![[Refereed]](http://www.istl.org/graphics/r2.gif){kind=small}
Public Progress, Data Management and the Land Grant Mission: A Survey of Agriculture Researchers' Practices and Attitudes at Two Land-Grant Institutions
Peter Fernandez
Unniversity Libraries
University of Tennessee, Knoxville
Knoxville, Tennessee
pfernand@utk.edu
orcid.org/0000-0002-9731-6567
Christopher Eaker
Unniversity Libraries
University of Tennessee, Knoxville
Knoxville, Tennessee
orcid.org/0000-0001-5881-1680
Shea Swauger
Auraria Library
University of Colorado Denver
Denver, Colorado
orcid.org/0000-0003-0125-5438
Miriam L. E. Steiner Davis
Oak Ridge Associated Universities
Oak Ridge, Tennessee
orcid.org/0000-0002-6454-6411
Abstract
This article reports results from a survey about data management practices and attitudes sent to agriculture researchers and extension personnel at the University of Tennessee Institute of Agriculture (UTIA) and the College of Agricultural Sciences and Warner College of Natural Resources at Colorado State University. Results confirm agriculture researchers, like many other scientists, continue to exhibit data management practices that fall short of generally accepted best practices. In addition, librarians, and others seeking to influence future behavior, may be informed by our finding of a relationship between the land-grant mission and researchers' data management practices.
Introduction
This article examines the current state of research data management and sharing practices and attitudes within colleges of agriculture at two land-grant universities. During a time of relatively rapid change in data sharing practices, this article captures how agriculture researchers at these institutions think about data sharing and in what ways those opinions manifest in their workflows. The results help explain broad trends, such as the relationship between scientific understanding and actual data management practices. By focusing on agriculture researchers, we were able to assess attitudes of a population less reflected in the literature, specifically, the implication(s) of the land-grant mission on their data management practices.
In general, awareness and understanding of the value of sharing scientific data is in a very different place today than it was even a couple of years ago. Librarians and other theorists have long been making the case that access to data can have a transformative effect on the research process (Hey and Trefethen 2003; Gold 2007; Witt 2008). One of the key changes in recent years has been institutional recognition of the benefits of sharing data. Funders, publishers, and research institutions have begun to change how they relate to and share research data (Murphy 2014). Libraries are increasingly hiring data librarians or training existing staff with these skills and publishers and various startups are investing in technology to host researchers' data (Pryor and Donnelly 2009; Heidorn 2011). Perhaps most significantly, this change is evident in the requirements that funders have placed on research data they fund (MacMillan 2014; Murphy 2014). With the combination of land-grant universities' missions to disseminate new knowledge, the increasing momentum of data sharing mandates from research funders, and the increase in tools and opportunities to facilitate more open and networked research practices, we also wanted to examine what role the land-grant mission played, if any, in researchers' evolving practices.
These three research questions guided survey development:
- What are the data sharing and data management practices of land-grant agriculture and natural resources researchers?
- How do these relate to the land-grant mission? (i.e., are there differences between those who see their work supporting the land-grant mission and those who don't?)
- How can librarians help land-grant agriculture and natural resources researchers better meet the land-grant mission?
Literature Review
An increasing consensus is developing that sharing data can produce tremendous downstream and upstream effects (MacMillan 2014). In the United States, in 2013, these theoretical concerns were given significant prominence through a Presidential executive order that directed federal grant funders to require the creation of data management plans for most researchers they fund. During the time we conducted our research in 2014, these changes were still in flux. Part of the value of this work is that it provides a snapshot of researcher opinions during this time.
The value of disseminating research data could have particular saliency for researchers working at land-grant universities, institutions whose mission includes sharing "...the development of practical applications of research knowledge and giving of instruction and practical demonstrations of existing or improved practices or technologies in agriculture..." (United States. Congress. House 1959). A significant goal of the land-grant university is the dissemination of new knowledge, applications, and technologies. Effective management of research data and subsequent sharing of that data are critical to mission success.
