Virtual Users Support Forum: Do Community Members Really Want to Help You?

Abstract

The survival of a virtual community is guaranteed by the users' creation of content. However, the literature has found that the percentage of users who create innovative content is very modest. The content contribution process can also be interpreted as a social collective action in which we-intentions play a primary role. Nevertheless, some people choose not to participate in the collective action, but to benefit from the community's resources and to maximize individual outcomes. In this study (N=250), we investigated the effects of the free-riding tendency, conceived as the willingness to maximize personal outcomes. The specific setting was a virtual support forum, the most common type of web platform, generally used instrumentally by web users to find information and solutions to specific problems. We used the theory of planned behavior theoretical framework, plus social influence variables to test the effect of the free-riding tendency as a drawback for contributions, considering both the role of individual and we-intentions on the observed behavior. Findings showed that neither we-intentions nor I-intentions predicted the actual contribution behavior. Both types of intentions and contribution behavior were negatively influenced only by the free-riding tendency construct. Considerations and future developments of these results are discussed.

Introduction

The development of the Web 2.0 has rapidly changed how people use the Internet by facilitating the relational processes. The web today represents a shift from the society of information to the society of participation. Many Websites offer content directly created by users [user-generated content (UGC¹)], primarily in text form. Some scholars have suggested that knowledge—that is, creation and sharing of UGC—is the main value around which an online community develops.^2–5 The present study focuses specifically on users support forums, virtual communities of users, who help each other to figure out the best ways to use a product. In such user-to-user platforms, some users ask questions and others provide answers concerning technical issues. Virtual forums are the most common web platform,⁶ and those of a technical nature are generally accessed instrumentally by users to find information.

The contribution process can be considered as a collective action carried out by the members of a virtual community. Many of the behaviors observable in virtual communities are described by their actors through the use of collective concepts, based on a strong sense of “we-ness”⁷: community members tend to consider themselves as “part of the social fabric of the virtual community.”^2(p7) Thus, online social interactions may be described as intentional social actions,⁸ a concept that draws its origins from the philosophy of intentionality.^9–12 For instance, Tuomela^13(p2) introduced the concept of “we-intention” as “a commitment of an individual to participate in joint action and involves an implicit or explicit agreement between the participants to engage in that joint action.” In we-intentions, the constitutive intentions of the individuals are interlocked: every agent has an intention in favor of the efficacy of others' intentions.¹⁴ From the members' viewpoint, the community offers reasons for thinking and acting collectively: the community goals, values, and beliefs can determine the reasons for the group action.¹⁵ The shared intentions may be conceived both in terms of a group action (“We intend to perform action X”) and of a person acting with the group (“I intend that we—other members and I—will do action X”).¹⁶ It is worthwhile noting that in these formulations, actions and intentions only have meaning if the group acts collectively. Individual action by oneself is not enough to attain one's goal in either case.¹⁷

Bagozzi and colleagues^2,16–18 studied online participation and social interactions in virtual communities, by reformulating these considerations and introducing the construct of we-intentions within a well-known theoretical model of behavioral prediction, such as the theory of planned behavior (TPB¹⁹). The TPB has been successfully employed in hundreds of applied studies, across different behavioral domains (e.g., see Refs.^20,21). This theory maintains that behavior is directly influenced by one's intention to act and the perceived control over the behavior; intention to act, in turn, is affected by the attitudes toward the act, subjective normative pressure to behave according to relevant referents' expectations, and perceived behavioral control (PBC). Intentions under the TPB have been construed in a strictly individual sense.¹⁷ For instance, Eagly and Chaiken^21(p168) interpret an intention as a “person's motivation in the sense of his or her conscious plan to exert effort to carry out a behavior by him or herself.” Bagozzi and colleagues^2,16–18 afterward adopted this as the definition of “I-intentions.”

The TPB and other contemporary models for behavioral predictions deriving from attitude, generally explain the motivations that lead to the formation of individual intentions and behavior, but do not consider social aspects in carrying out group acts. Thus, in their work, Bagozzi and colleagues^2,16–18 captured the group dimension of a virtual community, by adding the concepts of group identification and group norms to the general model of behavioral prediction as additional determinants of we-intentions, alongside the other antecedents of intentions. These studies focused primarily on the processes leading to the formation of the we-intentions, confirming the predictive and explicatory power of TPB in this behavioral domain. Nevertheless, the literature has not yet adequately investigated the relationship between the collective intentions, the individual intentions, and the actual behavior of contribution; the current study is aimed at filling this gap.

