Disinhibition on the Internet

For the past three years, Dr Niki Harré of the Psychology Department of the University of Auckland has invited Liz Butterfield of ISG to make a presentation on interesting Internet research topics to the Community Psychology class at the University of Auckland.  This year Quentin Atkinson took up the topic of disinhibition and has written a fascinating paper.  We would like to thank Quentin for allowing us to post this paper on NetSafe and also to thank Niki for her on-going encouragement of new research looking at different aspects of the psychological impact of communication technologies.

Disinhibition on the Internet: Implications and Intervention
Quentin Atkinson

Department of Psychology
University of Auckland
June 2002
 
ACKNOWLEDGEMENTS

This research was undertaken as a class assignment for an honours level course in
Community Psychology at the University of Auckland. I would like to thank Liz
Butterfield, Niki Harré and Nathan Gaunt for their guidance and inspirational enthusiasm for the topic.
 
Key issues and recommendations

• There is a considerable body of evidence that Computer Mediated Communication (CMC), like e-mail and Internet Relay Chat, generates disinhibited verbal behaviour both in the form of flaming and excessive self-disclosure.
  
• A growing number of studies suggest that browsing behaviour may also be disinhibited.
  
• Disinhibited behaviour on the Internet can be harmful to the victims of such behaviour, and may also be harmful to those exhibiting disinhibited behaviour.
  
  • Those who exhibit disinhibited behaviour online may become more disinhibited in real life.
    
  • Those who exhibit disinhibited behaviour online may develop negative self-perception as a result of their inappropriate behaviour on the internet.
    
• Disinhibition on the Internet is not merely the result of conditions of anonymity. The influence of online norms and problems associated with the bandwidth of CMC must also be considered.
  
• Because of the influence of norms in online behaviour, promoting Netiquette may be an important means of reducing disinhibited behaviour online.
  
• Educating people about the potentially disinhibiting effects of the Internet and empowering them to act responsibly online is suggested as a means of reducing Internet disinhibition.
  
• More research is needed into the prevalence and implications of Internet disinhibition.  and also the efficacy of possible interventions.
  

Disinhibition on the Internet - a potential problem

The internet is performing an increasingly diverse range of functions in our lives. As we adapt it to our lifestyle and our lifestyle to it, as computers make their way out of the office and into the living-room, as people, young and old, begin even to carry it around with them in their pockets, issues surrounding the internet's impact on our society and psyche become progressively more important.  Consequently, there is an increase in the need for organisations that seek to monitor this impact and intervene where necessary.  NetSafe, a New Zealand based Internet safety organization, is such an entity with interests as diverse as law-enforcement on the Internet and child safety at school.  NetSafe is well resourced to tackle most internet safety issues although focuses primarily on child safety online, both in terms of the risk of exposure to inappropriate material and the risk of sexual predation. In talking with Liz Butterfield, from NetSafe, one of the areas in which the organization lacked expertise was disinhibition on the Internet.

There is an ever-increasing body of evidence in the literature that the Internet, specifically in the context of chat-rooms and e-mails, causes 'disinhibited behaviour' (Joinson, 1998).  Such behaviour may constitute a real threat to both child and adult safety and enjoyment online.  Thus the prevalence of disinhibition on the Internet, what causes it and what we can do about it, are questions that need to be addressed.
 
Objective

The objective of this report is twofold: - 
1) First, a comprehensive review of the literature on Internet disinhibition is sought. Current theoretical approaches to disinhibited behaviour, both on the Internet and in general, will be discussed as well as empirical studies looking at the prevalence and impact of disinhibition on the net.  The goal here is to gain insight into the risks and nature of disinhibition related problems on the net, with emphasis on the lesser-studied areas - like the effect of antisocial disinhibited behaviour on those exhibiting the behaviour. It is hoped that this information can be used by NetSafe to assess the importance of Internet Disinhibition to the work that they are doing.
2) The second objective is to present, in the light of this information, a possible intervention method and research program that could be employed by NetSafe or any other organization interested in making the Internet a safer place for users.  The focus will be on helping individuals exhibiting disinhibited or antisocial behaviour to make informed decisions about the way they act online. 

Literature Review - Disinhibition on the Internet

What is Disinhibition on the Internet?

The phenomena was first noticed during the mid eighties by psychologists studying verbal behaviour in e-mailing (Sproull & Kiesler, 1986), bulletin boards and Multi-User Dimensions (MUDs) - collectively known as Computer Mediated Communication or CMC (see Siegel, Dubrovsky, Kiesler & McGuire (1986) for a particularly good example of these early studies).

