PUBLIC DELIBERATION GOES ON-LINE?
AN ANALYSIS OF CITIZENS' POLITICAL DISCUSSIONS ON THE INTERNET PRIOR TO THE FINNISH PARLIAMENTARY ELECTIONS IN 2007
KIM STRANDBERG
o
Abstract
This article examines on-line political discussions' potential of becoming truly deliberative discussions, capable of bringing about democratic benefits, through combining two theoretically important aspects found in the literature
-	concerns regarding citizens' participation in on-line politics and the quality (or lack thereof) of on-line discussions
-	in one analytical framework. Specifically, the article firstly examines how many, and more importantly, which types of
citizens participate in on-line discussions. This part of the analysis adds to the scholarly debate concerning whether on-line politics is reaching beyond politically active and interested segments of the public. Secondly, the article examines the discussions on four Finnish political discussion boards during the last three weeks before the Finnish parliamentary election in March 2007. The quality of the discussions is assessed and discussed in light of several criteria based on the literature concerning deliberative democracy. In combining these two aspects, the article fills a gap in the research field where these aspects have mostly been examined separately. The findings of the article generally demonstrate that on-line discussions are not, at least for the time being, truly deliberative. The debates analysed generally did not meet deliberative standards in terms of quality and only politically very active and interested citizens seemed to take part in them. The question thus still remains if, and how, on-line citizens' discussions can ever
become truly deliberative.
Q.
a.
Kim Strandberg is a researcher in the Department of social sciences, Äbo Akademi University, Vasa; e-mail: kistrand@abo.fi.
00 o o rN
m
£
£ 3
a
0
£ *
1 *
(R 0 C
K
Introduction
Scholars have argued that indications of declining civic political engagement and participation are evident in modern representative democracies (Dalton 2000; Lusoli et al 2002; Norris 2001a; Wattenberg 2000). It is proposed that citizens are becoming increasingly detached from traditional institutions of representative democracy and feel unenthusiastic about traditional political mobilisation agencies (Dalton 2000; Kippen & Jenkins 2004, 254; Norris 2001a, 217). The membership of political parties along with electoral turnout has experienced downward trends (Djupsund & Carlson 2003; Mair & van Biezen 2001; Scarrow 2000). Modern representative democracy has even been labeled as a "thin" form of democracy, failing to involve the citizens in decision making (Barber 1984).
These trends have seen several scholars calling for democratic processes which are better capable of involving citizens (e.g. Barber 1984; Dahl 1989). Essentially, the ideals of participatory and deliberative democracy have been called upon as alternatives to the top-down vertical communication symptomatic of representative democracies. Notions such as 'strong democracy' (Barber 1984) and 'the public sphere' (e.g. Habermas 1962), which both strongly emphasise horizontal citizen deliberation and communication, have been put forth. Drawing on these democratic theories, scholars have especially perceived the internet as providing excellent tools for realising horizontal communications and consequently possibly reviving citizen deliberation and participation in public matters (e.g. Budge 1996; Dahl 1989; Rheingold 1993). Several web-specific features hold a potential to increase citizens' knowledge, political engagement and participation (Kavanaugh et al 2005, 21-22; Polat 2005, 436-437; Papacharissi 2002, 9-10). The vast amount of information available on-line, and the accessible opportunities for horizontal communication between citizens facilitated by on-line discussion boards are often mentioned in the literature. For instance, Coleman and Goetze (2001, 17) write that the internet "makes manageable large-scale, many-to-many discussion and deliberation". Some y scholars even contemplate whether citizen discussions on the internet constitute a new 'virtual public sphere' (e.g. Fuchs 2006; Woo-Young 2005).
While on-line citizen discussions contain a clear potential for facilitating citizen deliberation and engagement in public and political matters, there is still much uncertainty whether the potential can actually be realised. In this article attention is drawn - both theoretically and empirically - to two aspects which are important to consider in relation to studies of on-line citizen discussions. Firstly, there are significant concerns regarding on-line political activity, especially participation in on-line discussions, not reaching beyond the already active segments of citizens. ^ Secondly, scholarly discussions and concerns vis-à-vis qualitative aspects of on-line citizen discussions are highlighted. Both of these aspects are assessed through empirical examinations of on-line citizen discussions in Finland, a country with a long history of being among the countries with the highest internet penetration rates in the world1 (Norris 2000). Two general questions are put in theoretical light in the two subsequent sections and empirically addressed thereafter; are on-line citizen
0
C >
discussions reaching beyond the active segments of the population and do they live up to the qualitative ideals of democratic deliberation? The findings mostly indicate that the answer to both of these questions is negative. On-line debaters are found
il
r\i
to belong to very active segments of the population at large and the discussions in which they engage fail to live up to rigid deliberative standards.	j^O
On-line Political Activity: Mobilising or Reinforcing?
Even though scholars have argued that the internet holds the potential to "inform, organise and engage those who are currently marginalised from the existing political system" (Norris 2001b, 218), it could nonetheless fail on all of these accounts. Two theoretical viewpoints concerning the internet's potential to engage citizens have been excellently described by Norris (2001b, 218-219); the mobilisation theory and the reinforcement theory.
The mobilisation theory recapitulates numerous optimistic visions regarding the internet's ability to affect citizens' political activity (e.g. Barber 1984; Budge 1996; Dahl 1989; Dertouzos 1997; Negroponte 1995; Rheingold 1993; Schwartz 1996). The theory contains four main arguments in defence of its assessment of the internet's impact on citizen participation (Norris 2001b, 218): Firstly, the internet provides abundant opportunities for political engagement. Secondly, the relative ease and low costs of receiving information via the internet could lower the barriers in order for citizens to learn about public matters (cf. Downs 1957). Thirdly, the large amount of information available on-line provides citizens with the opportunities to become more informed about public affairs, and thus more prone to become active concerning public matters. Finally, seeing that the internet enables two-way communication, it could strengthen and improve the links between citizens and mediating organisations (cf. Coleman & Goetze 2001; Kamarck 1999). To summarise, the internet is considered to form a distinct type of opportunity for political participation which considerably diverges from traditional participation channels; it possesses the ability to inform, activate and engage citizens.