The literature has shown that the percentage of members who actively produce content is definitely very low. Nielsen²² empirically analyzed the user content production rate of several virtual communities. The author found that participation generally follows a ratio of 1-9-90: 1 percent of users represents the active content producers, 9 percent is composed of editors, and 90 percent is defined as audience.

This empirical evidence, when considering the collective intentionality framework, raises an important question: why do only few community members actively contribute, even when it is not in their interest to do so, whereas most members do not?

The literature on social dilemmas suggests that in social groups—as users support forums—whose survival depends on cooperation,²³ the choice is between maximizing the outcomes for the group, choosing not to increase individual outcomes, or vice versa, maximizing personal outcomes without contributing to the group.^24,25 This latter phenomenon is generally known as free-riding. Tuomela²⁶ offered a philosophical analysis of free-riding in close relation to a group acting collectively to produce a public good. The author emphasized the role of the individual decision in not participating in the collective action, considering the defection as more advantageous.

On a closer inspection, we may consider online free-riding tendency as a continuum with two extremes: contributing in turn to the community resources or escaping this task.²⁴ The 1-9-90 rule²² emphasizes the imbalance in contribution behavior. For these reasons, our aim was to better understand the psychological processes that could impede or promote the actual contribution process in participating in online support forums.

To reveal such processes, we used an augmented version of the TPB (see Fig. 1 in the Results section) to capture users' readiness to perform a given behavior in a virtual community. We considered both we-intentions and individual intentions as proximal determinants of contribution behaviors; we also hypothesized intentions (shared and individual) to be positively affected by the TPB determinants (attitudes toward the behavior, PBC, and perceived social pressure; see Refs.^2,16,17). Further, to shape the social dimension of the contribution behavior in a virtual community, social identity and group norms were added as predictors of both intentions (see Refs.^2,16–18 for a similar approach). The collective action, in fact, may be shaped by the influence that each group member exerts on the other members, through the intentions.

FIG. 1.

Findings for TPB augmented model, completely standardized parameter (N=250). The gray arrows represent hypothesized, but nonsignificant paths. Errors and correlations are omitted for the sake of simplicity. *p<0.05; **p<0.01; ***p<0.001; ^†p=0.06; a=fixed parameter.

In the considered behavioral context, both individual and shared users' intentions might be inhibited by the willingness to maximize individual outcomes (free-riding tendency), leading to a lower rate in contribution. However, we assumed that the free-riding tendency takes place more on an individual level. As a consequence, the free-riding tendency was hypothesized to directly negatively affect both types of intentions and the contribution behavioral goal. To our knowledge, this is the first study considering the inhibition effects of free-riding on intentional and actual behavioral processes.

Methods

Procedure

Data collection was performed on a single online support forum, HTCBLOG (www.htcblog.com), aimed at solving technical issues related to a specific smartphone brand. The community was composed approximately of 30,000 registered members who share information on their experiences in using mobile devices through comments and forum discussions.

Participants, who took part in the survey voluntarily, were recruited through an invitation e-mail sent by the community administrators. We also put a banner, with a hyperlink to the survey, in the community home page and in some pop-up messages. To encourage the questionnaire completion, we also launched a prize competition.

We designed a two-phase data collection study with both self-reported and objective behavioral measures. In the first wave, data collection was realized through a web-survey programmed with PHP language and supported by a MySQL database. The survey was hosted on the web-community's server and was graphically similar to the htcblog.com Website, so not to confuse the participants.²⁷ The web master granted the privacy of personal data treatment. We measured all the variables hypothesized as antecedents of contribution behavior: attitudes, PBC, subjective and group norms, social identity, I- and we-intentions, and free-riding tendency. In the follow-up phase, about one month later, we measured the actual contribution of every single user, by counting his/her interventions on the forum.

Sample characteristics

A total of 263 participants completed the survey; 219 were men (84.9 percent) and 44 were women (15.1 percent). Their mean age was 33.26 years (SD=14.34).

The average access to the community, measured on a categorical scale with four intervals, revealed that 17.8 percent of total users accessed the community at least once a day, 27.9 percent at least 5 times a week, 24.8 percent between 5 and 15 times a month, and 29.5 percent accessed the community<15 times a month.