Under normal, face-to-face social conditions, conversation is governed by a myriad of quite stringent social norms and rules - people are, for the most part, kind and considerate in their interactions with others, heated conflict is unusual and hateful verbal abuse is very rare indeed.  However, CMC users appear to follow different rules or, as some have argued, no rules at all.  An issue that would have caused only minor disagreement face-to-face may often result in things getting out of hand in CMC as communication degenerated into hostile word-slinging.  Indeed the behaviour is so common that it has been given a name - 'flaming'.

The JennyMUSH discussion group, a self-help web site for victims of abuse, is used by Reid (1998) to provide a particularly vivid example of Internet disinhibition. On the JennyMUSH site, a user took advantage of the system to transform him-/her- self into a 'virtual manifestation of every other user's fears'. Changing their gender to male and name to 'Daddy' the user was able to send messages describing 'virtual assaults in graphic and violent terms' to all others logged on to the site at the time.  Some of the victims of this attack logged off but others moved to the same virtual location as the attacker and pleaded with him to stop.  Eventually a Wizard arrived to find the remaining users being 'obscenely taunted' by the attacker.  Taking matters into his own hands the wizard took away the attacker's ability to communicate, changed his name to 'Vermin' and in the description field that accompanies each user on-line wrote, "This is the lowest scum, the most pathetic dismal object which a human being can become." Despite the awful nature of what had happened, the worst was yet to come. With their attacker effectively bound and gagged the remaining JennyMUSH users took "dreadful virtual revenge."  "They described all the most violent punishments they would like to enact on this and all other attackers, emoting all the hatred and rage that JennyMUSH had been established to help people come to terms with" (Reid, 1998).  The shocking behaviour of not only the instigator but the victims of this hateful display has, been attributed to the disinhibiting effect of CMC.

As well as the cases of anti-social disinhibited behaviour, psychologists also observed a more pro-social effect. People appeared more open when using CMC. Intimate friendships were formed quickly and easily over what was thought to be a relatively restricted means of communication. Indeed, people were friendly to a fault. Where someone might talk self-consciously with a stranger face-to-face, being protective and careful about giving away personal details, meetings with strangers on CMC were less awkward and resulted in the divulgence of an alarming amount of sensitive, personal and potentially dangerous information.

There is, today, no shortage of evidence that CMC is associated with apparently disinhibited behaviour and there is a growing body of evidence that even browsing behaviour is subject to disinhibition.  In short, it has been found that on the Internet people say what they would not normally say, do what they would not normally do, and perhaps even go, at the click of a button, where they would not normally go. The definition of exactly what constitutes disinhibited behaviour is somewhat contentious however.  In the literature, particularly in that which deals with CMC specifically, disinhibition is often simply defined as flaming.  Though the study of flaming behaviour in CMC is important in it's own right, defining disinhibition as flaming ignores the potentially more interesting and informative issues surrounding that broader group of behaviours characterised by disinhibition.  If flaming, excessive disclosure and disinhibited browsing behaviour are associated then it would be of benefit to psychologists to study these collectively.  Similar problems occur with definitions that incorporate the idea of 'anonymity' a priori into the concept of disinhibition - there is evidence that not all disinhibited behaviour is under strictly anonymous conditions (see section on psychological explanations, below).

According to Joinson (1998), "Disinhibition on the Internet is any behaviour that is characterised by an apparent reduction in concerns for self-presentation and judgement of others".  Joinson's definition is broad enough to encompass all of the types of disinhibited behaviour discussed so far, but it also gives a clear idea of the sort of behaviour we are looking for.  Furthermore, his deliberate use of the word 'apparent' makes this definition neutral with respect to theory. A theory neutral definition is preferable when we discuss possible causes of disinhibition on the Internet later. For these reasons, unless explicitly stated, Joinson's definition of disinhibition will be used henceforth.

When and where does disinhibition on the Internet occur?

Are the above examples of behaviour, like JennyMUSH, isolated cases or evidence of a universal phenomenon? Do such examples demonstrate the effect of the Internet on everyday people or are they more indicative of the type of people using the Internet?  These are points that need to be considered, however, as we will see, the evidence suggests that the internet does, indeed, cause disinhibited behaviour of the type described above in otherwise 'normal' people.

The prevalence of flaming in CMC is perhaps the best-documented example of disinhibition on the Internet. According to Coles (1997), "over half of Americans say they have received office flame mail.'' Selfe and Meyer (1991) maintain that 'heated, emotional, sometimes anonymous, venting' is a common, if not universal, feature of computer based conferences'.  Though they offer support to the hypothesis that the Internet (in this case CMC) causes disinhibition, neither of these studies give direct evidence that such behaviour does not occur in face-to-face communication in a similar context.  Dyer, Green, Pitts & Millwards (1995) do better by comparing CMC to face-to-face communication in general and claim that flaming occurs four times as often in CMC than in face-to-face interactions.  Unfortunately, no attempt is made to control for subject matter - perhaps topics discussed on CMC are just more flame-worthy!  