The passage of the Morrill Act of 1862 first established land-grant universities. Since then, the perceived responsibility of land-grant universities to share with the public the products of their labor has evolved. Likewise, the scholarly communication system has evolved. Neilson claims recent technological innovations can enable us to propagate knowledge at a prolific scale, and that the larger issues inhibiting this are cultural. He writes, "We're at a unique moment in history: for the first time we have an open-ended ability to build powerful new tools for thought. We have an opportunity to change the way knowledge is constructed. But the scientific community, which ought to be in the vanguard, is instead bringing up the rear..." (Nielsen 2012).
The Morrill Act has been interpreted a variety of ways in different eras and at different institutions, but overall, agriculture tends to be a very prominent part of most land-grant institutions. After the establishment of cooperative extension agencies, these institutions also began to function as government agencies (Ramussen 2002). As such, many of the findings about government employees in Douglass et al. (2014) become equally true for all researchers operating using government funding, particularly those associated with the extension service.
The comprehensive research conducted by the DataONE project captures many of the practices and perceptions of scientists in regard to their data management and sharing (Tenopir et al. 2011). While research such as this provides a great deal of information about the data sharing practices among scientists in general, there is relatively less available concerning discipline- or subject-specific practices (Kirlew 2011; Sayogo and Pardo 2013; MacMillan 2014). Within agriculture in particular, there are studies by both Diekmann (2012) and Williams (2012) that highlight the importance of this disciplinary knowledge. Williams' work focuses on a bibliographic study of the data practices of crop scientists that highlighted a reliance on supplemental files to share data, and also provides a general overview of their practices. Diekmann reported on a more general study conducted in 2008-2009, that provides a good baseline for rapid change in this area. Finally, Bracke (2011) highlights the role that agriculture librarians can play in providing support to researchers in this area. This is particularly important in light of a survey conducted in 2012 that indicated researchers in general often wanted to share their data, but lacked the skills and awareness of the resources their institutions may provide to assist them (Diekema et al. 2014).
By focusing on the practices and attitudes that prevail within a subject or discipline, our research is able to explore these researchers' behavior at a more granular level. Previous studies have noted that data management practices and perceptions can vary significantly by academic discipline (Tenopir et al. 2011). Agriculture provides a particularly compelling test case for investigating an academic discipline area in more depth, because it is a topic that includes a wide variety of departments and subjects. The rate at which agriculture researchers at land-grant institutions adopt data sharing practices could indicate how much, if any, an institution's mission effects the research practices of its members. Understanding the relationship between a researcher's attitude about data sharing and a researcher's practice of data sharing, particularly when there are discrepancies between them, could help institutions develop better research services around data management.
Methods
Participants
This study was an exploratory project seeking to describe the current data management and sharing practices and attitudes of agriculture researchers. The study population included all people involved in agricultural research and/or extension at the University of Tennessee, Knoxville (UTK) Institute of Agriculture (UTIA) and the College of Agricultural Sciences and Warner College of Natural Resources at Colorado State University. We constructed a list from the online directories of names and e-mail addresses of all agricultural research personnel, including faculty, students, research scientists, extension personnel, and administrators. For UTK, this list included all names listed on personnel listings for all departments in UTIA, except for the College of Veterinary Medicine. For CSU, this list included all names listed on personnel listings for the College of Agricultural Sciences and the Warner College of Natural Resources. Our survey was sent to the entire population of 1,862 names and e-mail addresses (787 from UTK and 1,075 from CSU).
Characteristics
Broken down by university affiliation, 61% (n=188) of the respondents were from Colorado State University, and 38% (n=117) were from the University of Tennessee. Less than 1% were from other institutions or a combination of institutions. Broken down by position, students represented the greatest percentage of respondents at 29% (n=88). This can be further broken down by type: 15% doctoral students (n=46), 14% masters students (n=42), and less than 1% undergraduate students (n=2). Most of the students were from CSU (97% n=86) because the personnel listings available on CSU's departmental websites included students while those on UTK's departmental web sites did not. Just over one quarter, 26% (n=82), of the respondents were faculty, which can be further broken down by rank: 9% (n=28) assistant professor, 4% (n=14) associate professor, and 13% (n=40) full professor. Research associates, technicians, and scientists made up 22% (n=67). Extension personnel made up 5% (n=15), and administrators 5% (n=14), of the respondents. Post-doctoral fellows represented 4% (n=12). Fewer than one percent of the respondents were lecturers or instructors.