Measures

Seven-point items, derived from earlier literature, were adapted to the behavioral domain of contribution. A summary of the measures is listed in Table 1.

Table 1.

List of Measures

Attitudes

“Accessing htcblog.com for contributing to the contents of the community and adding comments to the discussions of the community, in the next month, would be…”

• Foolish–Wise

• Pleasant–Unpleasant

• Bad–Good

• Depressing–Funny

One was assigned to the negative pole and 7 to the positive one, with 4 indicating neither/nor.

Subjective norms

• Most people who are important in my life think that I should (7)/should not (1) contribute to this community by entering one or more comments during the next few months.

• Most people who are important to me would approve (7)/disapprove (1) of me contributing to this community by entering one or more comments during the next few months.

Group norms

“Using htcblog.com to contribute by entering comments about HTC smartphones and problems with phones based on Android and Windows Mobile with other people with similar interests can be considered a goal. For the listed community members, please estimate the strength with which everyone pursues this goal”:

• I pursue the goal of contributing by entering comments on htcblog.com;

• The active member pursues the goal of contributing by entering comments on htcblog.com;

• The reader member pursues the goal of contributing by entering comments on htcblog.com;

• The moderators of the community pursue the goal of contributing by entering comments on htcblog.com.

Seven-point scales anchored by not at all (1) and very strongly (7), and 4 indicating quite enough.

Perceived behavioral control

• How much control do you think you have over contributing to this community by entering one or more comments in htcblog.com during the next few months?

Seven-point scale ranging from (1) no control, to (7) total control, and the midpoint (4) moderate control.

• How easy or difficult do you think it is for you to contribute with one or more comments in htcblog.com during the next month?

Seven-point scale ranging from (1) very difficult to (7) very easy.

Social identity

• I evaluate positively being a member of htcblog.com

• Being a member of htcblog.com provides me with prestige

• I feel tied to the other members of htcblog.com

• I feel attached to the other members of htcblog.com

• Being one of the htcblog.com members is something I often think about

• I often think about being a member of htcblog.com during the day

• I do not usually behave as a typical member of htcblog.com (reversed score)

• I would feel uncomfortable if the media (TV, newspapers, or other web communities) criticized the members of htcblog.com

Seven-point scales anchored by (1) strongly disagree and (7) strongly agree (4=neither agree/nor disagree).

I-intentions

• How likely is your intention as an individual to contribute to the contents of htcblog.com, by adding comments or suggestions to the discussions of the community in the next month?

Seven-point scale anchored by (1) very unlikely and (7) very likely (4=neither agree/nor disagree).

• As an individual, I would like to contribute to the contents of htcblog.com, by adding comments or suggestions to the discussions of the community, during the next month.

Seven-point scale anchored by (1) strongly disagree and (7) strongly agree (4=neither agree/nor disagree).

We-intentions

• I intend that our group, the other members of htcblog.com and I, contribute to the contents of htcblog.com by adding comments or suggestions to the discussions of the community, during the next month.

• We—the other members of htcblog.com and I—intend to contribute to the contents of the community by adding comments or suggestions to the discussions of the community, during the next month.

Seven-point scales anchored by (1) strongly disagree and (7) strongly agree (4=neither agree/nor disagree).

Free-riding tendency

• By not contributing with comments on htcblog.com, I won't end up worse off than anyone else.

• If everyone else contributes a lot with comments on htcblog.com, then I get the best of both situations if I do not.

• If others put a large amount of effort into contributions on htcblog.com and I do not, that is my gain and their loss.

• Let the suckers put a lot of effort into contributing on htcblog.com

• By not contributing with comments on htcblog.com, I will come out better off than those who contribute a lot.

Seven-point scales anchored by (1) strongly disagree and (7) strongly agree (4=neither agree/nor disagree).

To measure attitudes toward contributions, we used four semantic differential items derived from Ajzen,¹⁹ whereas two items for PBC were derived from Perugini and Bagozzi.²⁸ Two and four items were employed to measure, respectively, subjective and group norms.¹⁷ The latter was measured with items concerning the degree of shared goals between the self and the other group members. To measure identification processes, we used a shortened adapted version of the scale proposed by Capozza et al.,²⁹ composed of eight agreement items. Two items were adapted from Bagozzi and Lee,³⁰ respectively, for we-intentions and I-intentions.