The best evidence for the disinhibiting effect of CMC is from the ongoing research program of Kiesler and colleagues. One study (Siegal, Dubrovsky, Kiesler and McGuire, 1986) compared verbal behaviour between groups of three people across four different conditions; face-to-face, anonymous CMC, non-anonymous CMC and e-mail.  They found that subjects in the conditions involving computer communication exhibited significantly more disinhibited behaviour (defined as hostile comments) than those in the face-to-face condition (34 episodes of uninhibited behaviour in CMC, and none in FTF conditions!).  Interestingly, the increase was significant regardless of whether or not the subjects were in the anonymous condition indicating that anonymity alone cannot account for the disinhibiting effect of CMC.  Siegal et al. (1986) also found that CMC verbal behaviour was more disinhibited than e-mail verbal behaviour.

The evidence appears incontestable, however Siegal and Kiesler's studies, which repeatedly find high levels of disinhibited behaviour, have been criticised on a number of levels. Lea, O'Shea, Fung, and Spears (1992) point out that subjects are usually students engaged in hypothetical role-play, interacting anonymously with strangers.  Hiltz et al. (1989) conducted a study similar to Siegal et al. using groups of managers, who were completing role-playing 'choice dilemmas' realistic for their organization, and found that about half the CMC groups had no incidence of disinhibited (flaming) behaviour at all.

Further, Baym (1995) states "that it is a mistake to see patterns in CMC as the direct effects of the medium itself.'' It may not be possible to specify in advance of the actual interaction what specific factors in what combination will be relevant to CMC outcomes in a particular group, or what the actual impact might be. Actually, Baym is probably being a bit too harsh here.  She is correct that the data of Kiesler and his associates can only identify trends in CMC behaviour in different contexts and not predict how particular interactions will unfold, but the conclusions are expressed as trends and not governing laws.  For this reason, the results of Baym, Hiltz et al. and Lea, et al. do not invalidate the findings of Kiesler and company.  What they do indicate though, is that caution is necessary when generalizing results across social and contextual boundaries.

Disinhibited behaviour other than flaming is harder to find in the literature, however the effects are just as substantial. Sproull and Kiesler (1991) looked at office communication and report that people are more frank and open in e-mails.  In addition, medical patients have been found to report more symptoms and undesirable behaviours in on-line interviews than face-to-face (Joinson, 1998) whilst clients at an STD clinic report more sexual partners, more previous visits and more symptoms to a computer than to a doctor (Robinson, 1992).

The disinhibiting power of the Internet may extend to WWW browsing also.  Joinson and Harris (1995) recorded the number of site hits a football team web site received after wins and losses. The data showed an expected increase in site hits post a win (dubbed ''basking in reflected glory'') but no significant drop in hits following a loss.  Joinson and Harris claim that, normally, we would expect information seeking for a team to drop following a loss.  No such drop was recorded on the Internet, hence the pair conclude that ''WWW surfing may not be subject to the same self protection needs as in real life (where people generally avoid threatening information).'' Joinson and Harris further suggest that this could be the result of disinhibition. Their results are far from conclusive, people could be visiting the site for other reasons (perhaps to bask in past glories) or, as Joinson points out, the baseline hit rate might represent a level of 'random' hits on the site and not the activity of team devotees. Even so, the results are certainly consistent with a theory of disinhibited Internet browsing behaviour and warrant further investigation.

Finally, the amount of illegal or undesirable material available on the Internet seems to suggest that people feel less inhibited about the exchange of such material electronically.

Implications of Internet Disinhibition

What, if anything, are the implications of disinhibited behaviour on the Internet? Should we be concerned by it, or is it just an interesting but essentially unimportant by-product of the information-super-highway? 

Certainly, disinhibition on the Internet need not always be construed as bad.  It can be enormously beneficial - the lonely can make new friends, the meek can speak their mind, the shy can explore new identities, and the embarrassed or ashamed can seek counsel. One American survey showed that teenagers felt the Internet freed them up to be their true selves (Lenhart et al., 2001).  Korean web-site designers capitalised on the liberating quality of CMC by opening up a 'swear room' where people could flame each other to let off steam (Yoon, 2001).  Health organisations, as well, are beginning to realize the benefits of people's increased propensity to seek or divulge potentially embarrassing or threatening medical information online. Unfortunately however, the effects of disinhibited behaviour on the Internet are not always desirable.  The following is a synopsis of some of the potential negative implications of disinhibition on the Internet: -

Flaming - The characteristics of flaming behaviour have already been discussed and there can be no doubt that at least some of the time this behaviour has a negative impact on child and adult victims alike.  The old adage 'Sticks and stones will break my bones but names will never hurt me' could never have been more inappropriate than on the Internet.