A number of research findings have been found in support of the mobilisation theory. The political on-line audience has grown considerably over time, largely due to the increased penetration of the medium, but also due to a change of user preferences in seeking out political information (see discussion in Lusoli 2005, 154-156). The internet is increasingly becoming an important source of political information for young people, a group of citizens normally less politically active off-line (e.g. Boogers & Voerman 2003, 25; Carlson & Strandberg 2005; Gibson et al 2005, 578; Norris 2003, 39-40). Likewise, Gibson and colleagues (2005, 578) argue that their results indicate that the internet is "offering a space for political engagement among those who might not have been otherwise active." Johnson and Kaye state that "though the Web has not yet changed the larger democratic process [...] The Web politically empowers individuals and increases their feelings of self-efficacy, levels of political involvement, political interest, campaign interest and likelihood of voting" (Johnson & Kaye 2003, 28-29; cf. Kaye & Johnson 2002, 65-67).
However, according to the opposing reinforcement theory, the internet would rather merely constitute an alternative channel for the politically motivated, active and engaged citizens to engage in familiar activities (Hill & Hughes 1998, 44; Norris 1999, 89). There is little reason to expect the internet to cause previously uninterested citizens to suddenly become 'political animals' (Davis 1999, 8; cf. Pa-pacharissi 2002, 22). Politics on the net will fail to politically activate and engage citizens even if current 'digital divides' in internet access are overcome. Thus,
personal motivation and preferences are central. Along these lines, Djupsund and Carlson (2003, 41) observe that the internet differs from traditional mass media on one crucial point: it requires activity rather than passivity from the user. While television exposes the passive viewer to political content, the on-line citizen must actively seek out the information he or she desires from an overabundance of choices (cf. Hill & Hughes 1998, 183). This leads to the core argument of the motivation-based reinforcement theory: "There are a million places to go and sites to see on the Internet. Unless they stumble across political content accidentally ... those who choose to visit political sites will probably have far higher than average civic interest" (Norris 2001b, 221).
Several observations which underline the importance of motivation and personal preferences in determining citizens' use of the internet have been put forth by social science scholars. Largely comparable socio-economic patterns to those found regarding traditional political activity and interest have also been noted upon examination of internet users engaging in on-line politics (Lusoli 2005, 262; Norris 1999, 87). Norris (2001b, 231) found that "party websites tended to attract those who were already among the most aware of public affairs, as well as those with higher socio-economic status." Hill and Hughes (1998, 183) emphasise that political use of the internet is an act of self-selection: "people go on-line to find out more information about a subject, not to be transformed". Stromer-Galley and colleagues (Stromer-Galley et al 2001, 24) note that only a fraction of citizens seek out political information on the internet, and that an even smaller fraction is involved in more engaging forms of on-line political activity, such as taking part in political discussions (cf. Cornfield et al 2003, 20; Norris 1999, 81-82; 2001b, 223; 2003, 36). Muhlberger (2004, 235) similarly explicitly states that "motivation matters" pertaining to citizens' propensity to engage in on-line political discussions.
The Deliberative Quality of On-line Citizen Discussions
According to Dahl (1989, 169-179), a normative ideal for a truly democratic pro-y cess contains certain criteria: effective participation, equal voting opportunities, and enlightened understanding. As Dahl himself notes (pp. 169-170), such a process is yet to appear and highly unlikely to ever do so. Likewise, Barber (1984, 24) claims that contemporary democracy, or what he calls "thin democracy," brings about "neither the pleasures of participation nor the fellowship of civic association, neither the autonomy and self-governance of continuous political activity nor the enlarging mutuality of shared public goods - of mutual deliberation, decision, and work." Several democratic theorists have instead called for more participatory, deliberative forms of democracy (e.g. Barber 1984; Dahl 1989; Habermas 1962). As mentioned in the introduction, scholars have especially looked to the internet as being potentially capable of achieving such participatory, deliberative democratic processes.
In essence, the general notion of deliberative democracy perceives political discussions to have benefits for democracy only if these meet certain qualitative criteria (e.g. Barber 1989; Fishkin 1995; Sunstein 2001, in Witschge 2004, 110). Drawing upon deliberative norms, Jankowski and Van Selm (2000, 153-154) pinpoint several important interrelated issues of concern to on-line citizen discussions: equality of the discussants, diversity of topics, reciprocity and tone of discussions. Dahlberg (2001, 2-3) similarly lists exchange and critique of criticisable moral-practical va-
C >
-šT rv
lidity claims, reflexivity, ideal role-taking, sincerity, and discursive inclusion and equality as central components of a deliberative, democratic discussion. Barber LO (1984, 178) states that deliberative discussions should include a wide array of par- ^ ticipants, cover a wide range of topics and the actual discussions should involve mutual understanding, respect, and "listening as well as speaking, feeling as well as thinking, and acting as well as reflecting."
However, correspondingly to the uncertainties concerning on-line politics not reaching beyond the active citizens, scholars have been rather inconclusive regarding whether internet discussions actually meet these standards. Bentivegna (1998, 8-9), on one hand, reports that on-line discussions show equality among members, and states that they are near equivalents of a public sphere. Other scholars, however, have found them to be dominated by a small number of very active citizens (cf. Fuchs 2006, 14; Jankowski & Van Selm 2000, 160, 190). Hence, there are few signs of either equality between, nor heterogeneity of participating citizens. Dahlberg (2001, 13-16) raises concerns regarding what he calls a risk of on-line interaction becoming merely an "elite public sphere." Also, both Wilhelm (1999, 172-183) and Hill and Hughes (1998, 74) similarly found tendencies towards "tribalisation" in as much as on-line discussants tended to engage in conversation with likeminded citizens resulting in situations where the discussions "emphasize particular topics from particular points of view" (Hill & Hughes 1998, 74; cf. Papacharissi 2002, 17; Polat 2005, 451). Therefore, diversity of discussion topics is usually found on different discussion boards but not to any wider extent within one discussion board. Bimber (2000, 332) talks about not one single, but many specific public spheres on the internet. Wilhelm (1999, 170-172) reports that some discussion boards do cover a rather wide range of topics, but that many nonetheless have well-defined agendas and strong in-group affiliation.
