Journal of Nanoparticle Research

, 15:1381

Tweeting nano: how public discourses about nanotechnology develop in social media environments

Authors

    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Sara K. Yeo
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Michael Cacciatore
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Dietram A. Scheufele
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Dominique Brossard
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Michael Xenos
    • Department of Communication ArtsUniversity of Wisconsin
  • Ashley Anderson
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Doo-hun Choi
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Jiyoun Kim
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Nan Li
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Xuan Liang
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Maria Stubbings
    • Department of Life Sciences CommunicationUniversity of Wisconsin
  • Leona Yi-Fan Su
    • Department of Life Sciences CommunicationUniversity of Wisconsin
Research Paper

DOI: 10.1007/s11051-012-1381-8

Cite this article as:
Runge, K.K., Yeo, S.K., Cacciatore, M. et al. J Nanopart Res (2013) 15: 1381. doi:10.1007/s11051-012-1381-8

Abstract

The growing popularity of social media as a channel for distributing and debating scientific information raises questions about the types of discourse that surround emerging technologies, such as nanotechnology, in online environments, as well as the different forms of information that audiences encounter when they use these online tools of information sharing. This study maps the landscape surrounding social media traffic about nanotechnology. Specifically, we use computational linguistic software to analyze a census of all English-language nanotechnology-related tweets expressing opinions posted on Twitter between September 1, 2010 and August 31, 2011. Results show that 55 % of tweets expressed certainty and 45 % expressed uncertainty. Twenty-seven percent of tweets expressed optimistic outlooks, 32 % expressed neutral outlooks and 41 % expressed pessimistic outlooks. Tweets were mapped by U.S. state, and our data show that tweets are more likely to originate from states with a federally funded National Nanotechnology Initiative center or network. The trend toward certainty in opinion coupled with the distinct geographic origins of much of the social media traffic on Twitter for nanotechnology-related opinion has significant implications for understanding how key online influencers are debating and positioning the issue of nanotechnology for lay and policy audiences.

Keywords

NanotechnologySocial mediaTwitterPublic opinionOnlinePolicy

Nanotechnology opinion in the U.S

On September 4, 2010, Google launched a doodle in honor of the 25th anniversary of the discovery of the buckyball. It was an important anniversary for nanotechnology research, but most Americans logging onto Google that day were probably unfamiliar with the reason behind the novel icon. Compared to their European counterparts, the U.S. public continues to be relatively unaware of nanotechnology and, as a result, less likely to attend to nanotechnology news or hold strong nanotechnology opinions (Scheufele and Lewenstein 2005). Even though nanotechnology is used in more than a thousand consumer products in the U.S. (The Project on Emerging Nanotechnologies 2012), ranging from sunscreen to squash racquets, 77 % of Americans reported that they had heard “just a little” or “nothing at all” about nanotechnology as late as 2010 (Smith et al. 2011).

Since the discovery of the buckyball, communication researchers have had the opportunity to study opinion development as awareness of nanotechnology slowly diffuses through the American public. Over the past decade scholars have consistently found the U.S. public to be largely unaware but also free of a strong negative attitude toward the technology. In 2002, only 9 % of respondents in a U.S. Internet survey agreed with the statement “Our most powerful 21st century technologies—robotics, genetic engineering, and nanotechnology—are threatening to make humans an endangered species” (Bainbridge 2002). In 2004, 83 % of respondents reported feeling hopeful about nanotechnology, and only 26 and 18 % claimed to feel worried or angry about it (Cobb and Macoubrie 2004). More recently, in 2010, awareness levels continued to remain low and the U.S. public still had not developed strong negative attitudes; 37 % of those surveyed thought that the benefits of nanotechnology will outweigh the harmful results, 45 % did not know and only 9 % thought the harmful results would outweigh the benefits (Smith et al. 2011).

However, it would be wrong to assume that low levels of knowledge and awareness mean that Americans are withholding judgment. Recent research has concluded that complex mechanisms involving strongly held values and other types of heuristic cues, such as those suggested by media portrayals or strongly held religious beliefs, help individuals form opinions about nanotechnology even in the absence of knowledge about the issue. For instance, while science media use and perceived benefits are related to positive attitudes toward nanotechnology, perception of risk and strong religious beliefs are negatively related to its acceptance. More importantly, factual knowledge plays a role in influencing acceptance for the technology only among those individuals lacking strong religious beliefs (Brossard et al. 2009).