Predation - The results of a NZ on-line survey of adolescent girls on the net suggest that the sort of behaviour the Internet can generate is alarmingly dangerous. 'Girls on the Net', as the survey was called, found that: 35.5% of those surveyed had given out personal information over the net; 26.5% had sent a photo of themselves to someone on the Net; 33.5% had met someone face to face that they had met online; 32% went to that meeting alone; 22.5% had felt unsafe or threatened while using the Internet (most commonly sexual threats).  These are disturbing results, and even more disturbing is that they are relatively conservative.  Whether the direct result of a disinhibitory effect or not, such figures are part of growing evidence that predation on the internet presents a real risk to youth.  Most predators operate via CMC, grooming unsuspecting children into a face-to-face meeting.  In this context, disinhibition may be something of a double-edged sword.  Not only could it increase the risk of children divulging information and agreeing to more than they otherwise would, but it may also impact on the behaviour of potential predators.  It is not hard to visualize a deviant who enjoys just talking to children over CMC progressing to 'predator' status by arranging meetings under the disinhibiting influence of CMC.

For better or for worse, the Internet's potential to facilitate deviant behaviour is becoming increasingly well known. In 1997 a Welsh man was accused of numerous acts of child abuse. His defence argued that the Internet acted as a catalyst (in Joinson, 1998).

Illegal or Undesirable material - There is no doubt that the internet enables users to access illegal or undesirable material that they would not otherwise be able to access, but does it do something more?  If the Internet causes disinhibition, then people will be more likely to actually act on a desire to procure illegal or undesirable material online.  Just as disinhibition on the information super-highway can make seeking medical information easy as well as appealing (Joinson, 1998), so too may it 'encourage' those who seek everything from an illegal copy of this weeks hit single to illicit pornography.

Online recruitment - Another risk that has been identified in association with Internet disinhibition is an increased susceptibility to online recruitment by gangs and cults.  As with illegal and undesirable material, the effect of the Internet may be two pronged.  Not only does it increase the likelihood of exposure, but also the chances of acting on any, otherwise latent, desires. 
This may help people to realize their 'true identity' - for instance the Internet has proven enormously helpful to people struggling with their sexual identity.  On the flip side though, an increased likelihood of subscribing to the KKK, pro-anorexia sites, religious cults or paedophile rings can only be negative. Gaunt (pers.  omm..) points out a salient example in the Heaven's Gate cult, a religious group that recruited members on the Internet, all of who committed suicide in 1997.  Though the effects of disinhibition in this area are potentially dramatic, research is sadly lacking.

Impact on the perpetrator - An important area of study that has received little attention to date is the psychological effect of disinhibited behaviour on the individual exhibiting the behaviour (Gaunt, pers. comm.).  Justifiably, much has been made of the devastating effects of disinhibited behaviour (esp. flaming) on its victims, but relatively little research looks at the effects on the perpetrator.  In the real world most of us have done or said something that we later regretted, so perhaps, that the phenomenon occurs online should not be of concern. Unlike in real-life though, on the Internet everything said and done is recorded (potentially on thousands of computers worldwide) and can be used against people verbatim, indefinitely.  In CMC this is made all the more distasteful by the fact that most, if not all communication is text based.  What you say and what you have said ARE who you are online. Because of this people find it very difficult to take back what they have said, preferring instead to vehemently defend a statement, even if it is not a true representation of their thoughts and feelings (Reid, 1998).  When the very nature of the medium is disinhibiting as well, this makes for a fragile cyber-community indeed. 

Lin & Chin-Chung (2000) found that those people who have become dependent on the Internet scored higher on overall sensation seeking and disinhibition than non-dependents.  This means that the very people who rely on the internet for social support and friendship are the ones who are likely to have it all come crashing down around them, and are unlikely to have a 'real-world' social support network to kick in when things get tough.  As the JennyMUSH example illustrated, the retribution sought by victims of flaming can be as tough as, if not tougher than, the crime itself. 'Victims' also have the power to completely ostracise the perpetrator in their MUD and potentially exclude them from other MUD's by passing on user details.  Fortunately or unfortunately, depending on your point of view, the same anonymity that may encourage disinhibited behaviour also serves to protect the perpetrator from real-life persecution and allows him the option of creating a new online character to replace the banished one.