Classified by primary subject area, participants represented over 40 different disciplines, with the most frequently represented discipline being forestry, wildlife, fisheries or conservation sciences (21%, n = 62). The least frequently represented discipline was agricultural leadership, education or communications (2%, n =6). The total distribution of disciplines is shown in Table 1.
Table 1. Distribution of Disciplines
| Response | n | % |
|---|---|---|
Forestry, Wildlife, Fisheries or Conservation Science(s) |
62 |
21% |
Other, please specify |
40 |
14% |
Plant Science(s) / Crop Science(s) |
37 |
13% |
Animal Science(s) |
25 |
9% |
Agricultural Economics / Natural Resource Economics |
25 |
9% |
Entomology / Pathology / Pest Management |
25 |
9% |
Human Dimensions |
19 |
7% |
Ecosystem Science / Sustainability Science |
17 |
6% |
Soil Science(s) |
13 |
4% |
Horticulture |
11 |
4% |
Food Science or Food Technology |
9 |
3% |
Agricultural Leadership, Education or Communications |
6 |
2% |
Total |
289 |
100% |
It should be noted that the disciplines represented here are a convenience sample that emphasizes the importance of the land-grant mission. It is not a representative sample of all researchers working in the area of agriculture. This limitation is slightly mitigated by partnering to include two land-grant universities rather than one. Agriculture is a broad discipline, and a wide variety of life science (and even social science) disciplines can be affiliated with the land grant mission at universities. The work they do can vary widely (National Research Council (U.S.) and Committee on the Future of the Colleges of Agriculture in the Land Grant University System 1995).
Research Design
The research design, participant criteria, and recruitment procedures were approved through both universities' Institutional Review Boards prior to commencing data collection. Data were collected through online surveys administered to respondents at both universities during the spring semester of 2014 through the University of Tennessee's Qualtrics account. The researchers sent everyone in the target population an initial e-mail, which contained the purpose and description of the study, why they were selected for involvement, and an invitation to complete the survey. The researchers did not perform any probabilistic or non-probabilistic sampling as they attempted to obtain a census. One week after the participants received the initial survey invitation, they were sent a reminder e-mail.
Of the total e-mails sent, 135 from UTK and 36 from CSU were undeliverable, reducing the number of people reached to 1,691. A total of 379 surveys were started, and 325 surveys were completed. Based on completed surveys, the response rate was 325/1,691*100 = 19.2%.
Results
The survey began by asking participants if they worked with research data as part of their job responsibilities. If a respondent answered that they did not work with research data, their survey ended. We were interested only in those people who actively worked with research data. The percentage of people who actively work with data was 87% (n = 310). The results discussed below pertain only to those participants stating they worked with research data.
Data Related Activities
The survey then asked participants to identify all the data-related activities they participated in through their work. The list of activities provided was developed based on the DataONE Research Data Life Cycle (Tenopir et al. 2011). The results indicate that over three-fourths spent their time analyzing data (85%, n = 247), writing up results concerning data (85%, n = 245), planning for data collection (84%, n = 243), and collecting data in the field or laboratory (78%, n = 226). Just slightly less than three-fourths spent time performing data quality control (74%, n = 215), and more than half spent time integrating different data sources (56%, n = 162), and searching for data or datasets (53%, n = 152). Fewer than half of the respondents indicated that they spent time describing and creating metadata for data (48%, n = 138) and preparing data for long term preservation (36%, n = 104).
Type and Format of Data
When asked about the type of data that best described what they typically work with, 48% (n = 134) say they work with experimental data, which was defined as data that involve some degree of manipulation. Slightly fewer, (46%, n = 128) use observational data, such as weather and crop observations. Biotic surveys and social science data (such as interviews, surveys, and focus groups) were next most frequent, 33% (n = 92) and 32% (n = 90), respectively. The remaining data types (data models, abiotic surveys, and remotely sensed data) were used by 20% or less of the respondents (20%, n = 57; 19%, n = 54; and 16%, n = 46, respectively).
