term it almost kind

of seems meaningless. Like sometimes I actually

find myself getting caught up in that. I‟m

like, „Does it really matter what this exact sedge

is?‟ Like if it‟s Juncus balticus or Juncus

nodosus, does it matter? But if you think about it in

long-term, it’s not just about that. [. . .] It’s not about the little identifying plants [. . .] (Brooke, IM-UR).

PURPOSE OF NUS STUDY

• Addressing a gap in knowledge

• How leaf litter affects nutrient uptake in streams

• Supporting Hypotheses

• Nutrient uptake depends on N:P on the leaves

• As the leaves decompose, C:N and C:P

increase and nutrient uptake in the different

leaf treatments becomes more similar

“We're doing it in this situation because we want

to test the mechanism. […] If it wasn‟t a

mechanism-driven question then it wouldn‟t be

appropriate to ask it in this setting.” (Jessica-PI)

The problem […] is that that wasn't the microbes

on the leaves that was [taking up the nutrients]. It

was algae and microbes and all that fine

particulate organic matter. That's why we had to

switch to ground water last Wednesday. Because

that's […] not what we're interested in. That wasn’t

the whole point of why we built all these

experimental channels: to grow algae and fine

particulate organic matter of unknown C to P to N ratios. (Jessica-PI)

TYPE DESIGNATIONS & DATA VALUATION

• Raw vs. Derived Data

• Baseline Data

• Ancillary Data

• Field vs. Controlled Experiment Data

“[. . .] the whole experiment was designed around

two questions, and you don't have other sorts of

variabilities. You don't have differences in

ambient concentrations, or differences in site, or

differences in channel dynamics that . . . You

know, they're all . . . It’s all for the exact same

thing . . . all designed just to answer two questions” (Jessica, PI).

“[…] this is an isolated experiment designed to

answer a simple question. […] it's done in these

artificial stream channels. I think, to a large

extent, the useful life of our actual numbers will

probably end when the paper comes out. If we

were doing something like this in a stream, or like

what we did last year, I think the useful life of that data is a lot longer […]” (Elizabeth-PI)

“[…] they're not comparable, really, to anything

else outside the system that we're working in.”

“I probably would not reach out to them about

this kind of data, because it's an experiment in

these channels as opposed to observations of the

natural system, which I might be more inclined

then to say, „Would you like some component of

this data?‟ because it would contribute to

baseline information or something. (Elizabeth-PI)

FIELD VS. CONTROLLED EXPERIMENT

• Data gathered through the study of a “natural”

system seen as having more broad value

• Could go back to the system

• Could combine with other data gathered

from same place

• Controlled experiment data

• Only valuable within the context

studied, which is transient and deliberately

unnatural

NEXT STEPS

• Cross-case comparison

• Further exploration of categories of data

meaning and those meanings implications

in data practices

Thank you.

dharma.akmon@gmail.com

Twitter: @dharmaakmon