Whilst an individual's cyber-characters are theoretically replaceable, the same cannot be said for their real-life persona, which may be permanently damaged by such incidents.  Not surprisingly, ostracism and rejection by one's cyber-peers has been found to negatively affect self-esteem and physiological and psychological health (Williams, Cheung and Choi, 2000). Putting aside the impact of rejection by one's supposed friends for a single moment of indiscretion, one must ask what is the impact of the behaviour itself on the individual's self-perception.  From CMC flaming to unsolicited hate mail, from downloading illegal software to frequenting the seedy cyber-underworld, what is the impact of disinhibited behaviour on the disinhibited?  As Joinson (1998) points out, according to Bem's (1972) self-perception theory, people conceptualise themselves and their attitudes by observing their own behaviours.  If an individual is repeatedly acting in a disinhibited way on the net, they may infer that this sort of behaviour is who they are.  This may be a traumatic inference to make if the behaviour is particularly distasteful.  What is more, there is the potential that, having formed a negative self-perception, the behaviour will diffuse into daily life (Gaunt, pers.com.). The Lin & Chin Chung study found that people who have become dependent on the Internet scored higher on overall sensation seeking and disinhibition than non-dependents.  This shows at least a correlation between dependent Internet use and disinhibited behaviour in other aspects of one's life.  More research is needed into whether there is a causative relationship between Internet use and disinhibition in other areas but the results certainly suggest a possibility.

Possible Psychological Explanations of Internet Disinhibition

Given that disinhibition on the Internet is a potential problem, what can we do about it?  The first step towards preventing dangerous or undesirable disinhibited behaviour is to attempt to explain it. Armed with a powerful explanation of what might be happening we can construct informative, theory driven research and develop interventions where necessary.

Most theories focus on the role of anonymity in fostering disinhibition and indeed the two ideas have long been linked in psychology.  In 1895, Le Bon incorporated both into his 'Study of the popular mind' (Joinson, 1998) and, more recently, we can see the link in applications of Social Identity Theory and Deindividuation Theory (Prentice-Dunn and Rogers, 1982) in explaining disinhibited crowd behaviour. Although largely focused on explaining group level phenomena, the ideas behind the two theories are still applicable to disinhibition on the Internet.  Both claim disinhibited behaviour results from a reduction in accountability (e.g. anonymity reduces concern for the public consequences of the behaviour) and a shift in attention from the private or personal level to the public or social level (reducing the salience of internal standards and self-regulation).  Though these theories are appealing because of the substantial body of research behind them, there is evidence that they are inadequate as an explanation of disinhibition on the Internet. 

Firstly, they predict that behaviour, like flaming, by anonymous users will not be governed by any norms. As is claimed by Postmed, Spears and Lea (2000) proceeding their investigation into group norms in CMC, this is clearly not the case.

''Results show that norms prescribing a particular use of technology are socially constructed over time at the level of locally defined groups and also show that the influence of these norms is limited to the boundaries of the group. It is concluded that the process of social construction is restrained by social identities that become salient over the course of interaction via CMC. These findings complement experimental evidence that stresses the importance of normative influence in CMC.''

Secondly, these theories imply that anonymity is of prime importance in the development of disinhibited behaviour. However, Siegal's study mentioned earlier, and others like it, have shown that the effect of anonymity is small when compared to the effect of the communication medium.  

Another group of theories that have been applied to explaining disinhibition on the Internet are those that focus on the nature of the communication medium itself.  Not surprisingly, there is evidence that understanding of each other (Straus & McGrath, 1994) and of group decisions (Adrianson & Hjelmquist, 1991) is poorer in CMC than face-to-face. The narrow bandwidth of CMC and e-mail is thought to lead to depersonalisation through reduced 'social presence' and to misunderstandings, disagreements and frustration because of reduced social cues and the limited amount of information that can be conveyed in any given time. 

Band-width arguments can explain hostile behaviour relatively easily, however, they have more difficulty explaining things like excessive self-disclosure and intimate Internet based friendships.  One argument is that precisely because of the limited amount of information available, users reveal personal information in an effort to compensate.  It is, as yet, a moot point as to whether the sort of disclosure seen in CMC is beyond mere compensation.