The free-riding tendency, conceived as maximization of individual outcomes, was assessed with five agreement items adapted from a scale developed by Markòczy³¹ to measure free-riding in relation to energy saving.

Finally, the dependent variable of the present study was the contribution behavior. For every participant, we counted the number of messages posted in the forum and the number of comments posted in the internal pages of htcblog.com, in the month following the date of questionnaire completion.

Results

Descriptive statistics

All variables were normally distributed and the reliability coefficients were greater than 0.69 for all the scales except for attitude, for which the elimination of one item improved the alpha to 0.82 (see Table 2 for descriptive statistics and reliabilities of each construct).

Table 2.

Means, Standard Deviations, and Reliabilities for Each Construct

	M	SD	α
Attitudes	5.55	1.11	0.82^a
Subjective norms	4.90	0.91	0.69
Group norms	4.46	1.15	0.78
Perceived behavioral control	4.15	1.34	0.70
Social identity	4.14	0.94	0.78
We-intentions	4.78	1.25	0.93
I-intentions	4.41	1.36	0.87
Free-riding tendency (maximization of individual outcomes)	3.10	1.36	0.91
Contributions (posting frequency of comments)	7.69	77.83	–

Alpha calculated on three items.

Both the I-intentions and the we-intentions to contribute to the community were positive. However, I-intentions of members registered for less than one year (n=193; M=4.90, SD=1.40) were higher than those of older members (n=57; M=4.25, SD=1.32), t(248)=3.11, p<0.01. No significant differences emerged for we-intentions.

Finally, the analysis of comments posting frequency indicated that 83 percent of the respondents never posted messages during the considered one-month period; 14 percent posted fewer than 10 times, 1.6 percent contributed more than 10 times, but less than 50 times, and only 1.4 percent contributed more than 50 messages (M=7.69, SD=77.83). It is worthwhile noting that the percentage distribution is consistent with Nielsen's²² observations. For the subsequent analyses, frequency of posted messages was recorded as follows*: never (0), from 1 to 10 messages (1), 10 to 50 (2), and more than 50 messages on the forum (3) (for a similar procedure, see Tsai and Bagozzi⁷).

Confirmatory factor analysis

Confirmatory factor analysis was used to investigate discriminant validity of all the considered constructs (LISREL 8.5 ver.³²). We ran a model with nine latent variables and two indicators for each latent construct, except for the contribution behavior that had only one observed variable (the input matrix is reported in Table 3). To reduce the number of parameters to be estimated, we used the partial disaggregation method³³ for the variables with more than two items.

Table 3.

Correlation Matrix of Observed Variables

	ATT 1	ATT 2	SN 1	SN 2	GRN 1	GRN 2	PBC 1	PBC 2	WEINT 1	WEINT 2	IINT 1	IINT 2	SID 1	SID 2	FREE 1	FREE 2	CONTR
ATT 1	1
ATT 2	0.64^**	1
SN 1	−0.02	0.05	1
SN 2	0.09	0.05	0.55^**	1
GRN 1	0.25^**	0.27^**	0.22^**	0.21^**	1
GRN 2	0.28^**	0.17^**	0.24^**	0.27^**	0.54^**	1
PBC 1	0.12	−0.06	0.09	0.10	0.07	0.30^**	1
PBC 2	0.13^*	−0.02	0.12	0.15^*	0.22^**	0.40^**	0.53^**	1
WEINT 1	0.34^**	0.29^**	0.07	0.11	0.46^**	0.43^**	0.16^**	0.42^**	1
WEINT 2	0.32^**	0.22^**	0.07	0.16^*	0.41^**	0.50^**	0.24^**	0.47^**	0.85^**	1
IINT 1	0.26^**	0.04	0.10	0.12	0.30^**	0.51^**	0.51^**	0.63^**	0.52^**	0.56^**	1
IINT 2	0.1^**	0.06	0.11	0.14^*	0.31^**	0.53^**	0.41^**	0.59^**	0.58^**	0.64^**	0.82^**	1
SID 1	0.30^**	0.13^*	0.07	0.19^**	0.38^**	0.55^**	0.30^**	0.46^**	0.46^**	0.51^**	0.56^**	0.57^**	1
SID 2	0.34^**	0.24^**	0.13^*	0.26^**	0.44^**	0.53^**	0.25^**	0.43^**	0.47^**	0.52^**	0.50^**	0.51^**	0.75^**	1
FREE 1	0.00	−0.14^*	0.07	0.00	−0.03	0.19^**	0.32^**	0.25^**	0.05	0.07	0.20^**	0.19^**	0.26^**	0.19^**	1
FREE 2	−0.04	−0.12	0.10	0.01	−0.07	0.14^*	0.26^**	0.22^**	0.05	0.05	0.14^*	0.14^*	0.25^**	0.15^*	0.78^**	1
CONTR	0.05	0.07	0.01	0.06	0.04	0.02	0.06	0.13^*	0.13^*	0.14^*	0.09	0.06	0.11	0.12	−0.11	−0.05	1