Turning to reciprocity, rationality, and tone of discussion - i.e. what Barber (1984, 223) mentions as some of the pillars of strong democracy - findings again point in fairly inconsistent directions. Bentivegna (1998, 9) argues that the horizontal nature of citizens' communication on on-line discussion boards correspond to communities "bound together by horizontal relationships of reciprocity and cooperation" Fuchs (2006, 24) concludes that the Austrian on-line discussions which he analysed contained a large degree of interactive discussion. He nonetheless expresses concerns regarding the significant presence of insulting and outright destructive postings. Papacharissi (2002, 16) also raises concerns regarding the abundance of "hasty opinions" and "flaming and conflict beyond reasonable boundaries" found in many analyses of on-line discussions (cf. Hill & Hughes 1998). Jankowski and van Os (2004, 190) write that asynchronous discussion boards appear to provide good opportunity for reading and reacting to the writings of other participants, i.e. preconditions for reciprocity and rational-critical debate. However, they also remark that this opportunity was not seized by the participants as there were little signs of achieving mutual understanding and agreement in the Dutch discussions which they analysed. Dahlberg (2001, 8) similarly notes that "few participants acknowledge the strength of criticism directed towards them and even fewer seem moved to change or compromise their positions in the course of argumentation." Wilhelm (1999, 170-171) similarly expresses concerns over on-line debaters not feeling responsible before each other and "not responding to the views of other group members" (cf. Strandberg 2005).
u
S
5
0
£ *
1 *
IR O
C
«
UD
Analytical Framework
The discussion thus far has pointed at two central theoretical perspectives often used in research concerning on-line political activity by citizens in general, and on-line citizens' discussions in particular. Evidently, there has been a considerable amount of research of citizens' on-line discussions. However, there have been few, if any, studies considering the two aspects in conjunction. This is somewhat surprising given that the requirements of truly deliberative discussions, mentioned by all democratic theorists, is that these include a wide range of participants and have a good deliberative discussion quality (e.g. Barber 1984; Dahl 1989; Fishkin 1995). Arguably, a dual approach in studying citizens' on-line discussions could give a more comprehensive picture, and a better understanding of the on-line citizen discussions than what has been present earlier. While it is true that the deliberative quality is an important aspect of on-line discussions, there is an apparent risk of exaggerating the impact of the debates without knowledge of their spread among the wider public. Concerning this, Muhlberger (2004, 236) argues that: "Online political discussion is too small a portion of overall political discussion to have appreciable political effects currently." Nixon and Johansson (1999, 146) also express concerns regarding the fact that political internet users constitute merely a small share of the entire population and are not representative of all citizens. If on-line discussions only become a venue for an 'elite' group of citizens already politically engaged, they will do little to affect democracy, regardless of their potential deliberative virtues. In such a case, the discussions can hardly be regarded as truly deliberative; rather they merely have the potential to become so if a larger share of the citizens were to take part in them. Both dimensions thus shed light on each other which renders a more comprehensive understanding of the phenomena. The theories of mobilisation and reinforcement serve to position the assessments of on-line discussions within a wider societal context.
I therefore argue that both the extent to which citizens in general engage in on-line discussions, and the aforementioned qualitative aspects, are important elements in an assessment of on-line political discussions. Schematically, an analytical framework is depicted in Figure 1.
Figure 1: Typology of Political On-line Citizens' Discussions
Deliberative quality of discussions High	Low
2 eg
Wide
Narrow
A	B
C	D
Although, as most typologies are, the figure is a simplification of reality, the framework nonetheless provides a tool for analysing the current state of on-line discussions thoroughly. In effect there are four different types of discussions depicted in the typology. Firstly, in situation A, discussions are so-to-speak truly deliberative, i.e. a large and heterogeneous group of citizens participate in discussions which
also meet the qualitative standards of deliberative discussions. The type of discussion depicted in situation B is somewhat ambiguous to describe. While there is a satisfying share of citizens taking part in the on-line discussions, the quality of the debates fails to live up to deliberative standards. On a positive note, people appear to be drawn to on-line political discussions, but, on a negative note, they do not appear to engage in discussions which meet deliberative ideals. Situation C in the typology describes on-line discussions which hold a clear potential to become truly deliberative but still fail to engage larger shares and more heterogeneous groups of citizens. Finally, in situation D, the discussions do not hold any qualitative potential and are not reaching out to larger shares of the public.
Drawing on this framework and the theoretical discussions in the previous sections, three empirical research questions are formed for the purpose of assessing on-line discussions in light of the analytical framework:
RQ1: To what extent are citizens engaging in on-line political discussion?
RQ2: What are the characteristics of the citizens that participate?
RQ3: Do on-line political discussions meet the deliberative ideals concerning the quality of discussions such as equality of discussants, diversity of topics, reciprocity, and rationality of discussions?
Research Methods and Data
The empirical focus of the study is on on-line citizen discussions found in Finland, a country which has a relatively long history of a high level of internet penetration. In fact, Finland was among the world leaders already in 1999 (Norris 2000), and still ranks fifth in the EU with its penetration of 62.3 per cent2. This high internet penetration is important as it reduces the risk of differences in internet access influencing the findings regarding the spread of on-line political discussions among the citizens. As a result, the focus of the analysis is turned towards motivation-based aspects, which are stated as important to consider by the reinforcement theory. This focus is also arguably essential considering the short history of the medium's diffusion among the general public. As the medium spreads, as it undoubtedly will (cf. Norris 2001b, 92), access to the technology will be less influential on citizens' on-line behaviour, thus accentuating the role of individual preferences, interests and motivation.
In order to analyse the extent to which the Finnish people engage in on-line political discussions and the characteristics of those citizens actually talking politics on the internet, survey data from the Finnish national election survey 2007 is used3. The findings from the national survey are complemented by a quantitative content analysis of discussion threads and messages posted during the last few weeks, i.e. March 5th to 18th, leading up to the 2007 parliamentary election (see appendix for detailed coding scheme). It can, of course, be argued that in opting for a coding period prior to an upcoming election, there is a risk of hasty partisan opinions dominating discussions and adversely affecting the deliberative quality (cf. Papacharissi 2002, 16). However, on the other hand, elections do raise the overall interest in politics (e.g. Asp 1986, 108) which ought to serve to increase the likelihood of people engaging in political discussions. A coding period in between elections could very well serve to underestimate the quantitative activity on political internet discussions just as much as a coding period in times of elections possibly
underestimates the quality of discussions. It should also be noted that content analyses, naturally, are only capable of providing rather superficial indications of the deliberative quality of on-line discussions. There are most likely many underlying dimensions which go unnoticed in any form of content analysis. Then again, on the other hand, any form of deeper more participatory form of analysis restrains the capacity to encapsulate a large enough sample in order to generalise from the findings.