Emergent Norm Theory (henceforth ENT; Turner, 1974) offers another possible explanation of disinhibition on the Internet. This is the idea that in unfamiliar or socially ambiguous situations (like in a chat room) we tend to look at the behaviour of others for guidance as to what is appropriate and we tend to only attend to behaviour that stands out in some way (like someone being aggressive or revealing a lot about themselves).  In this way, quite anti-social behaviour can become the norm.  Unfortunately, the potential for ENT to explain disinhibited behaviour on the Internet has not been investigated.  Admittedly, ENT can say little about behaviour in situations that are already highly normative and many chat rooms might fall into this category.  Nonetheless, ENT could have quite a lot to say about what sort of norms we could expect to be created in new chat rooms with new users. 

In the past, the puzzle has been to explain how apparently anti-social behaviour (e.g. flaming) can be so common when all of the evidence suggests that Internet communication is highly governed by social norms (Postmed, Spears and Lea, 2000). ENT, in combination with some of the other theories, offers a solution:

In a chat room full of 'strangers', under conditions of deindividuation, users will initially find themselves in a socially ambiguous situation.  Those who are regular CMC users might bring with them a pre-existing set of norms that may or may not include an expectation of disinhibited behaviour.  Regardless, in most cases nobody will know who does and who does not know how to behave in this chat room and so, as ENT states, all will be primed to look for any 'emergent norms'.  Given the potentially frustrating nature of CMC communication and the necessity to convey personal information to make up for a lack of contextual and social presence cues (attributable to bandwidth problems), the sorts of behaviour that will stand out in CMC will be the sorts of behaviour we label as disinhibited. 

Social norms theory or ENT might also be able to explain disinhibited browsing behaviour.  Whatever the illness, whatever the penchant, no doubt one will find thousands of web-sites on just what one is looking for.  Not only does this give one access to the desired material but also it implies that the material is normal.  In some cases, like in the acquisition of medical information, the knowledge that you are 'not the only one' may be enormously beneficial, but it may also allow someone to justify to themselves browsing behaviour that they would not previously have deemed appropriate. 

Admittedly, if browsing behaviour or CMC verbal behaviour is affected in the way described above, calling the behaviour 'disinhibited' is misleading in that it is still under the control of social norms - emergent or otherwise.  However, as Joinson stresses in his definition of disinhibition, the 'reduction in concerns for self-presentation and judgement of others' need only be apparent.  That the term 'disinhibition' may be misleading should not detract from theories that seek to explain the phenomena.

Research is needed to establish exactly which of the above processes may operate and when.

Intervention, research and other suggestions

Without an empirically supported explanation of what is going on when people become disinhibited it is difficult to design any sort of intervention so perhaps the most important step forward at the moment is to conduct more research - research actually comparing the efficacy of potential methods of reducing disinhibition is currently non-existent. Nonetheless, we can apply research in other areas to evaluate some of the possibilities.  Here we will look at ways to reduce the impact of disinhibited behaviour on the disinhibited by elucidating two of the most promising classes of intervention; one preventative, the other something of a treatment. Other possible interventions, like decreasing anonymity on the Internet, may well be effective (Suler, 1997) but have ethical and practical issues associated with them that make them problematic.

Promoting Netiquette and changing browsing behaviour - If disinhibition on the Internet (particularly CMC) is governed by norms, then one obvious way to reduce the incidence of disinhibited behaviour and thereby reduce the negative consequences of such behaviour is to attempt to change those norms.  As Berson (2001) points out, ''young users often lack cultural sensitivity that can foster collaboration in a global community. Young people are especially prone to misperceive the perspectives and opinions of others and to refrain from respectful interactions.''  Netiquette, a set of rules for Internet etiquette, has the potential to decrease anti-social disinhibited behaviour not only by explicit prohibition but also by establishing more appropriate norms in online communities. 

Although not such a problem in schools (where chat rooms are not generally used), disinhibited behaviour in online chat rooms at home could have negative implications.  Between 70% and 80% of teenagers are apparently using some sort of Internet chat or instant messaging service. Given these figures and given the risks and implications of disinhibited behaviour in CMC, it is important that information to children (and adults) on how to behave online includes advice on chat room etiquette. Internet safety groups, including NetSafe, already encourage netiquette as part of a contract of Internet use for children.  Part of this 'Netiquette' package needs to mention communication problems specific to CMC and how to deal with them.  NetSafe address the issue of excessive disclosure in this manner - 'I will never give out my real name, address and telephone number to someone on the Internet.'; 'I will never agree to get together with someone I meet on the Internet without checking with my parents first.'; 'If I receive a message which makes me feel uncomfortable, I will tell my parents.'; and perhaps something pertaining to flaming behaviour should also be included. For example, Ciskowski and Benedikt (1995) in a guide on MUD etiquette state, "You should be aware of mistakes that can make what you have to say sound, even if by accident, obnoxious and mean-spirited." This sort of information may help avoid anti-social behaviour caused by misunderstandings and the 'bandwidth' problems mentioned earlier.  Something explicitly asking for careful avoidance of abusive language and for a general empathy towards those with whom one is chatting would also be helpful - e.g. "You should always consider the feelings of those who you are e-mailing/chatting with online.  The fact that you are in a chat room is no excuse for verbally abusing someone."