p<0.05; ^**p<0.01; ^***p<0.001; ATT, attitudes; SN, subjective norms; GRN, group norms; PBC, perceived behavioral control; WEINT, we-intention; IINT, I-intention; SID, social identity; FREE, free-riding tendency; CONTR, contribution behavior.

The model's goodness-of-fit was evaluated by the χ² test, the comparative fit index (CFI), and the standardized root mean square residual (SRMR). Satisfactory model fits are obtained when χ² is nonsignificant, CFI is ≥0.95, and when SRMR is ≤0.05.³⁴ This baseline model fits the data well: χ²(84)=164.03, p≈0.00, CFI=0.98; SRMR=0.047. Even if the χ² test was significant, all the other indexes were satisfactory. Factor loadings were consistent and all between 0.64 and 0.96. Discriminant validity emerged from the fact that latent variables showed correlations lower than 1.0 (Table 4) and the confidence interval did not include the perfect correlation.

Table 4.

Correlation Matrix of Latent Constructs

	ATT	SN	GRN	PBC	WEINT	IINT	SID	FREE	CONTR
ATT	1.00
SN	0.08	1.00
GRN	0.38^***	0.42	1.00
PBC	0.10	0.21^*	0.52^***	1.00
WEINT	0.39^***	0.17^*	0.63^***	0.53^***	1.00
IINT	0.22	0.17^*	0.66^**	0.81^**	0.69^**	1.00
SID	0.37^***	0.26^***	0.73^***	0.60^***	0.61^***	0.70^***	1.00
FREE	−0.08	0.04	0.16^*	0.37^***	0.07	0.22^**	0.28^***	1.00
CONTR	0.07	0.05	0.11	0.58^*	0.18^**	0.15	0.11^*	−0.10^*	1.00

p<0.05; ^**p<0.01; ^***p<0.001; ATT, attitudes; SN, subjective norms; GRN, group norms; PBC, perceived behavioral control; WEINT, we-intentions; IINT, I-intentions; SID, social identity; FREE, free-riding tendency; CONTR, contribution behavior.

The correlation between I- and we-intentions was moderately large (Φ=0.69, p<0.001, SE=0.04). Similar results were also reported in a number of previous studies.^2,16,17 However, since the distinction between the two types of intentions was decisive for our hypotheses, we used also a more restrictive evaluation procedure. We compared the baseline model—where the correlation between the latent construct of I- and we-intentions was freely estimated—with a second model where such a correlation was constrained to be equal to 1 [χ²(85)=316.82, p≈0.0]. The chi-square difference test was then used³⁵: χ²_d(1)=152.79, p<0.001. The significance of the test indicated that the two models differ from each other and consequently the two types of intentions were to be considered as distinct constructs.

Structural models

We then used structural equation modeling with latent constructs (LISREL 8.5 ver.³²) to test the hypothesized model: an augmented version of TPB considering also social influence variables and free-riding tendency (Fig. 1).

Findings showed satisfactory fit statistics: χ²(88)=169.29, p≈0.00, CFI=0.98; SRMR=0.047. Attitude positively determined we-intentions (γ=0.15, p<0.01), but not I-intentions. PBC positively influenced both I- (γ=0.62, p<0.001) and we-intentions (γ=0.30, p<0.001), as well as group norms did (γ=0.19, p<0.01 and γ=0.28, p<0.001, respectively). Social identity emerged as a predictor of I-intentions (γ=0.23, p<0.01) and marginally of we-intentions (γ=0.21, p=0.06). Subjective norms did not predict the contribution intentions.