The content analysis focuses on discussions located both on the Usenet and on the World Wide Web. When turning to the choice of actual discussion forums, forums which appear to have a clear potential for relatively large scale, diverse discussions taking place were chosen. In search of such on-line discussion forums, where so-to-speak large enough, diverse and varying topics may occur, discussion boards situated on three popular websites in Finland in terms of user traffic4 were opted for; www.iltalehti.fi,www.suomi24.fi and www.jippii.fi. Additionally, discussions on one of the few existing Finnish political Usenet boards; sfnet.keskustelu. politiikka, were analysed. All of these sites host forums for several different topics, politics being just one of them. The political discussion on the website of one of Finland's two leading tabloids, www.iltalehti.fi is not focused on fixed topics. Instead, participants are free to start discussions around any political topic. The www.jippii.fi website - a web portal - hosts one "general politics" discussion, also without fixed topics. The political discussions on the most popular Finnish web portal, www.suomi24.fi are categorised into several sub-themes within which there are no fixed topics. Several of these forums were, however, devoted to discussions concerning individual political parties, which per se appears likely to mainly attract citizens with similar views. Therefore, I opted to only analyse the suomi24. fi discussions on the section labelled "politics in general." The sfnet.keskustelu. politiikka discussion group has the character of a "general politics" discussion without fixed topics.
Regarding the content analysis of the discussions, a random sample (20 per cent) of discussion threads was coded5. More specifically, the threads, consisting of several individual messages, were analysed using a coding scheme (see appendix for detailed description of the scheme) drawing upon existing research within the field (e.g. Bentivegna 1998; Hill & Hughes 1998; Wilhelm 1999; Strandberg 2005). In particular, the number of participants and messages, as well as the character of the start message of each thread were analysed. Moreover, the occurrence of conflicts Q or disagreements within threads, and whether mutual understanding nonetheless could be achieved, was noted. Secondly, the main topic, type of content, use of validations and tone were analysed for each individual message posted to the forums during the coding period (see appendix for details).
gj On-line Political Discussions: Amount of Activity and Characteristics of Participants
The first part of the analysis focuses on the extent to which Finnish citizens engage in on-line political discussion. Firstly, Table 1 contrasts the share of citizens ß engaging in on-line political discussions to the share of citizens obtaining election-related information from various media channels.
00 rv
U
S
Table 1: The Share of Respondents Following the Election through Different Media
	%*	N
TV news/current affairs programmes	65.5	931
Newspapers	48.8	694
Radio	16.8	239
Internet news	12.3	175
Party- or candidate websites	6.8	96
Blogs	2.7	38
Candidate selectors	13.2	187
On-line political discussion**	14.1	200
Total:	100	1422
* The share of respondents claiming they engaged in the activities either "very much" or "quite much."
** The share of respondents claiming to have engaged in at least one on-line political discussion during the last 4 years.
Evidently, internet sources are lagging far behind traditional media sources, disregarding radio, in terms of the share of citizens using them for obtaining election-related information. Notably, a substantial share of Finnish citizens claims to have engaged in at least one on-line political discussion during the last four years. However, due to the phrasing of the questionnaire item, this finding is not to be regarded as representing the share of citizens regularly engaged in political discussion on the internet. That share is most likely lower (cf. Cornfield et al 2003, 20). Turning to the characteristics of the on-line debaters, Table 2 firstly presents a comparison of Finnish internet users in general and on-line debaters. Specifically, demographical, as well as political and media activity items are compared.
Table 2 shows several interesting statistically significant differences between the two groups. Most importantly, the differences found bear a strong resemblance to the main argument of the reinforcement theory stating that the internet is mostly a new channel for the politically motivated, active and engaged citizens to engage in familiar activities (Hill & Hughes 1998, 44; Norris 1999, 89). In support of this, the findings show that the on-line debaters voted to a significantly higher degree than general internet users. Moreover, they used all forms of media for following the election significantly more than general internet users. Their level of political interest was also significantly much higher than general internet users.
Still, Table 2 does not reveal any information on the independent and relative importance of these citizen-specific variables in predicting the engagement in on-line political discussion. This issue is further explored in Table 3 in which a logistic regression model predicting citizens' participation in on-line debates is presented.
Table 2: The Characteristics of Finnish Internet Users and On-line Debaters
	Internet users in general		On-line debaters	
	%	N	%	N
Education				
Primary	10.0	75	11.0	22
Vocational	37.7	282	37.5	75
Upper secondary	23.7	177	24.5	49
Polytechnic	9.8	73	7.0	14
University	18.9	141	20.0	40
Household income				
I (poorest)	22.6	154	28.6	50
II	31.9	217	34.9	61
III	33.1	225	26.9	47
IV (richest)	12.4	84	9.7	17
Age**				
18-24	15.1	114	10.5	21
25-34	21.7	164	19.5	39
35-44	17.4	132	15.5	31
45-54	19.2	145	17.5	35
55-64	17.3	131	16.5	33
64+	9.4	71	20.5	41
Gender				
Male	49.8	480	51.5	103
Female	50.2	477	48.5	97
Voted in election**				
Yes	84.0	634	91.0	182
No	16.0	121	9.0	18
Use of media				
TV news/current affairs***	59.5	450	76.5	153
Radio***	10.8	82	26.5	53
Newspapers***	44.4	336	68.0	136
Internet news***	13.3	101	34.5	69
Candidate/party websites***	7.0	53	19.0	38
Blogs***	2.3	17	9.0	18
Candidate selectors*	17.3	131	26.5	53
Political interest***				
Very low	6.3	48	1.5	3
Low	35.0	265	16.0	32
High	45.4	344	43.5	87
Very high	13.2	100	39.0	78
il
O
00
* p<.05; ** p<.01; ***p<.001 Chi-square tests and Fisher's exact tests of differences between the two groups. Notes: Education=the highest education obtained; Household income=quartiles; Use of media=the
share of respondents claiming to have used the media in question "quite much" or "very much" for
following the election.