Next, survey participants were asked about the format of data they work with. Spreadsheets are the most widely used format for data (86%, n = 241). After spreadsheets, the majority of people use statistical packages, such as SPSS, SAS, or R (63%, n = 178). Just over 50% of responses indicated that they use databases, such as MySQL or Microsoft Access in which they store their data (51%, n = 142). Though slightly less than half (46%, n = 130), a large percentage of people use physical paper and handwritten notes, such as field notes, laboratory notebooks, data collection sheets, and specimen cards. Digital photographs (35%, n = 98), geographic information systems shape files (33%, n = 92), physical specimens and samples (30%, n = 84), digital text documents (28%, n = 80), audio/video files (12%, n = 33), and print photographs (7%, n = 19) were all also used to some degree, in some cases by as many as one-third of respondents.
Data Storage
The survey asked the participants to indicate where they stored their research data, given the following locations: desktop or laptop computer, external device, server, non-digital, or other. They were asked to indicate how much of their data were stored on those locations: None (0%), Some (<50%), Most (>50%), and All (100%). Most people (58.7%, n = 128) indicated that at least some of their data were located in non-digital formats, such as field notes or paper files. Of note for those interested in the long term preservation of data, about half of our participants indicated that all of their data were located on a personal or laboratory desktop or laptop computer (51.9%, n = 140).
Assistance
Nearly 90% of respondents either feel they could benefit from assistance in working with data (65%, n = 173) or were unsure of whether they would benefit from such assistance (23% n = 62). Only 12% (n = 31) felt that they did not need any assistance in working with their data.
For researchers who are unsure of how to manage their data, libraries are only one potential source of information (Table 2). Over half (55%, n = 110) responded that they would seek help from colleagues. Perhaps even more significantly, they were widely divided as to where they would go after colleagues, with the next most common response being designated data managers (10%, n = 19) and libraries tied with the office of information technology (7%, n = 14 and 13, respectively).
Table 2: Seek Assistance with Data Management
| Response | n | % |
|---|---|---|
Colleagues |
110 |
55% |
Designated data managers |
19 |
10% |
Office of Research and Engagement(including the Office of Sponsored Programs) |
13 |
7% |
Office of Information Technology |
13 |
7% |
University Libraries |
14 |
7% |
Professional associations |
9 |
5% |
Conference workshops |
8 |
4% |
Other, please specify |
7 |
4% |
Administrative offices (deans, department heads, provost, etc) |
6 |
3% |
Data organizations, such as DataONE or data repositories |
1 |
1% |
Total |
200 |
100% |
In this context, it is worth noting that both universities surveyed employed full-time librarians whose job duties included helping to providing data management guidance and instruction, although most respondents were unaware of this service. The percentage of respondents who would seek help from the library was significantly lower than the percentage of respondents who were aware that their libraries offered services in this area. When asked about entities that provided training on data management, a total of 34% (n = 62) mentioned the library as a potential source of training.
Advancement of Science/Land-grant Mission
In addition to these general questions about the state of agriculture researchers' data sharing and management practices, our survey also gathered information about the relationship between their motivation (if any) for sharing data, and their actual data sharing practices. Overall, their responses to this question were centered between various entities that may require or encourage sharing, and other reasons that relate to the advancement of science (Table 3).
Table 3. Data Sharing Motivations
| Response | N | % |
|---|---|---|
Others may be able to use my data to answer their research questions |
125 |
65% |
It is good practice to share research data |
125 |
65% |
Data sharing supports the land-grant mission |
92 |
48% |
Solving environmental problems requires access to multiple datasets |
66 |
34% |
The funder REQUIRES it |
41 |
21% |
The funder ENCOURAGES it |
40 |
21% |
My university or department ENCOURAGES it |
36 |
19% |
The journal where I published results REQUIRES it |
19 |
10% |
The journal where I published results ENCOURAGES it |
19 |
10% |
My university or department REQUIRES it |
13 |
7% |
Other, please specify |
10 |
5% |
To better understand this relationship, we condensed the responses, "Solving environmental problems requires access to multiple datasets," "Others may be able to use my data to answer their research questions," "It is good practice to share research data," and "Data sharing supports the land-grant mission" into a single variable about the general advancement of science, as opposed to pressure from institutions or funders. These responses may indicate that respondents had, to some degree, internalized the value of sharing data.