These findings are the result of a nationally representative survey, and the literature is not settled on how survey results like this translate to unsolicited, spontaneous expression in the regular world. We do not reliably know how often sentiment toward nanotechnology is expressed spontaneously, and whether spontaneous expression is optimistic, or if those expressing sentiment are confident or certain in their language. The purpose of this paper is to bridge the gap by large-scale textual analysis to examine nanotechnology-related social media messages expressing opinion in the microblogging service Twitter. Chosen because of the spontaneous and unsolicited nature of its messages, Twitter should provide evidence that would further broaden understanding of how nanotechnology-related sentiment is expressed. Before explaining our methods, we give an overview of the literature relevant to our study.

Media, nanotechnology … and Twitter

What different American subpublics know about nanotechnology tends to co-vary with individual levels of media use. Online science news consumers are different from the general population. Recent research has shown that 54.9 % of science Internet users have earned college degrees, almost twice the rate of the general U.S. population, and nearly all have completed high school. They are more evenly distributed across age ranges than other media users: 26.4 % of science Internet users are 18–34 years old, 28.6 % are 45–54 years old, and 25.3 % are above 55, and slightly more likely to be male (51.9 %) than female (Anderson et al. 2010). Well educated and generationally diverse, these media users appear to have a preference for interactive forms of media when consuming science news.

National surveys on public awareness and support for nanotechnology highlight the importance of online sources for informing the general public about this area of science (Lee and Scheufele 2006). This trend is corroborated by the most recent iteration of the National Science Board’s Science and Engineering Indicators study (National Science Board 2012), which showed that 59 % of all Americans routinely turn to the Internet as their primary source when they are looking for information about specific scientific topics. Still, in an era of concurrent increases in online news use and decreases in newspaper readership, it is not clear if these individuals are shifting attention from print newspapers to online versions of the same or instead seeking entirely new sources of online news. Perhaps more importantly for this study is the finding that more than one in four Americans now report that they “regularly or sometimes get news or news headlines through Facebook, Twitter or other social networking sites” (Pew Research Center 2011b).

Because this is a relatively recent phenomena, the existing literature on online news use does not take into account the potential changes in patterns of exposure resulting from use of social networks for dissemination in contrast to more traditional forms of disseminating online news. The major academic works in this area (Tewksbury 2005, 2001; Sunstein 2007) were written either before or at the launch of the major social networks (Twitter and Facebook) and consequently could not adequately examine the possibility that these new networks would change online behavior. Unlike typical online monitoring of news, which depends on the user selecting specific online news outlets or blogs, Twitter users follow persons or organizations, and the topics that are covered are often issue-based, e.g., nanotechnology, and not media based. In addition, when users conduct topical searches on Twitter they are not guaranteed search results specific to their intent. For example, Twitter users searching for information on commercial products using the nanotechnology-related keywords “buckyball” and “nano” are likely to receive nanotechnology-related results along with results for buckyball magnets and iPod nanos. It is in this way that incidental exposure to nanotechnology seems plausible for Twitter users. Tewksbury et al. (2001) found that incidental exposure to online news as a positive predictor of current affairs knowledge in his 1998 study, and presumably this would be true for science news as well as general news. While this study is nearly 15-years old, the cognitive mechanisms underlying this finding are still valid. However, the concern has been that as online media outlets become increasingly specialized, and audiences become more selective, the online public will be exposed to less commonly read content (Tewksbury 2005). To date, there has been no conclusive evidence that social networks, like Twitter, actually result in greater content specialization for users. Based on the experience of this research team in conducting nanotechnology keyword searches during the project, it seems highly unlikely that nanotechnology keyword searches produce a “pure” result devoid of incidental exposure to unrelated concepts that share overlapping keywords.