	B	SE	Exp(B)
Education	-.307	.415	.736
Income	-.870	.445	.419
Age/100	-.705	.714	.494
Gender	.104	.196	1.110
Voted in the election	-.132	.312	.877
Activity in off-line politics	5.211***	.623	183.340
Use of traditional media	2.368**	.858	10.677
Internet use	.839*	.428	2.443
Political interest	1.533*	.632	4.631
Constant	-5.281***		
N	1,041		
% correct	86.0		
Cox-Snell R2	.152		
Nagelkerke R2	.262		
* p<.05; ** p<.01; ***p<.001
Note: The dependent variable: 0=citizen did not take part in on-line political debate, 1=citizen took part in at least one discussion during the last 4 years. Predictors: Gender: Male (1); Female (0). All other predictors are recoded dummies on a scale from 0 to 1 based on the categories presented in Table 2. Activity in off-line politics is also a dummy (0-1) based on a 12 point scale assigning 1 point for each of 12 different political activity-items.
Activity on the Political Discussion Boards: Quantity and Deliberative Quality
The second theoretically important aspect of on-line citizen discussions mentioned in the theoretical framework is whether these meet deliberative ideals. This aspect is now addressed using the results of the content analysis of four Finnish political on-line discussion boards.
Table 4 displays several quantitative indicators of the activity on the four discussion boards. In particular, drawing upon the research within the field (e.g.
With due caution to the somewhat low explanatory power of the model, the findings reported in Table 3 confirm the pattern observed in Table 2. The regression model thus reveals several significant predictors which strongly support the reinforcement theory: compared to the general public, Finnish on-line political debaters are significantly much more likely to be active in off-line politics, to use off-line media for political purposes and to have a high level of political interest. Moreover, quite naturally, they also use the internet to a higher extent. Indisputably, the Finnish citizens taking part in on-line political discussions are politically very active and have a high degree of civic interest (cf. Norris 2001b, 221). Bearing democratic theories in mind, on-line discussions are thus apparently not including a wide range of participants (cf. Barber 1984; Dahl 1989; Fishkin 1995). On the contrary, on-line political debaters belong to a very specific segment of the general public.
Table 3: Determinants of Participating in On-line Political Discussions in Finland (logistic regression)
Bentivegna 1998; Hill & Hughes 1998; Strandberg 2005), three measures are in focus: the degree to which the debates where monopolised by few debaters (i.e. messages posted by each debater), the liveliness of the debates (i.e. a standardised measure of the relationship between original and reply posts) and intensity (a measure indicating the length and complexity of discussion threads).
Table 4: Four On-line Discussion Boards: Quantitative Indicators of Activity in Discussion Threads
	jippii.fi (N=16)		suomi24.fi (N=39)		iltalehti.fi (N=164)		sfnet (N=21)		Total (N=240)	
	Mean	Std. dev.	Mean	Std. dev.	Mean	Std. dev.	Mean	Std. dev.	Mean	Total
Messages*	16.69	29.22	2.36	1.29	12.65	22.94	16.33	22.28	11.57	21.74
Debaters*	4.19	4.17	2.18	1.05	6.98	11.11	6.43	6.11	5.96	9.59
Monopolisation***	2.47	2.00	1.07	.20	1.52	.75	1.91	1.04	1.54	.92
Liveliness (0-1) **	.56	.45	.45	.29	.68	.33	.66	.39	.63	.35
Intensity***	59.5	119.31	2.44	1.41	45.27	116.49	206.62	397.52	53.38	160.95
* p<.05; ** p<.01; ***p<.001 One-way ANOVA tested for difference between group means.
Judging from the figures concerning the number of individual messages, all forums except the suomi24-board display a rather satisfying level of activity given the Finnish context (cf. Strandberg 2005). It is, however, noteworthy that the standard deviation is quite high which means that there is much variation in thread length. This is also confirmed by the degree of liveliness and intensity which also both have high standard deviations. Consequently, it seems as if there are two types of discussions; short, lifeless and low-intensity or lengthy, intensive and lively debates. In theoretical terms, there are discussions whose quantitative activity meets deliberative ideals but there are also an equally large share of discussions which do not. y Still, many discussions are long enough to hold a potential for deliberation to take place. Beyond these features, Table 4 shows that the discussion threads comprise an average of over ten individual messages and roughly six individual debaters and are thus not dominated by a few citizens as often found in other studies (cf. Fuchs 2006; Jankowski & Van Selm 2000). Viewed in light of deliberative norms, the equality of discussants appears to be quite good (cf. Dahlberg 2001, 2-3; Jankowski & Van Selm 2000, 153-154).
Moving onward with the analysis, focus is turned to the contents of the discussions. Concerning this, many of the democratic theorists' (e.g. Barber 1984; Dahl 1989; ^ Fishkin 1995) perceptions of deliberative discussions are again central to the analysis. Table 5 shows the four forums according to certain qualitative indicators.
0
C >
Several scholars have, as discussed earlier, raised concerns regarding the fact that topics in on-line discussion usually vary more between forums than within forums (cf. Hill & Hughes 1998; Papacharissi 2002; Polat 2005). The Finnish findings displayed in Table 5 do not entirely support this: the forums have a rather varied
distribution of topics inasmuch as no discussion is predominantly focused on just one type of topic. On the other hand, however, the topic with the largest share does QQ vary significantly between the different discussion boards which is more in line
Table 5: Four On-line Discussion Boards: Qualitative Indicators of Deliberative
Quality in Individual Messages PO _ 00
	jippii.fi		suomi24.fi		iltalehti.fi		sfnet		Total	
	(N= %	65) N	(N= %	88) N	(N= %	675) N	(N= %	80) N	(N= %	908) N
Topic										
Politican/ Party image	27.7	18	23.9	21	45.8	309	13.8	11	39.5	359
Issues	27.7	18	47.7	42	22.5	152	35.0	28	26.4	240
Political process	15.4	10	17.0	15	13.6	92	7.5	6	13.5	123
Metatheme	18.5	12	9.1	8	13.2	89	41.3	33	15.6	142
Type of content										
Provide fact	6.2	4	2.3	2	0.0	0	0.0	0	0.7	6
Opinion	18.5	12	25.0	22	24.3	164	38.8	31	25.2	229
Question	26.2	17	15.9	14	15.0	101	17.5	14	16.1	146
Statement	32.3	21	25.0	22	27.9	188	17.5	14	27.0	245
Announcement	16.9	11	31.8	28	31.1	210	25.0	20	29.6	269
Tone of message										
Negative	53.8	35	52.3	46	68.9	465	70.0	56	66.3	602
Neutral	33.8	22	22.7	20	17.0	115	21.3	17	19.2	174
Positive	12.3	8	19.3	17	12.1	82	7.5	6	12.4	113
Use of validations										
None		50		81		664		80		875
Internal		11	0.0	0	0.0	0	0.0	0	1.2	11
External	6.2	4	8.0	7	1.6	11	0.0	0	2.4	22
Note: All differences in the distribution of share between the forums are statistically significant at WHAT the .05 level (chi-square tests).
with the "topics vary more between than within forums" perspective. Nonetheless, the discussions, to a certain extent, meet the deliberative ideal of diversity of discussion topics.