Both types of intentions failed to predict contribution,^36,† and the observed behavior was negatively influenced by the free-riding tendency (γ=−0.19, p<0.01), which inhibited also I-intentions (γ=−0.11, p<0.01) and we-intentions (γ=−13, p<0.01).

The model accounted for a great amount of variance in I- (77 percent) and we-intentions (52 percent), but a smaller amount in contribution behavior (7 percent).

We also performed formal tests of mediation to assess the presence of residual direct effects of the antecedents on the behavior not specified in the hypothesized model.^2,17 The inspection of chi-square difference³⁵ in nested models revealed that each additional path was not predictive of the target behavior (see Table 5 for tests of rival hypotheses).

Table 5.

Mediation Tests for Rival Hypotheses

Baseline model M1 χ²(88)=169.29, p≈0.00; CFI=0.98; SRMR=0.047
	Model	Goodness-of-fit	Tests of hypotheses
M₂	Attitude→Contribution	χ²(87)=165.96, p=0.0	M₁−M₂: χ²_d(1)=3.33, p=0.07
M₃	Subjective norms→Contribution	χ²(87)=165.78, p=0.0	M₁−M₄: χ²_d(1)=3.51, p=0.06
M₄	Group norms→Contribution	χ²(87)=165.87, p=0.0	M₁−M₅: χ²_d(1)=3.42, p=0.065
M₅	Social identity→Contribution	χ²(87)=165.53, p=0.0	M₁−M₆: χ²_d(1)=3.73, p=0.06

Discussion

Recent research has highlighted that the long-term life of virtual communities depends critically on members' contribution.³⁷ The principal aim of this study was to investigate the effect of the free-riding tendency on the behavioral goal of contribution and shed a first light on the processes that may inhibit such a performance in the specific setting of a virtual user support forum.

We included the free-riding tendency in an augmented version of the TPB¹⁹ (see Fig. 1). In line with previous findings,^2,16,17 perceived control, group norms, and social identity positively influenced the we-intentions to contribute. The same variables determined I-intentions too, but the effect of the perception of control over the behavior—which includes internal and external factors promoting or impeding the action—was much stronger. Attitude toward contribution—framed as a collective action—determined only we-intentions.

In this behavioral domain, subjective norms did not show significant effects. This is not surprising since participation in virtual communities is anonymous, and members may feel little need to comply with relevant referents' expectations.^7,38

Group norms showed strong influence on I- and we-intentions. Usually, within a discussion forum, the forum moderators and administrators who ensure it is observed make netiquette^‡ salient. We speculate that even if most of the members access the forum mainly to satisfy personal needs, certain norms must be respected so that the community will not extinguish, and relevant group members' expectations are not deceived.

Our results reinforce the importance of distinguishing between individual intentions and we-intentions in studying collective action in a virtual community, since psychological processes that underlie these constructs are distinct and show different effects.

The hypothesized model accounted for a great amount of variance in both individuals and we-intentions. Previous studies^7,38 suggested that web community members manifested collective intentions toward contributions; still, in our model, we-intentions did not influence the observed behavior. Since the sample was composed of a consistent number of novice users who reported a significantly greater score for I-intentions than older members, and given the instrumental nature of a virtual forum, one could expect that individual intentions would represent a stronger predictor for contributions. Nevertheless, even I-intentions showed no significant effects on it. Indeed, findings revealed that the free-riding tendency, acting at the individual level, could prevent both intentions and directly hamper the contribution performance.

According to Bagozzi and Lee,³⁰ social action may be classified on a continuum in which at one extreme there are fully cooperative actions, whereas on the other side there are minimally cooperative actions. These latter actions refer to situations in which members of a group share common goals, but not a joint action to cooperate together. We posit that forum members in such a situation, if influenced by the free-riding tendency, may lack the personal motivation—part of the original postulation of we-intentions¹²—in playing their role in the contribution process.

Moving from this point, we suggest that a virtual support forum may be perceived as an extended repository of information that is consulted, or even exploited, only when needed. Members identify with the community group and follow its norms and rules; although, if influenced by the maximization of individual outcomes, once the information sought is found they miss the motivational boost to contribute.

It is worthwhile noting that this is one of the few studies measuring the actual behavior; however, the model accounted for only 7 percent of variance in the behavioral goal. As suggested by the literature,^39,40 the predictive power of behavioral models is often low. We think that this is especially true in the case of virtual contributions, which is a complex behavior: to contribute, people have to subscribe to the community and then login in.