This has a couple of implications: First, the literature suggests that science Internet users with nanotechnology knowledge are intentional in their use of the Internet to learn and communicate about nanotechnology. Second, provided that there is a nanotechnology presence on social media, the activity of nanotechnology-specific science users on social media could expose others in their networks to nanotechnology news. Third, searches using nanotechnology-related language, like “buckyball” or “nano,” could result in exposure to nanotechnology news even if the intended search was unrelated to the subject, and the social media user does not have a nanotechnology-specific science user in his network; extrapolating from Tewksbury, in this scenario, the chances of incidental exposure increases the more an individual is online in social networks, and the result is that such exposure could yield an increase in awareness about nanotechnology (Tewksbury et al. 2001).

We may be able to make a logical case for increased exposure to nanotechnology through social networks, but there is little evidence in the literature to support such an argument. Few, if any, studies focus on the use of social networking as a medium for science news users searching or attending to news about a specific topic, like nanotechnology. It is on one of these networks, Twitter, that we will focus.

Our study focus: Twitter and nanotechnology

Twitter is unique among Internet-based communication channels. With a limit of 140 characters per message, Twitter’s content consists of short bursts of text that allow users to conduct public asynchronous conversations through the use of websites, mobile Internet devices, and SMS. Twitter reports over 140 million active, unique users worldwide who send more than 340 million tweets per day. Search queries by active and non-active users exceed 1.6 billion per day, worldwide (Twitter 2012). Twitter users are typically young: 26 % of Internet users aged 18–29 use Twitter, nearly double the rate of those aged 30–49 (Smith and Brenner 2012). African-American audiences have flocked to Twitter as well with 28 % of African-American Internet users utilizing the social media platform (Smith and Brenner 2012). While it is not clear how many tweets are actually attended to on a daily basis, the service has become a critical hub for linking anonymous users to one another to form collective actions, communicate emergencies, or disseminate news that is not widely covered in the mainstream media. Communication scholars are increasingly recognizing and studying its role in events in facilitating spontaneous, real-time communication to interested, but otherwise disconnected individuals for events as diverse as the Arab Spring and American tornado warnings. (Christensen 2011; Hughes and Palen 2009; Lasorsa 2011; Murthy 2011; Papacharissi and de Fatima Oliveira 2012). Nanotechnology-related traffic may not have the urgency associated with crisis events, but it is a topic that is only modestly covered in U.S. news (Dudo et al. 2011a, b; Friedman and Egolf 2005). Like the Arab Spring and American tornado warnings, the relative lack of accessible and timely mainstream news coverage, and the possible desire among nanotechnology partisans to communicate with one another, could make Twitter a platform for disseminating information and creating connections.

Nanotechnology has been a popular issue for content analyses (Cacciatore et al. 2012; Dudo et al. 2011a, b; Weaver and Bimber 2008), but there is very limited data on how nanotechnology or other developing technologies have emerged in the rapidly expanding world of Twitter. More particularly, there is very little empirical research on the types of nanotechnology-related content that audiences are exposed to in these emerging online environments. We know from research on other scientific topics and controversies that the nature of coverage often follows fairly well-defined issue cycles (Brossard et al. 2004; Nisbet et al. 2003). For example, early coverage of stem cell technology was driven by an initial excitement about the scientific potential and the economic impacts of the technology and later shifted to more cautious frames surrounding regulations, unintended consequences, and science–public disconnects (Nisbet and Scheufele 2009). Content analyses of nanotechnology suggest that media coverage of the issue has moved beyond the initial period of excitement and has begun to incorporate frames related to risks and other potentially undesirable consequences of nanotechnology (Dudo et al. 2011b). As a result, coverage has become increasingly balanced in terms of mentions of risks and benefits (Cacciatore et al. 2012; Dudo et al. 2011b). Indeed, there is even evidence that nanotechnology risk information is appearing to a greater extent than benefit information in the online realm (Anderson et al. 2010). However, how such trends in nanotechnology coverage have translated to the world of social media, and whether the trend is repeating that seen for stem cell technology, is difficult to predict. Therefore, we propose the following research question:

RQ1

What is the opinion valence in nanotechnology-related Twitter tweets?