Regarding the remaining three indicators, however, deliberative ideals are hardly met. Similar to many existing studies of on-line discussion, the debates analysed here display a low degree of reciprocity and mutual respect, the participants do not feel responsible before each other and they are not "listening as well as speaking, feeling as well as thinking nor acting as well as reflecting" (Barber 1984, 178). This is indicated by the large share of negative messages and the dominance of opinions, statements and announcements over questions and facts (cf. Fuchs 2006; Papacharissi 2002; Wilhelm 1999, 169-170). The relatively low share of questions is particularly worrisome; this could indicate that the discussions were not truly "interactive exchanges" (cf. Wilhem 1999, 170). Also, very few messages included any validations, neither internal (e.g. argumentation based on previous posts) nor external (e.g. links to facts or articles), supporting the posted message. The degree of "rational-critical argumentation" is thus very low (cf. Wilhelm 1999, 173-174). Correspondingly, a separate analysis of discussion threads (not reported
in detail here) shows that roughly one in four discussions turned into conflicts and that any kind of mutual agreement only could be achieved in 20 per cent of these conflicts.
Conclusions
Naturally, as in most studies of any on-line phenomena, the analysis in this study is bound to have missed out on thousands of on-line political discussions. Seeing as the forums analysed in this study are located on very popular Finnish websites, and the most popular Usenet root, I tentatively suggest that the analysis facilitates some general conclusions concerning the current state of mainstream citizens' on-line political discussions. Likewise, as discussed earlier, the content analysis per se is also bound to have merely scratched the surface of the deliberative quality of on-line political discussions. Returning to the analytical framework established earlier, bearing these reservations in mind, what conclusions can be drawn from the empirical analysis?
Four ideal types of on-line discussions were depicted in the analytical framework (Figure 1): a truly deliberative discussion, a potentially deliberative discussion, a non-deliberative wide audience discussion and a narrow audience non-deliberative discussion. The empirical analysis of the citizens taking part in on-line political discussions and the content analysis of actual discussions have brought about rather discouraging results. The on-line political discussions in Finland - a country which clearly is very accustomed to the internet and its use for political activity (cf. Gibson & Römmele 2005, 10) - are hard to classify as anything but narrow audience non-deliberative discussions, or at best as narrow audience semi-deliberative discussions. Even though a substantial share of Finnish citizens claimed to have engaged in on-line political discussions at least once during the last four years, the analysis revealed that these citizens were noticeably biased towards being politically very active and interested. Nixon's and Johansson's (1999, 146) fear of on-line discussions becoming a venue for an 'elite' group of politically engaged citizens y is thus very true, at least in light of this analysis. The notion contained within the reinforcement theory of motivation being a crucial driving force of citizens' political use of the internet (e.g. Hill & Hughes 1998, 44; Norris 2001b, 221) is also supported. Pertaining to the analytical typology (Figure 1), the Finnish discussions analysed in this study are clearly located on the "narrow" end of the vertical - penetration among the public - dimension.
Given that on-line debaters are politically active and motivated, it is surprising that the content analysis of the four Finnish discussion boards revealed mostly sparse signs of true deliberation, as described by democratic theorists, taking place. Firstly though, on a positive note, some discussions were rather long, intense and lively and therefore held a potential for fruitful debates taking place. Also, the debates were not dominated by few active discussants and the diversity of discus-
l-^M^i/if TiTTf f n^if(Tri«(T 1 Vr>	«dfrnl-nrfi fi/ii-»	4TAT	4-Krt fi i vt-K At" TnilTTfir
C >
sion topics was satisfying. On the negative side, however, the further analysis of discussion contents showed that the discussions showed little signs of being truly deliberative. The degree of reciprocity and level of mutual respect were both quite
low. Moreover, there was an abundance of opinions, announcements and state ments leaving little opportunity for discussions to take place. The messages were not backed up by rational argumentation, as measured by the use of validations
-šT 00
in this article, which truly deliberative discussions ought to feature (e.g. Dahlberg 2001). A large share of the messages was also negative in tone. This is not per se in LO contrast to deliberative ideals, in fact, critical argumentation is a central part of such discussions. However, taking the large share of unresolved conflicts and disagreements into account (i.e. 80% of the threads where conflicts appeared), one might question the degree of critical argumentation in the noted negativism. Turning to the analytical typology, the deliberative quality of the discussions analysed mostly corresponds to situation D, i.e. a low deliberative quality, whilst containing, to a lesser extent, some elements which are symptomatic of good-quality deliberative discussions.
In conclusion, the study has demonstrated that public deliberation is generally not found on-line, at least not in Finland. The discussions mostly resemble narrow audience non-deliberative discussions, as described in the analytical framework (Figure 1). According to the general deliberative theory, the discussions thus do little to enhance democracy (cf. Fishkin 1995; Sunstein 2001). That said, neither does deliberation exist off-line in Finland, where a systematic deliberative experiment has only recently been conducted (Setälä et al 2007). Finnish society in general also has a clear emphasis on quietness and solitude over conversation (e.g. Carling 2007). This could be one explanation why politically active and interested citizens fail to engage in truly deliberative on-line discussions. It simply doesn't come naturally to Finnish citizens. Concerning this, however, it is important to bear in mind that studies from other countries have also found similar results. This would indicate that the nature of on-line discussion boards themselves, perhaps most importantly the user anonymity, is not a perfect venue for truly deliberative discussions. Nonetheless, these findings do not denote that the internet is unsuitable for deliberative discussions given the right conditions. Most likely, some sort of institutional arrangements and rules of conduct enforced through moderation would enhance the quality of on-line debates (cf. Wright & Street 2007). Even so, the major obstacle in any attempt towards citizen deliberation both on- and off-line, which was quite evident in this study as well, is how to engage citizens beyond the active segments. To summarise, the opportunities for on-line political discussions on the internet do not automatically bring about democratic benefits nor do they succeed in attracting citizens who are not already politically active and interested. In other words, public deliberation in its true form has not yet moved on-line.