A greater awareness of 'chat room rules', and when they are broken, should reduce the amount of anti-social disinhibited behaviour directly (through prohibition), but also by creating norms for appropriate behaviour in chat rooms. It may also reduce the impact of the behaviour on victims. Further, Berson suggests that educating children (and adults alike) about how to behave online could be applied to reduce the impact of inappropriate, disinhibited browsing. 'When the rules for appropriate conduct are combined with skills in information literacy, young people are more capable to critically evaluate information found on the Internet.'

As well as promoting netiquette, there are other ways to censure browsing behaviour.  Parents checking up on kids or surfing with the child, having the computer out in the open and the use of effective filtering software can all help to reduce inappropriate browsing (Lenhart, 2001) and, derivatively, anti-social, disinhibited behaviour.  NetSafe already addresses all of these issues in its information pamphlets.

Increasing Awareness -  Another strategy to reduce the negative psychological impact of disinhibited behaviour would be to target those who feel they might be disinhibited and offer information about the potentially disinhibiting effects of the Internet.  A heightened awareness of internet disinhibition and its negative effects might cause people to check their behaviour when they feel it is becoming disinhibited and might also reduce the impact of the behaviour when it does occur. 

Although such intervention has no direct empirical support, it seems intuitively plausible and is similar in many respects to other community intervention programs that seek to change behaviour and attitudes through education rather than manipulation.  Incorporating information on disinhibition into a web page would be a good starting point, giving advice for people who think they might be disinhibited as well as information for victims and those who are just curious.  This is something that the people at NetSafe have expressed an interest in and could be implemented relatively easily.

Appendix 1 is a proposed web page, designed on the basis of the current literature to inform users about the risks of disinhibition on the Internet and ultimately to reduce undesirable disinhibition. One of the advantages of this type of intervention is that it empowers the individual to act on his/her newfound knowledge and change his/her behaviour. This is especially helpful in the case of disinhibition since disinhibited behaviour can sometimes be a good thing.  We do not want to eliminate disinhibition entirely, but rather, inform and empower users to decide for themselves what is and what is not appropriate behaviour online. The proposed intervention does this by offering a relatively brief summary of what disinhibition is and suggesting a simple way of dealing with it - to Stop and think. 

Whether the page came up automatically when users logged on to their ISP or whilst people browsed through the NetSafe site or whether it could be appended to a pre-existing page are issues that would need to be considered.

Future Research - A problem with the proposed intervention methods is that they are, as yet, completely untested.  In the short term, it would be beneficial to test the web page, and any netiquette advice, on a target group - probably teenagers who use the Internet regularly. Some sort of Internet based survey would be ideal here since Internet users are the targets of the intervention. Conceivably, a simple question at the end of the disinhibition web page could ask for visitors' comments on what they thought of the site, what they got out of it and how they thought it could be improved. Site hits are also fairly easily measured and could be used in conjunction with the comments to determine the best way to deliver the page. Having the page crop up automatically might get the most hits but may annoy people or cause them to dismiss it as irrelevant. Conversely, if visiting the page requires one to click on a link, those who visit the site voluntarily may view it more positively but site hits will undoubtedly be fewer. Attaching the page to the NetSafe site may prove to be a feasible option. Importantly, not all visitors to the site are concerned parents or school principals. Children, teenagers and even paedophiles (looking for information on the risks of being caught), all use the site and could all be affected by disinhibition on the Internet.

More broadly, further research is required into the prevalence and implications of disinhibited behaviour online. There is a special need for research into how those who exhibit the behaviour are affected in the wider context of their lives.

Conclusions and recommendations

Disinhibited behaviour is well documented in CMC and may also play a part in peoples' general browsing behaviour online.  The negative impact of disinhibited verbal behaviour on the victims of this behaviour is well evidenced in the literature and perhaps for this reason alone we should endeavour to stop any disinhibited behaviour in CMC that we can.

What is more contentious, and will remain so until further research is carried out, is the extent to which disinhibited behaviour negatively affects those who exhibit the behaviour. Despite the limited research in the area there is evidence that the negative implications of disinhibition on the Internet may extend beyond its effect on the victims of verbal abuse.  Disinhibited verbal behaviour in CMC, as well as disinhibited browsing, could have a negative, perhaps profound, impact on the lives of some Internet users.  More research is needed into whether or not this is the case and, if so, what we can do about it. 