Overall, our empirical understanding of the nature of online content about nanotechnology is much more limited than our understanding of traditional content (Ladwig et al. 2010). Previous studies of nanotechnology media coverage have shown that research is the most dominant conceptual theme present in print news stories about the issue (Dudo et al. 2011b), and environmental themes are more dominant in online news (Cacciatore et al. 2012). Uncertainty themes, one of the themes we are interested in for this study, have been relatively absent from both online and print news coverage of the issue (Cacciatore et al. 2012; Dudo et al. 2011b). We conceptualize “uncertainty” as tweets referring to unclear or unknown consequences related to nanotechnology. Given the absence of uncertainty themes in previous studies of this issue, we can predict with some confidence that Twitter tweets will be relatively more certain than uncertain in their portrayals of nanotechnology. With this information in mind, we propose the following hypothesis:

H1

There is a higher proportion of certainty tweets than uncertainty tweets.

The origins of nanotechnology discourse on Twitter is a second important question. By examining the strands of discourse driven by corporate Twitter accounts, nanotechnology activists, citizens, and scientists, among others, we can map the emergence of different themes of conversations and—maybe more importantly—their geographic origins around federally funded nanotechnology research centers or networks, as well as other research centers. In 2012, the President’s Council of Advisors on Science and Policy (PCAST) recommended that “[a]n effective program in societal implications would have … clearly articulated outcomes as well as plans for assessing and evaluating those outcomes” (President’s Council of Advisors on Science and Technology 2010, p. 15).

This paper also provides the first systematic analysis of how federal funding may have helped shape the discourse surrounding nanotechnology. Specifically, we would expect to see Twitter tweets clustering around technology centers or regions with higher levels of education, where science Internet users are more likely to be exposed to emerging technologies and have a greater chance of possessing nanotechnology knowledge. Based on these considerations, we put forth the following hypothesis.

H2

States with a federally funded National Nanotechnology Initiative center or network will have higher incidences of tweets about nanotechnology.

Methods

Using automated nonparametric content analysis software, Crimson Hexagon Forsight, we collected and analyzed a census of all Twitter tweets posted between September 1, 2010 and August 31, 2011. A total of 495,195 English-language nanotechnology-related tweets expressing opinions were collected during this time period. Crimson Hexagon is designed to collect all tweets during the specified time period; therefore, n = 495,195 represents the whole population of English-language nanotechnology-related tweets expressing opinion communicated through Twitter between September 1, 2010 and August 31, 2011, based on key words used in comparable analyses of traditional news content (Dudo et al. 2011b). When available, the Forsight program collects the geographic origins of tweets. 186,348 of the collected tweets were geotagged, enabling us to map the volume and proportion of responses by state for a portion of the census.

Our analyses gauged the volume and tone of nanotechnology content on Twitter and, when possible, the geographic origins of collected tweets. Crimson Hexagon uses algorithms to automatically track linguistic patterns—representative of various underlying concepts and identified by human coders—across large amounts of textual data (Hopkins and King 2010). Researchers in the computer sciences have used such analysis, often known as sentiment analysis or opinion mining, to assess opinions in a range of contexts, such as discussion threads in online news websites, U.S. Congressional floor debate, blogs, and public comments on proposed regulations (Bansal et al. 2008; Chmiel et al. 2011; Conrad and Schilder 2007; Kwon et al. 2006; also see Pang and Lee 2008 for a broad overview of sentiment analysis). Based on a carefully constructed keyword search, a series of Twitter tweets are first randomly pulled from public Internet content and then analyzed by trained coders.

Our goal was to categorize the proportion of social media messages into five categories along two dimensions. The first dimension identified attitude toward nanotechnology as optimistic–neutral–pessimistic, and the second dimension distinguished between certainty and uncertainty (Table 1). Optimism was conceptualized as language indicating the positive or beneficial outcome related to the technology, or positive commentary on the technology. Neutral was conceptualized as language indicating no judgment relative to a positive or negative outcome, and pessimistic was conceptualized as language indicating a negative or harmful outcome related to nanotechnology or a negative judgment on the technology. Certainty was conceptualized as language indicating clear or known consequences related to nanotechnology, and uncertainty was conceptualized as language indicating unclear or unknown consequences. Table 1 includes sample nanotechnology–related tweets coded according to our conceptualizations of optimist–neutral–pessimistic and certain–uncertain. Results were based on word stem patterns, and each tweet was included and analyzed in its entirety.
Table 1