There was a significant association between this new variable and the likelihood that the respondent would actually make their data available to others (Pearson Chi-square=86.79; degrees of freedom=3; exact p-value<.001). Within the overall population the responses to the question, "In the past, how much of your data have you made available to others?" were: All (12%, n=34); Most (22%, n = 60); Some (37%, n = 101); None (10%, n = 27); and Not applicable/Not my decision to share data (19%, n= 51) (Table 4).
Respondents who indicated that data sharing supports science all made at least some of their data available when they were in a position to make that determination. None of those respondents failed to share at least some of their data. Only about 54% of those who shared data because of other factors actually shared their research data.
Table 4. Supports Science and Data Made Available Crosstabulation
Data Made Available |
Total |
||||||
|---|---|---|---|---|---|---|---|
None |
Some |
Most |
All |
||||
Data Sharing |
Supports science |
Count |
0 |
82 |
54 |
27 |
163 |
Expected Count |
19.8 |
74.2 |
44.1 |
25.0 |
163.0 |
||
% within Data Sharing |
.0% |
50.3% |
33.1% |
16.6% |
100.0% |
||
Adjusted Residual |
-9.2 |
2.4 |
3.4 |
.9 |
|||
Does not support science |
Count |
27 |
19 |
6 |
7 |
59 |
|
Expected Count |
7.2 |
26.8 |
15.9 |
9.0 |
59.0 |
||
% within Data Sharing |
45.8% |
32.2% |
10.2% |
11.9% |
100.0% |
||
Adjusted Residual |
9.2 |
-2.4 |
-3.4 |
-.9 |
|||
Total |
Count |
27 |
101 |
60 |
34 |
222 |
|
Expected Count |
27.0 |
101.0 |
60.0 |
34.0 |
222.0 |
||
% within Data Sharing |
12.2% |
45.5% |
27.0% |
15.3% |
100.0% |
||
When looking only at the land-grant mission, rather than the broader effect of scientific advancement, this same effect was observed (Table 5). All those who think data sharing supports the land-grant mission do in fact share their data, while about 82% (n=104) of respondents who do not agree that data sharing supports the mission make data available to others. This effect was most pronounced among those who shared most or all of their data.
Table 5. Supports Land Grant Mission and Data Made Available Crosstabulation
Data Made Available |
Total |
||||||
|---|---|---|---|---|---|---|---|
None |
Some |
Most |
All |
||||
Supports Land Grant Mission |
No |
Count |
27 |
60 |
28 |
16 |
131 |
Expected Count |
15.9 |
59.6 |
35.4 |
20.1 |
131.0 |
||
% within Supports Land Grant Mission |
20.6% |
45.8% |
21.4% |
12.2% |
100.0% |
||
Adjusted Residual |
4.6 |
.1 |
-2.3 |
-1.5 |
|||
Yes |
Count |
0 |
41 |
32 |
18 |
91 |
|
Expected Count |
11.1 |
41.4 |
24.6 |
13.9 |
91.0 |
||
% within Supports Land Grant Mission |
.0% |
45.1% |
35.2% |
19.8% |
100.0% |
||
Adjusted Residual |
-4.6 |
-.1 |
2.3 |
1.5 |
|||
Total |
Count |
27 |
101 |
60 |
34 |
222 |
|
Expected Count |
27.0 |
101.0 |
60.0 |
34.0 |
222.0 |
||
% within Supports Land Grant Mission |
12.2% |
45.5% |
27.0% |
15.3% |
100.0% |
||
Respondents' motivations had less of an effect on the rest of their data management practices. For instance, sharing data can become significantly more impactful when that data is accompanied by useful metadata. For the overall population 17% (n = 47) of respondents created metadata for all of their data, 54% (n= 151) created metadata for some of their data, 11% (n = 31) created metadata for none of their data, and 18% (n = 49) were not certain. For both the advancement of science (Pearson Chi-square=2.846;), as well as for the land-grant mission (Pearson Chi-square=3.927), there was not a statistically significant relationship between their motivations, and the amount of metadata they created.