Acknowledgements
The author wishes to thank Professor Heikki Paloheimo at the University of Tampere for providing the dataset from the national election survey. Naturally, I am solely responsible for any shortcomings in the presentation and use of the data. The data had been collected as part of The Academy of Finland project nr. 114851 named "Styles of Political Participation and Modes of Democracy."
Notes:
1.	Finland still ranks among the top 30 countries in the world according to the website of Internet World Stats, www.internetworldstats.com.
2.	Source: www.internetworldstats.com accessed in January 2007.
3.	The national election survey was conducted with funding granted by the Academy of Finland.
4.	According to Alexa web rankings website - www.alexa.com - accessed February 8th 2007, the www.suomi24.fi website is the 9th most visited site in Finland. The www.iltalehti.fi website is ranked 15th and the www.jippii.fi website 88th. The sfnet.keskustelu.politiikka group has 99 subscribers which is the highest number I noted among Finnish political newsgroups.
5.	The reliability of the scheme was checked using the formula for interreliability suggested by Holsti (1969). After initial briefing, the result of the test was a very satisfying .91 (recommended minimum is .90).
References:
Asp, Kent. 1986. Mäktiga Massmedier: Studier I Politisk Opinionsbildning. Stockholm: Akademilitteratur. Barber, Benjamin. 1984. Strong Democracy: Participatory Politics for a New Age. Los Angeles: University of California Press.
Bentivegna, Sara. 1998. Talking Politics on the Net. Research paper. The Joan Shorenstein Center on
Press, Politics and Public Policy. Bimber, Bruce. 2000. The Study of Information Technology and Civic Engagement. Political
Communication 17, 4, 329-333. Boogers, Marcel and Gerrit Voerman. 2003. Surfing Citizens and Floating Voters: Results of an Online Survey of Visitors to Political Web Sites during the Dutch 2002 General Elections. Information Polity 8, 1-2, 17-27.
Budge, Ian. 1996. The New Challenge of Direct Democracy. Oxford: Blackwell.
Carling, Maria. 2007. No Need for Words: In Finland, Silence Serves to Connect. Psychology Today 40, 1, 28.
Carlson, Tom and Kim Strandberg. 2005. The 2004 European Parliament Election on the Web: Finnish
Actor Strategies and Voter Responses. Information Polity 10, 3-4, 189-204. Coleman, Stephen and John Goetze. 2001. Bowling Together: Online Public Engagement in Policy
Deliberation. London: Hansard Society. Cornfield, Michael, Lee Rainie and John Horrigan. 2003. Untuned Keyboards, Online Campaigners,
Citizens, and Portals in the 2002 Elections. Report. Pew Internet & American Life Project. Dahl, Robert. 1989. Demokratin och Dess Antagonister. New Haven: Yale University. Dahlberg, Lincoln. 2001. Computer-Mediated Communication and the Public Sphere: A Critical Analysis. Journal of Computer Mediated Communication 7, 1 <http://www.ascusc.org/jcmc/vol7/ issue1/dahlberg.html>
ÜDalton, Russell. The Decline of Party Identifications. In R. Dalton and M. Wattenberg (eds.), Parties Without Partisans. Political Change in Advanced Industrial Democracies, 19-37. New York: Oxford jQ	University Press.
^	Davis, Richard. 1999. The Web of Politics: The Internet's Impact on the American Political System. New
York: Oxford University Press. Dertouzos, Michael. 1997. What Will Be: How the New Information Marketplace Will Change Our Lives. San Fransisco: Harper.
Djupsund, Göran and Tom Carlson. 2003. Catching the "Wired Voters"? Campaigning on the Internet.
Nordicom Information 25, 1-2, 39-50. Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper. Fishkin, James. 1995. The Voice of the People: Public Opinion and Democracy. New Haven: Yale University Press.
^	Fuchs, Christian. 2006. eParticipation Research: A Case Study on Political Online Debate in Austria.
ICT&S research paper No. 1, University of Salzburg. Gibson, Rachel, Lusoli Wainer and Stephen Ward. 2005. Online Participation in the UK: Testing a "Contextualized" Model of Internet Effects. British Journal of Politics and International Relations 7, 561-583.
Gibson, Rachel and Andrea Römmele. 2005. Truth and Consequence in Web Campaigning: Is There an Academic Digital Divide. European Political Science 4, 3, 273-287. ^	Habermas, Jürgen. 1962. The Structural Transformation of the Public Sphere: An Inquiry into a Category
QQ	of a Bourgeois Society. Cambridge: MIT Press.
Hacker, Kenneth and Jan Van Dijk. 1999. What is Digital Democracy? In K. Hacker and J. Van Dijk (eds.), Digital Democracy: Issues of Theory and Practice, 1-9. London: Sage.
Hill, Kevin and John Hughes. 1998. Cyberpolitics: Citizen Activism in the Age of the Internet. New York:	00
Rowman & Littlefield.
Holsti, Ole. 1969. Content Analysis for the Social Sciences and Humanities. Reading: Addison-Wesley.
Jankowski, Nicholas and Martine Van Selm. 2000. The Promise and Practice of Public Debate in Cyberspace. In K. Hacker and J. Van Dijk (eds.), Digital Democracy: Issues of Theory and Practice, 149-165. London: Sage.
Jankowski, Nicholas and Renée Van Os. 2004. Internet-Based Political Discourse: A Case Study of Electronic Democracy in Hoogeveen. In P. Shane (ed.), Democracy Online: The Prospects for Political Renewal Through the Internet, 181-194. New York: Routledge.
Johnson, Thomas and Barbara Kaye. 2003. A Boost or Bust for Democracy? How the Web Influenced Political Attitudes and Behaviors in the 1996 and 2000 Presidential Elections. Harvard International Journal of Press/Politics 8, 3, 9-34.
Kamarck, Elaine. 1999. Campaigning on the Internet in the Elections of 1998. In E. Kamarck and J. Nye (eds.), Democracy.com? Governance in a Networked World, 81-103. Hollis: Hollis Publishing.
Kavanaugh, Andrea, Philip Isenhour, Matthew Cooper, John Carroll, Mary Beth Rosson, and Joseph Schmitz. 2005. Information Technology in Support of Public Deliberation. In P. Van den Bessalaar, G. de Michelis, J. Preece and C. Simone (eds.), Communities and Technologies II, 19-40. Dordrecht: Kluwer Academic Publishers.