Examples of coded nanotechnology-related tweets expressing opinion on Twitter between September 1, 2010 and August 31, 2011

 

Certain

Uncertain

Optimistic

New blog post: UNC study sends nano-particles to attack ovarian cancer http://t.co/rJ5TlT6

Breast cancer treatment | nanoscale targeting may improve breast cancer treatment http://t.co/BDK7SeA

[AU] $1.1mn funding coup for Sunshine Coast nanotechnology research to improve sustainable energy technologies http://t.co/7thhYKH

:) nano New Spire patent could attack cancer cells http://t.co/KqnbRti

Neutral

Silicide-induced multi-wall carbon nanotube growth on silicon nanowires http://dlvr.it/DXvQR

What is “frozen smoke,” aka “multiwalled carbon nanotube (MCNT) aerogel?” (gizmag) http://bit.ly/h413t5

Nanotubes… so thin and all tangled up together. #MakingStuff fibrous.

Update, What’s the most impressive nanotechnology that exists today? - http://tinyurl.com/4pqpokk

Pessimistic

Escaped nanoparticles hazardous to crops, says study http://t.co/gwoucyk #nanoagro #nanoproduction #wheat #nanocrops

American Chemical Society Podcast: questions about the safety of nanoparticles in food crops http://t.co/fgK7rOx #nanotechnology

Fearing “nano-cyborgs,” international terrorist group targets nanotech researchers http://t.co/qGkoDi7 #greygoo

Poll: Do you think Nanotechnology poses a threat to the environment and human health? http://tiny.cc/9acga

During an initial software programing period, a total of 10 University of Wisconsin Life Sciences Communication graduate students worked in interrelated, rotating teams of three to train the software and manually code Twitter tweets into the five sentiment categories. A sixth category, “not relevant,” was used to program the software to distinguish between nanotechnology-related tweets and those that used nanotechnology language to refer to unrelated concepts. This important category allowed us to eliminate mentions of iPod nanos, buckyball magnets and other non-nanotechnology-related concepts from the coded tweets used in the final analysis.

Programing occurred over a three-month period during which the software drew random samples of nanotechnology language tweets from Twitter. Tweets were coded in whole rather than truncates. A single codebook containing rules for opinion categorization was shared by the interrelated teams, and all team members attended a weekly meeting to discuss any ambiguous tweets. This continued until the Crimson Hexagon software had a statistically reliable number of coded word tweets in each dimension. An initial trial for validity and reliability was conducted. After adjustments and additional verification, the programing was considered complete and the census of tweets was analyzed. The software identified 751,119 tweets using nanotechnology language, of which 495,195 qualified as valid nanotechnology-related tweets expressing opinions. The coders then sampled the final software-coded tweets to check validity and reliability.

In order to analyze the geographic origins of the 186,348 geotagged tweets, we identified federally funded centers and networks focusing on nanotechnology-related research and development using a list of centers and networks published on the National Nanotechnology Initiative website (National Nanotechnology Initiative 2012), and collected population and educational attainment rates by state (United States Census Bureau 2012). Using volume of tweets as our dependent variable, we ran an ordinary least squares regression model to determine if state educational attainment rates or the presence of a federally funded nanotechnology research center or network better predicted the volume of tweets per million residents. Independent variables were entered in blocks, with educational attainment as block 1, population in millions as block 2, and National Nanotechnology Initiative centers and networks as block 3 (Table 4).