By far, the most popular method of sharing data within our population was by request from colleagues at 87% (n = 166) (Table 6). In comparison all of the different methods of sharing data through the internet (FTP, Data Repository and through the web) add up to 63% (n = 119).
Table 6. Methods of Data Sharing
| Response | N | % |
|---|---|---|
Colleagues can request the data from me |
166 |
87% |
It can be downloaded from the web |
51 |
27% |
It is viewable online |
30 |
16% |
It can be downloaded from a data repository |
28 |
15% |
Other, please specify |
12 |
6% |
It can be downloaded from an FTP site |
10 |
5% |
Given the significant number of respondents who share their data through these informal means, we tested to see if motivations would play a role in how likely respondents were to share their data openly. For scientific advancement in general (Pearson Chi-square=1.936; degrees of freedom=2; exact p-value=.316) and the land-grant mission in particular (Pearson Chi-square=3.387; degrees of freedom=2; exact p-value=.303), there was no significant relationship.
This relationship between motivation and how data is shared was also a consideration in a separate question when respondents were asked to complete a sentence about their data (Table 7), indicating that either "It is freely and openly available to everyone" or "It is available to others only under certain conditions." In the overall population, 33% (n = 65) made their data available to everyone, and 67% (n = 130) made it available with conditions. Respondents motivated by scientific advancement were not significantly more or less likely to share their data freely (Pearson Chi-square=3.664; degrees of freedom=1; exact p-value=.066). However, almost 43% of respondents who agreed that data sharing supports the land-grant mission indicated that they made their data freely accessible, while only about 25% of those who did not think data sharing supports the land-grant mission had freely accessible data.
Table 7. Supports Land-grant Mission and Availability of Data Crosstabulation
Availability of Data |
Total |
||||
|---|---|---|---|---|---|
Freely |
Conditional |
||||
Supports Land-grant Mission |
No |
Count |
26 |
78 |
104 |
Expected Count |
34.7 |
69.3 |
104.0 |
||
% within Supports Land-grant Mission |
25.0% |
75.0% |
100.0% |
||
Adjusted Residual |
-2.6 |
2.6 |
|||
Yes |
Count |
39 |
52 |
91 |
|
Expected Count |
30.3 |
60.7 |
91.0 |
||
% within Supports Land-grant Mission |
42.9% |
57.1% |
100.0% |
||
Adjusted Residual |
2.6 |
-2.6 |
|||
Total |
Count |
65 |
130 |
195 |
|
Expected Count |
65.0 |
130.0 |
195.0 |
||
% within Supports Land-grant Mission |
33.3% |
66.7% |
100.0% |
||
When asked to provide more details about the conditions under which they would share their data, respondents gave a variety of responses. Once these free-text responses were coded, the most common responses were "collaboration" (n = 32), "by request" (n = 20), "with approval" (n = 13), and "not until publication" (n = 9). Although the question was about the conditions under which they would share data, 11 respondents also mentioned a significant barrier to sharing data, which was that it contained sensitive information.
Discussion and Conclusions
The agriculture researchers we surveyed have shared at least some or most of their data with others. Yet, while they have some sense that there are higher ideals at stake, they also tend to view their data in practical terms. They share data most frequently only under certain conditions rather than freely. Even when they want to share data, they experience a wide variety of barriers, including time and knowledge. Librarians are ideally suited to help them overcome these barriers, even as the evidence shows that librarians are frequently not positioned to be the first place researchers at these universities turn to for help.
In general, participants in our survey who considered sharing data to be part of advancing science or the land-grant mission were more likely than others to actually share their data. As reported in our results, they framed this motivation in a variety of ways, ranging from a general understanding of it as a good practice, to recognizing the need for access to multiple datasets to solve environmental problems. Significantly in the context of agricultural research, every single researcher who indicated that the land-grant mission motivated their data management practice actually shared their data.
This finding highlights the importance of the land-grant mission to understanding and influencing at least some agriculture researchers. As we have previously discussed, the land-grant mission represents a particular articulation of the obligations researchers have to the larger world. Future research might explore in more depth how salient this articulation of the obligation researchers have to the wider public is, and if this relationship differs at different institutions. For librarians, this relationship further reinforces the idea that convincing researchers about the importance of sharing data for scientific progress broadly or the land-grant mission in particular, can have an impact on the actual practices of those scientists.