Tsai and Bagozzi⁷ found that the model of goal-directed behavior²⁸ (MGB—an augmented version of TPB) plus we-intentions accounted for 57 percent in we-intentions, and a higher level, 15 percent, than the present study of contribution. Such a difference could be mainly due to the nature and the aims of the considered communities: Tsai and Bagozzi analyzed a community aimed at sharing photos and information about personal travel experiences. A virtual support forum, instead, is structured mainly for finding information and solutions to problems. However, since also earlier literature in other domains has highlighted that MGB often has a superior predictive and explicative power than TPB,^41,42 future research should address this issue in the virtual support forum context.

Moreover, although a greater improvement is needed in its measurement on the Internet, the free-riding tendency may represent a process that hinders contributions. As expressed by Richetin et al.⁴³ performing a behavior is not simply opposed to not performing the same behavior, because they rely on separate goals. For these reasons, future research should focus on contribution abstention as a separate goal, and investigate the variables that lead (and not prevent) a user not to contribute, looking at the other side of the process of participation in a virtual community.

Footnotes

Acknowledgment

The authors would like to thank the web master as well as the members of htcblog.com for supporting us in the data collection process.

Author Disclosure Statement

No competing financial interests exist.

*

We divided the number of contributions using five intervals, as suggested by Bagozzi and Tsai.⁷ In this way, the variable was normalized and the data variability was reduced.

†

As suggested by Dholakia and Bagozzi,³⁶ we also ran a series of regressions to test whether, considering subsamples of novice users and experienced users, the effects of I- and we-intentions on contribution behavior were significant. However, no regression coefficient was statistically significant.

‡

Netiquette (short for “network etiquette” or “Internet etiquette”) is a set of social conventions and code of conduct that facilitate interaction in virtual environments as mailing lists, blogs, and forums.

References

, Agarwal

. Through a glass darkly: information technology design, identity verification, and knowledge contribution in online communities. Information Systems Research, 2007; 18:42–67.

Bagozzi

, Dholakia

. Intentional social action in virtual communities. Journal of Interactive Marketing, 2002; 16:2–21.

Wang

, Cheng

. Knowledge contribution in the online virtual community: capability and motivation. Lecture Notes in Computer Science, 2006; 4092:442–453.

Wang

, Noe

. Knowledge sharing: a review and directions for future research. Human Resource Management Review, 2010; 20:115–131.

Wasko

, Faraj

. Why should I share? Examining social capital and knowledge contribution in electronic networks of practice. MIS Quarterly, 2005; 29:35–57.

Bafoutsou

, Mentzas

. Review and functional classification of collaborative systems. International Journal of Information Management, 2002; 22:281–305.

Tsai

EHT

, Bagozzi

. Contribution behavior in virtual communities: cognitive, emotional and social influences. MIS Quarterly (in press).

Bagozzi

. On the concept of intentional social action in consumer behavior. Journal of Consumer Research Policy, 2000; 27:388–396.

Bratman

. 1999. Faces of intention. Cambridge, UK: Cambridge University Press.

10.

Gilbert

. 1989. On social facts. New York: Routledge.

11.

Searle

. 1990. Collective intentions and actions. Cohen

, Morgan

, Pollack

. Intentions in communication. Cambridge, MA: MIT Press, 401–415.

12.

Tuomela

. We will do it, an analysis of group-intentions. Philosophy and Phenomenological Research, 1991; 51:249–277.

13.

Tuomela

. 1995. The importance of us: a philosophical study of basic social notions. Stanford: Stanford University Press.

14.

Bratman

. Shared intention. Ethics, 1993; 104:97–113.

15.

Tuomela

. 2007. The philosophy of sociality. Oxford, UK: Oxford University Press.

16.

Bagozzi

, Dholakia

, Pearo

. Antecedents and consequences of online social interactions. Media Psychology, 2007; 9:77–114.

17.

Bagozzi

, Dholakia

. Antecedents and purchase consequences of customer participation in small group brand communities. International Journal of Research in Marketing, 2006; 23:45–61.

18.

Bagozzi

, Dholakia

, Mookerjee

. Individual and group bases of social influence in online environments. Media Psychology, 2006; 8:95–126.

19.

Ajzen

. The theory of planned behavior. Organizational Behavior and Human Decision Processes, 1991; 50:179–211.