Results

In answer to our first research question pertaining to the valence of the tweets, 32 % of all tweets expressing opinion were neutral, 27 % were optimistic, and 41 % were pessimistic (Table 2). When considering the sum of the optimistic and neutral tweets, these results are consistent with the literature on nanotechnology attitude. However, that the single largest portion was pessimistic (41 %) indicates that negative opinions toward the emerging technology on Twitter are occurring at higher rates than optimistic opinions. The fact that pessimistic opinions outweigh optimistic ones also makes Twitter tweets distinctively different from traditional news coverage in which positive mentions of nanotechnology continue to dominate the discourse.
Table 2

Valence of nanotechnology-related tweets expressing opinion on Twitter between September 1, 2010 and August 31, 2011

 

% Optimism

% Neutral

% Pessimism

Total

% Certainty

12

16

27

55

% Uncertainty

15

16

14

45

Total

27

32

41

100

H1 was supported. Certainty was expressed in 55 % of all tweets. While it is interesting to note that certainty is proportionally greater than uncertainty, the significance of this finding is not yet clear. This could suggest that Twitter discourse about nanotechnology is not dominated by vague predictions about potential applications down the road, instead focuses on fairly concrete predictions or developments in the field, for example, “New blog post: UNC study sends nano-particles to attack ovarian cancer.” Or it may indicate that tweets expressing certainty are re-tweeted more often than tweets expressing uncertainty.

In order to investigate H2, we used the states as units of analysis, and volume of nanotechnology-related tweets expressing opinion per state as the dependent variable. H2 was also supported by our data. Among tweets that were geographically identifiable (n = 186,348), California, New York, Texas, Florida, and Indiana generated the largest number of nanotechnology-related tweets expressing opinion by volume (Table 3). Hawaii, Alabama, Indiana, Arizona, and California generated the most tweets per million residents. The top five states by volume and tweets per million residents were all host to a federally funded nanotechnology research center or network during the time period studied (National Science Foundation 2012).
Table 3

Geographic distribution of nanotechnology-related tweets expressing opinion collected between September 1, 2010 and August 31, 2011

State

No. of tweets/million people

No. of tweets

Vol. (%)

State

No. of tweets/million people

No. of tweets

Vol. (%)

Alabama

128.29

6,132

3.3

Montana

81.87

810

0.4

Alaska

80.4

571

0.3

Nebraska

44.52

813

0.4

Arizona

95.99

6,136

3.3

Nevada

75.21

2,031

1.1

Arkansas

24.97

728

0.4

New Hampshire

61.38

808

0.4

California

93.56

34,854

18.7

New Jersey

42.93

3,774

2.0

Colorado

69.57

3,499

1.9

New Mexico

54.68

1,126

0.6

Connecticut

51.03

1,824

1.0

New York

91.23

17,679

9.5

Delaware

72.48

653

0.4

North Carolina

60.13

5,734

3.1

Florida

47.58

8,945

4.8

North Dakota

49.66

334

0.2

Georgia

35.83

3,471

1.9

Ohio

42.63

4,918

2.6

Hawaii

130.71

1,778

1.0

Oklahoma

28.5

1,069

0.6

Idaho

51.42

806

0.4

Oregon

78.91

3,023

1.6

Illinois

47.21

6,057

3.3

Pennsylvania

38.58

4,901

2.6

Indiana

118.22

7,665

4.1

Rhode Island

40.19

423

0.2

Iowa

46.58

1,419

0.8

South Carolina

28.67

1,326

0.7

Kansas

27.27

778

0.4

South Dakota

28.37

231

0.1

Kentucky

28.35

1,230

0.7

Tennessee

34.64

2,198

1.2

Louisiana

45.24

2,051

1.1

Texas

51.2

12,874

6.9

Maine

84.09

1,117

0.6

Utah

50.29

1,390

0.7

Maryland

63.72

3,679

2.0

Vermont

84.54

529

0.3

Massachusetts

86.11

5,638

3.0

Virginia

57.84

4,628

2.5

Michigan

40.31

3,984

2.1

Washington

61.0

4,102

2.2

Minnesota

51.34

2,723

1.5

West Virginia

29.47

546

0.3

Mississippi

19.31

573

0.3

Wisconsin

42.25

2,403

1.3

Missouri

36.15

2,165

1.2

Wyoming

35.84

202

0.1

In order to put H2 to a more stringent test, we used an ordinary least squares regression model to test if state educational attainment rates, state population or the presence of a federally funded nanotechnology research center or network better predicted the difference in the volume of tweets in a specific state. Our analysis showed that state population (β =0 .653, p ≤ 0.001) and the presence of a National Nanotechnology Initiative center or network (β =0 .359, p ≤0 .001) were positively correlated with the volume of tweets per state (Table 4), but educational attainment was not a significant predictor.
Table 4