Kaye, Barbara and Thomas Johnson. 2002. Online and in the Know: Uses and Gratifications of the Web for Political Information. Journal of Broadcasting & Electronic Media 46, 1, 54-71.
Kippens, Grant and Gordon Jenkins. 2004. The Challenge of E-Democracy for Political Parties. In P. Shane (ed.), Democracy Online: The Prospects for Political Renewal Through the Internet, 253-265. New York: Routledge.
Lusoli, Wainer. 2005. A Second-order Medium? The Internet as a Source of Electoral Information in 25 European Countries. Information Polity 10, 3-4, 247-265.
Lusoli, Wainer, Stephen Ward and Rachel Gibson. 2002. Political Organizations and Online
Mobilization: Different Media - Same Outcomes? New Review of Information Networking 8, 89107.
Mair, Peter and Ingrid Van Biezen. 2002. Party Membership in Twenty European Democracies, 19892000. Party Politics 7, 1, 5-21.
Muhlberger, Peter. 2004. Access, Skill, and Motivation in Online Political Discussion: Testing
Cyberrealism. In P. Shane (ed.), Democracy Online: The Prospects for Political Renewal Through the Internet, 225-238. New York: Routledge.
Negroponte, Nicholas. 1995. Being Digital. New York: Knopf.
Nixon, Paul and Hans Johansson. 1999. Transparency Through Technology: The Internet and Political Parties. In B. Hague and B. Loader (eds.), Digital Democracy: Discourse and Decision Making in the Information Age, 135-153. London: Routledge.
Norris, Pippa. 1999. Who Surfs? New Technology, Old Voters and Virtual Democracy. In E. Kamarck and J. Nye (eds.), Democracy.com? Governance in a Networked World, 71-94. Hollis: Hollis Publishing.
Norris, Pippa. 2001a. A Virtuous Circle: Political Communication in Post-Industrial Societies. New York: Cambridge University Press.
Norris, Pippa. 2001b. Digital Divide? Civic Engagement, Information Poverty & the Internet Worldwide. New York: Cambridge University Press.
Norris, Pippa. 2002. Democratic Phoenix: Political Activism Worldwide. New York: Cambridge University Press.
Norris, Pippa. 2003. Preaching to the Converted? Pluralism, Participation and Party Websites. Party Politics 9, 1, 21-45.
Papacharissi, Zizi. 2002. The Virtual Sphere: The Internet as a Public Sphere. New Media & Society 4, 1, 9-27.
Papacharissi, Zizi. 2004. Democracy Online: Civility, Politeness, and the Democratic Potential of Online Political Discussion Groups. New Media & Society 6, 2, 259-283.
Polat, Rabia. 2005. The Internet and Political Participation: Exploring the Explanatory Links. European Journal of Communication 20, 4, 435-459.
Rheingold, Howard. 1993. The Virtual Community: Homesteading on the Electronic Frontier. Reading: Addison-Wesley.
Scarrow, Susan. 2000. Parties without Members? Party Organization in a Changing Electoral
Environment. In R. Dalton and M. Wattenberg (eds.), Parties Without Partisans. Political Change in Advanced Industrial Democracies, 79-101. New York: Oxford University Press.
Schwartz, Edward. 1996. Net activism: How Citizens Use the Internet. Sebastopol: Songline Studios.
Setälä, Maija, Kimmo Grönlund and Kaisa Herne. 2007. Kohti Valistunutta Kansalaismielipidettä?: Kansalaiskeskustelukokeilu Turussa Syksyllä 2006. Oikeus 36, 2, 231-240.
Strandberg, Kim. 2005. "Town Hall" Meetings for the Masses or "Social Clubs" for the Motivated? - A Study of Citizens' Discussions on the Internet. World Political Science Review 1, 1, 1-19.
Stromer-Galley, Jennifer, Kristen Foot, Stephen Schneider and Elena Larsen. 2001. How Citizens Used the Internet in Election 2000. In S. Coleman (ed.), Elections in the Age of the Internet: Lessons From the United States, 21-26. London: Hansard Society.
Sunstein, Cass. 2001. Republic.com. Princeton: Princeton University Press.
Wattenberg, Martin. 2000. The Decline of Party Mobilization. In R. Dalton and M. Wattenberg (eds.), Parties Without Partisans. Political Change in Advanced Industrial Democracies, 64-77. New York: Oxford University Press.
Wilhelm, Anthony. 1999. Virtual Sounding Boards: How Deliberative is Online Political Discussion? In B. Hague and B. Loader (eds.), Digital Democracy: Discourse and Decision Making in the Information Age, 154-77. London: Routledge.
Witschge, Tamara. 2004. Online Deliberation: Possibilities of the Internet for Deliberative Democracy. In P. Shane (ed.), Democracy Online: The Prospects for Political Renewal Through the Internet, 109122. New York: Routledge.
Woo-Young, Chang. 2005. Online Civic Participation, and Political Empowerment: Online Media and Public Opinion Formation in Korea. Media, Culture & Society 27, 6, 925-935.
Wright, Scott and John Street. 2007. Democracy, Deliberation and Design: the Case of Online Discussion Forums. New Media & Society 9, 5, 849-869.
S
i
0
£ *
1 *
(R 0
C
«
00 00
Appendix: Coding Scheme for Content Analysis of Threads and Messages
Threads:	00
Number of participants
Number of individual messages
Number of reply levels
The topic of the first message in each thread:
1.	Political issue
2.	Politician's/party's image
3.	Political process
4.	Metatheme
5.	Other
The type of content in the first message in each thread:
1.	Fact
2.	Opinion
3.	Question
4.	Announcement
5.	Statement
6.	Other
The tone of the first message in each thread:
1.	Negative
2.	Neutral
3.	Positive
Use of validation in first message of each thread:
1.	Internal validation (i.e. referring to arguments already posted in the forum)
2.	External validation (i.e. referring to external sources outside of forum)
3.	None
Occurrence of conflict in thread: Yes or No
Is some form of middle ground/mutual agreement achieved in thread with conflict: Yes or No
Individual messages:
Type of message: Original (first in thread) or Reply The topic of the message:
1.	Political issue
2.	Politician's/party's image
3.	Political process
4.	Metatheme
5.	Other
The type of content in the message:
1.	Fact
2.	Opinion
3.	Question
4.	Announcement
5.	Statement
6.	Other
The tone of the message:
1.	Negative
2.	Neutral
3.	Positive