Our survey does not demonstrate what the relationship between these two factors is. It may be the case that highlighting how sharing data can further the land-grant mission could inspire other agriculture researchers to share their data. However, it may also be the case that those who are already inclined to share their data are simply more likely to understand their practice in the context of the land-grant mission. Similar conclusions can be drawn for any of the answers that were coded as the advancement of science in our results. The one notable exception is that respondents motivated by the land-grant mission were likely to state that they share their data without conditions. However, those motivated by scientific advancement did not have a significant correlation between those two statements.
This is a potentially important finding for librarians seeking to influence researchers because it highlights a connection between a particular understanding of the mission of the university and the researchers' behavior. This is a potential leverage point for facilitating change, and suggests (although does not prove) that institutions might be able to influence data sharing practices by encouraging this understanding of both the mission and its connection to data sharing practices. It should be noted however, that it could be that those who are already inclined to share data for other reasons might instead also be drawn to understand the intuitions' mission in a particular way. Either way, it remains a potential rhetorical tool for librarians and other institutional actors.
Because only 32% of these researchers were willing to share their data freely and without conditions, what those conditions are becomes particularly important. The most common response, collaboration, is a worthwhile goal but does not necessarily lead to sharing of data more broadly. It is particularly notable since the second and third most commonly cited reasons were "by request" and "with approval." Although not conclusive, all of these responses are consistent with researchers acting as gatekeepers for their data and only allowing it to be used by collaborators or those with whom they have existing relationships. This supports existing research on researchers at the National Science Foundation Science and Technology Center, that identifies "the gift culture of scholarship, in which goods are bartered between trusted colleagues rather than treated as commodities." (Wallis et al. 2013).
Similarly, our results also indicate the majority of our respondents did not create and share quality metadata. This was equally true regardless of what the motivations of the respondent were for sharing their data. This indicates that while many researchers recognize the value of sharing data, best practices established for sharing data have not fully integrated themselves into their behaviors. For librarians who would like to influence this behavior, closing this gap between motivations and practices is an important goal, as these best practices would make it possible for data to be easily shared with the widest possible audience.
Although this survey was not designed to provide a comprehensive answer to the question of why these behaviors occur, a few answers do suggest some of the barriers. Overall, the number of respondents who are responsible for describing and preserving data is almost half that of those who are responsible for analyzing and collecting data. Describing and preserving data is time-consuming work that is vital to sharing data but is less directly connected to existing reward mechanisms, such as publications. Given the number of researchers who only wish to share data when asked by colleagues, it is clear that informal networks along with traditional publication and citation practices continue to be very important to these researchers. While they may, or may not, have larger ideals, the community of researchers they know, and the publication processes they are familiar with, also have an important role to play in their actions.
For librarians seeking to affect the behaviors of agriculture researchers, these findings offer a number of intriguing possibilities. First, researchers are somewhat aware that librarians can be of assistance, since 34% (n = 62) mentioned the library as a potential source of training. Libraries were also ranked ahead of almost every other potential source of assistance except for colleagues, but still were only ranked at 7% (n = 11). Libraries are a potential source of assistance, but not the place researchers would likely turn. Given the importance of colleagues, librarians could take a creative approach, potentially drawing from social network analysis research. They could identify influencers within their departments, i.e., those colleagues who others turn to for assistance. If those key influencers could be convinced to refer their colleagues to the library for training, it could dramatically change the dynamic of librarian interactions. Additionally, by identifying and providing those key influencers with tools and training, librarians could have a disproportionate effect on how others in the department behave. Seen in this light, it may not be as important to conduct widespread training, as to train the right people.
However they approach researchers, being aware of how they think about their data can be a valuable asset. Agriculture researchers may share many features with other scientists, but they continue to be motivated by collaboration and their colleagues within their discipline. This is true in their motivations for sharing data and supported by how they prefer to share data, as well as to whom they reach out for help. As institutions and funders continue to evolve how they relate to the data management lifecycle, there will be many opportunities to assist researchers. Land-grant universities are institutions that have a mission to serve the public good, in part by creating and disseminating knowledge. Many researchers and librarians have inferred a strong connection between that mission and data sharing best practices. By acknowledging and honoring where these researchers currently are, it becomes easier to see how to influence additional researchers to share data in ways that allow them to effectively and easily serve their own interests, and hopefully align those actions with practices that benefit the general public, as well as their research community.