Predicting volume of nanotechnology-related tweets expressing opinion by state

 

Zero-order

Model 1

Model 2

Model 3

Block 1: educational attainment

 % residents with high school degree or more

−.402

−.670***

−.040

−.040

 % residents with baccalaureate degree or more

.141

.850*

.134

.145

 % residents with post-graduate degree or more

.174

−.455

−.055

−.178

 Incremental R2 (%)

 

31.0

  

Block 2: State population

 Population in millions

.923

 

.901***

.653***

 Incremental R2 (%)

  

54.8

 

Block 3: federal funding for nanotechnology research

 Presence of National Nanotechnology Initiative center or network

.824

  

.359***

 Incremental R2

   

5.7

Total R2 (%)

   

90.5

* p ≤ .05; ** p ≤ .01; *** p ≤ .001s

Discussion

In this article, we offer the first systematic and exhaustive analysis of all nanotechnology-related Twitter tweets expressing opinion over a specific time period. Unlike research based on content sampling, which has been notoriously challenging for online media, we are able to make conclusions based on patterns across a census of messages from an entire population of users on Twitter. This information illuminates a number of phenomena that, until now, were closed off to empirical research.

The prevalence of tweets on Twitter indicates that Internet users who search for science content, specifically nanotechnology, are likely applying traditional media behavior to online media channel selection. The number of opinions detected, 495,195, over a twelve-month period is a small number compared to the 340 million tweets posted daily. Although there are distinct peaks and valleys, the two highest volume days, November 1 (4,217 nanotechnology-related tweets expressing opinion) and February 14 (6,235 nanotechnology-related tweets expressing opinion) still have relatively low overall volumes. A cursory review of tweets by day indicates that, like mainstream Twitter tweets, topics trend within nanotechnology and certain tweets are adopted and re-tweeted by users other than their original authors.

The ability to capture an entire census of Twitter users reveals a correlation between federal funding (in the form of a National Nanotechnology Initiative center or network), research, and communication that would normally be undetectable. In addition, the correlation between nanotechnology Twitter volume by state and federal funding is consistent with earlier work showing peer-reviewed publication rates, patents, and collaboration for nanotechnology cluster into nanodistricts centered around 7 U.S. regions (Shapira and Youtie 2008; Youtie and Shapira 2008), each the recipient of significant investment in nanotechnology research and development by the federal government.

Further, if we consider these results in the context of earlier work on the geographic clustering of R&D intensive industries (Audretsch and Feldman 1996), the likelihood of patent citations to reference other geographically proximate patents as evidence of knowledge spillovers at the regional level (Jaffe et al. 1993), and the synergistic effects of local knowledge, technical, and scientific information exchanges as bases for the rise of regional high-tech industries (Carlino et al. 2009; Heinze 2006; Krugman 2009), then these observations may be a glimpse into a community of like-minded nanotechnology innovators who are using Twitter for the purpose of communicating with one another. Using a census, rather than a sample, we have detected an ongoing conversation among a small, but distinct group of nanotechnology partisans, a sort of Nanotechnology Twitterati, with a particularly strong motivation to communicate with others also interested in nanotechnology. While the geographic distribution supports this idea, it is undetermined if Twitter is used primarily to reinforce local knowledge exchanges, or if it is used to overcome distance and lack of acquaintance to communicate with like-minded folks outside of Twitter user’s nanodistricts.

A second important finding is the geographic clustering of tweets around nanotechnology innovation centers, for which we use federally funding in the form of National Nanotechnology Initiative centers or networks as a proxy. This pattern, along with the results of the regression, supports the notion that these areas have a higher frequency of people with particular interest in, or knowledge of, nanotechnology. The presence of a federally funded nanotechnology research center or network as a predictor of a state’s nanotechnology-related Twitter traffic is important with regards to the societal implications for investment in nanotechnology infrastructure development. The U.S. government is one of the world’s leading investors in nanotechnology research and development. Through the National Nanotechnology Initiative (NNI), the Federal government disbursed $18 billion in funding between 2001 and 2012; at the time of this writing, an additional $1.8 billion was budgeted for disbursement in 2013 (National Nanotechnology Initiative 2012). Logically, it would follow that geographic areas receiving these funds would have higher proportions of all types of media users that are knowledgeable about nanotechnology, attend to nanotechnology news, and are motivated to communicate about the subject.