20.

Armitage

, Conner

. Efficacy of the theory of planned behaviour: a meta-analytic review. British Journal of Social Psychology, 2001; 40:471–499.

21.

Eagly

, Chaiken

. 1993. The psychology of attitudes. Fort Worth, TX: Harcourt Brace Jovanovich.

22.

Nielsen

. 2006, 20–05–2011).Participation inequality: encouraging more users to contribute. www.useit.com/alertbox/participation_inequality.html. 2012 July 10.

23.

Kramer

, Brewer

, Hanna

. 1996. Collective trust and collective action. Kramer

, Tyler

. Trust in organizations: frontiers of theory and research. Thousand Oaks, CA: Sage, 357–389.

24.

Kollock

. 1998. Transforming social dilemmas: group identity and cooperation. Denielson

. Modeling Rational and Moral Agents. Oxford, UK: Oxford University Press, 185–209.

25.

Kuhlman

, Brown

, Teta

. 1992. Cooperation, individualism, and competition as moderators of verbal and social orientation interpretations of cooperation and defection in social dilemmas. Messick

, Wilke

, Liebrand

WBG

. Social dilemmas: theoretical issues and research findings. Oxford, England: Pergamon Press, 111–132.

26.

Tuomela

. On the structural aspects of collective action and free-riding. Theory and Decision, 1992; 32:165–202.

27.

Gabbiadini

, Mari

, Volpato

. Internet come strumento di ricerca: linee guida per la creazione di web survey. Psicologia Sociale, 2011; 2:237–258.

28.

Perugini

, Bagozzi

. The role of desires and anticipated emotions in goal-directed behaviours: broadening and deepening the theory of planned behaviour. British Journal of Social Psychology, 2001; 40:79–98.

29.

Capozza

, Brown

, Aharpour

et al. 2006. A comparison of motivational theories of identification. Brown

, Capozza

. Social identities: motivational, emotional, cultural influences. Hove, UK: Psychology Press, 51–72.

30.

Bagozzi

, Lee

. Multiple routes for social influence: the role of compliance, internalization, and social identity. Social Psychology Quarterly, 2002; 65:226–247.

31.

Markòczy

. Utilitarians are not always fair and the fair are not always utilitarian: distinct motives for cooperation. Journal of Applied Social Psychology, 2007; 37:1931–1955.

32.

Jöreskog

, Sörbom

. 2006. LISREL for Windows. Lincolnwood, IL: Scientific Software International, Inc.

33.

Bagozzi

, Heatherton

. A general approach to representing multifaceted personality constructs: application to state self-esteem. Structural Equation Modeling, 1994; 1:35–67.

34.

, Bentler

. Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Structural Equation Modeling, 1999; 1:51–55.

35.

Kline

. 1998. Principles and practice of structural equation modeling. New York: Guilford.

36.

Dholakia

, Bagozzi

. Open source software user communities: a study of participation in Linux user groups. Management Science, 2006; 52:1099–1115.

37.

Bock

, Zmud

, Kim

et al. Behavioral intention formation in knowledge sharing: examining the roles of extrinsic motivators, social-psychological forces, and organizational climate. MIS Quarterly, 2005; 29:87–111.

38.

Dholakia

, Bagozzi

, Pearo

. A social influence model of consumer participation in network and small group based virtual communities. International Journal of Research in Marketing, 2004; 21:241–263.

39.

Sutton

. Predicting and explaining intentions and behavior: how well are we doing? Journal of Applied Social Psychology, 1998; 28:1317–1328.

40.

Wright

, Velicer

, Prochaska

. An analysis of the predictive power of the transtheoretical model of behavior change applied to dietary fat intake. Health Educational Research, 2009; 24:224–236.

41.

Capozza

, Mari

, Falvo

. Intentions of reducing or eliminating one's own alcohol consumption in adolescents. New Review of Social Psychology, 2003; 2:332–342.

42.

Mari

, Capozza

, Falvo

et al. Dieta Mediterranea vs. dieta Tradizionale: la scelta di uno stile alimentare nel contesto italiano. Ricerche di Psicologia, 2007; 3:33–62.

43.

Richetin

, Conner

, Perugini

. Not Doing is not the opposite of Doing: implications for attitudinal models of behavioral prediction. Personality and Social Psychology Bulletin, 2001; 37:40–54.