Additional Resources
The data, survey questions, and data dictionary from this research project are publicly available and may be downloaded from Trace, Tennessee Research and Creative Exchange, at the following DOI: http://dx.doi.org/10.7290/V7KS6PHQ.
References
Bracke, M.S. 2011. Emerging data curation roles for librarians: a case study of agricultural data. Journal of Agricultural & Food Information 12:65-74.
Diekema, A.R., Wesolek, A., Walters, C.D. 2014. The NSF/NIH effect: surveying the effect of data management requirements on faculty, sponsored programs, and institutional repositories. The Journal of Academic Librarianship 40:322-331.
Diekmann, F. 2012. Data practices of agricultural scientists: results from an exploratory study. Journal of Agricultural & Food Information 13:14-34.
Douglass, K., Allard, S., Tenopir, C., Wu, L., Frame, M. 2014. Managing scientific data as public assets: Data sharing practices and policies among full-time government employees. Journal Associationn for Information Science and Technology 65:251-262.
Gold, A.K. 2007. Cyberinfrastructure, data, and libraries, part 2: libraries and the data challenge: roles and actions for libraries. D-Lib Magazine 13. [accessed 2015 Oct 18]. Available from: http://dlib.org/dlib/september07/gold/09gold-pt2.html
Heidorn, P.B. 2011. The emerging role of libraries in data curation and e-science. Journal of Library Administration 51:662-672.
Hey, T. & Trefethen, A. 2003. The Data Deluge: An e-Science Perspective. In: Grid Computing: Making the Global Infrastructure a Reality. Wiley and Sons. p. 809-824.
Kirlew, P.W. 2011. Life science data repositories in the publications of scientists and librarians. Issues in science and technology librarianship 5. DOI: 10.5062/F4X63JT2
MacMillan, D. 2014. Data sharing and discovery: what librarians need to know. The Journal of Academic Librarianship 40:541-549.
Murphy, F. 2014. Data and scholarly publishing: the transforming landscape. Learned Publishing 27:3-7.
National Research Council (U.S.), Committee on the Future of the Colleges of Agriculture in the Land Grant University System. 1995. Colleges of Agriculture at the Land Grant Universities: A Profile. Washington D.C.: National Academy Press.
Nielsen, M. 2012. Reinventing Discovery: The New Era of Networked Science. Princeton University Press.
Pryor, G. & Donnelly, M. 2009. Skilling up to do data: whose role, whose responsibility, whose career? International Journal of Digital Curation 4:158-170.
Ramussen, W.D. 2002. Taking the university to the people: Seventy-five years of cooperative extension. Purdue University Press.
Sayogo, D.S. & Pardo, T.A. 2013. Exploring the determinants of scientific data sharing: Understanding the motivation to publish research data. Government Information Quarterly 30, Supplement 1:S19-S31.
Tenopir, C., Allard, S., Douglass, K., Aydinoglu, A.U., Wu, L., Read, E., Manoff, M. & Frame, M. 2011. Data sharing by scientists: practices and perceptions. PLoS ONE 6:e21101. DOI: 10.1371/journal.pone.0021101
United States. Congress. House. 1959. Hearings, report, and debate : Smith-Lever act of 1914. Blacksburg, Va: Reproduced by the Virginia Agricultural Extension Service.
Wallis, J.C., Rolando, E. & Borgman, C.L. 2013. If we share data, will anyone use them? Data sharing and reuse in the long tail of science and technology. PLoS ONE 8:e67332. DOI: 10.1371/journal.pone.0067332
Williams, S.C. 2012. Data practices in the crop sciences: a review of selected faculty publications. Journal of Agricultural & Food Information 13:308-325.
Witt, M. 2008. Institutional repositories and research data curation in a distributed environment. Library Trends 57:191-201.
| Previous | Contents | Next |
This work is licensed under a Creative Commons Attribution 4.0 International License. 