Implications for future research

The geographic clustering of nanotechnology-related tweets expressing opinion in combination with literature identifying science Internet news users as relatively more educated, more age-diverse and more likely to use interactive forms of media is interesting given the opinions observed in this study. Although few solid conclusions can be made about the identity of those tweeting nanotechnology-related tweets expressing opinions, these observations provide the grounds for future research examining this population as distinct from the larger population of Twitter users. Future work should focus on verifying evidence indicating that nanotechnology-related tweets expressing opinion originate from Twitter users who are close to the technology and perhaps among the leaders in the field, the Nanotechnology Twitterati. Sampling tweets from this study and surveying their authors would verify this observation and allow us to further explore motivations for participating in nanotechnology-related social media.

Our finding that the presence of a federally funded nanotechnology research center or network is positively correlated with nanotechnology tweets by state (Table 4) should be explored in-depth. Research should consider how unsolicited, spontaneous, and widely broadcast nanotechnology messages (tweets) impact lay public understanding of the technology and whether this form of exposure increases general knowledge, aids in the formation of opinion, or provides a pathway for the general public to access authoritative figures or institutions. Within the nanotechnology community itself, future research should examine both the reasons that National Nanotechnology Initiative centers and networks seem to be associated with increased nanotechnology social media traffic, and if these regions are becoming, or have the potential to become, social media nanotechnology information hubs linking geographically dispersed researchers to one another as well as to interested members of the lay public. The discovery of a spontaneous, self-organizing, geographically dispersed network of experts and laymen would have significant implications for how all science, not just nanotechnology, might be informally communicated.

A second logical step is to subject Tweets collected in this census to an analysis to determine if specific news items or topics vary by region and date, as well as the networked paths news items take when diffusing among nanotechnology Twitter users. This type of in-depth analysis would provide evidence supporting the notion that nanotechnology Twitter users are using the medium to communicate with those in local knowledge exchanges and/or build bridges between nanodistricts; it would also lead to a better understanding of how particular news items, memes, or ideas flow through Twitter via individual or networked connections between geographically proximate or disparate individuals and communities.

A third step to analyze our data will involve identifying common frames for collected tweets. While the Forsight software tallies word frequency, an analysis of the way in which tweets are framed would have to be performed by human coders. Based on the prevalence of scientific and medical words in frequencies, we would expect that nanotechnology-related tweets expressing opinion rely more often on technical frames than other types, but we hesitate to make this conclusion without a comprehensive audit of the census. Confirmation of this observation in conjunction with a greater understanding of how National Nanotechnology Initiative centers and networks are associated with increased nanotechnology social media traffic would give us insight into how the technical and lay communities are similar or different in their framing of nanotechnology on social media.

Finally, this study provides a glimpse into spontaneous, unsolicited communication about nanotechnology. The method used allows us to aggregate and analyze communication that was previously inaccessible to researchers. That this type of analysis can only occur in social media is significant and should be explored further. Specifically, a longitudinal study tracking opinion would be useful in understanding how social media on the Internet reflects and/or shapes public opinion of nanotechnology, and more practically could be used as an evaluation tool for the National Nanotechnology Initiative. In addition, combining a longitudinal study of opinion dynamics with ongoing research identifying those tweeting nanotechnology-related opinion would allow us to understand when and why different members of the nanotechnology public become involved in online conversation about the technology.

Acknowledgments

This material is based upon study supported by grants from the National Science Foundation to the Center for Nanotechnology in Society at Arizona State University (Grant No. SES-0937591) and the UW-Madison Nanoscale Science and Engineering Center in Templated Synthesis and Assembly at the Nanoscale (Grant No. SES-DMR-0832760). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.

Copyright information

© Springer Science+Business Media Dordrecht 2013