In the mean time, the interventions proposed in this study proffer a method of decreasing disinhibited behaviour online.  The disinhibition web page offers an easy to implement and potentially empowering method of educating users about Internet disinhibition, whilst some minor additions to the netiquette that NetSafe already promotes may help to change some of the undesirable norms that chat rooms seem to generate. 
 
References

Adrianson, L., & Hjelmquist, E. (1991). Group processes in face-to-face and computer-mediated communication. Behaviour & Information Technology, 10, 281-296.

Baym, N. K. (1995) The Emergence of Community in Computer-Mediated Communications. In Cybersociety: Computer-Mediated Communication and Community, Pg. 138-164, Sage Publishing, London.

Bem, D. J. (1972). Self-perception theory. In L. Berkowitz(ed.), Advances in Experimental Social Psychology, (vol. 6, pg 1-62). New York. Academic Press.

Berson, M. (2002) Lessons Learned about Schools and Their Responsibility to Foster Safety Online. Presented at conference NetSafe: Society, Safety and the Internet

Ciskowski, D. & Scheier, M. F. (1995) In Gackenback, J. (1998) Psychology and the Internet: Intrapersonal, Interpersonal, and Transpersonal Implications. California. Academic Press. (Harcourt, Brace & Company).

Coles, J. (1997) Log off! Clear my Screen. The Gaudian, 19 July 1997, p. 5 In Gackenback, J. (1998) Psychology and the Internet: Intrapersonal, Interpersonal, and Transpersonal Implications. California. Academic Press. (Harcourt, Brace & Company).

Dyer, Green, Pitts & Millward (1995). What's the flaming problem? CMC: deindividuating or disinhibiting? In M.A.R. Kirby, A.J. Dix, and J.E. Finlay (Eds.), People and Computers X.  Cambridge: Cambridge University Press.

Gackenbach, J. (1998). Psychology and the Internet: Intrapersonal, Interpersonal, and Transpersonal Implications. California. Academic Press (Harcourt, Brace & Company).

Hiltz, S. R., Turoff, M., & Johnson, K. (1989). Experiments in group decision making, 3: Disinhibition, deindividuation, and group process in pen name and
real name computer conferences. Decision Support Systems, 5, 217-232.

Hrachovec, H. (2001) New Kids on the Net in Community informatics: shaping computer-mediated social relations / edited by Leigh Keeble and Brian D. Loader. New York : Routledge, 2002

Joinson, A (1998) Causes and Implications of Disinhibited Behaviour on the Net in Gackenbach, J. (1998). Psychology and the Internet: Intrapersonal, Interpersonal, and Transpersonal Implications. California. Academic Press (Harcourt, Brace & Company).

Joinson, A. N., & Harris, P.R. (1995). Self-enhancement and self-perception on the Internet: A study of football fans on the WWW. Presented at BPS London Conference and in Joinson, A (1998). Causes and Implications of Disinhibited Behaviour on the Net

Lea, M., O'Shea, T., Fung, P., & Spears, R. (1992). 'Flaming' in computer-mediated communication. In M. Lea (Ed), Contexts in computer mediated communication. London: Harvester Wheatsheaf.

Lenhart, A., Rainie, L., & Lewis, O. (2001). Teenage life online: The rise of the instant-message generation and the Internet?s impact on friendships and family relationships. Washington. Pew Internet and American Life Project. Available: http://www.pewinternet.org/

Lin, S. S. J. & Tsai, Chin-Chung (2000). Sensation Seeking and Internet Dependence of Taiwanese High School Adolescents. Availability ? Microfiche.

Postmes, T., Spears, R. & Lea, M. (2000) The Formation of Group Norms in Computer-Mediated Communication. Human Communication Research, 26(3), 341-72.

Prentice-Dunn, S., & Rogers, R. W. (1982) Effects of private and public self-awareness on deindividuation and aggression, Journal of Personality and Social Psychology, 43, 503-513

Reid, E. (1998) The Self and the Internet: Variations on the Illusion of One Self In: Gackenbach, J. (1998). Psychology and the Internet: Intrapersonal, Interpersonal, and Transpersonal Implications. California. Academic Press (Harcourt, Brace & Company).

Robinson, R. & West, R. (1992) A comparison of computer and questionnaire methods of history-taking in a genito-urinary clinic. Psychology and Health, 6, 77-84.

Selfe, C. L. & Meyer, P. R. (1991) Testing claims for online conferences. Written Communication, 8, 163-192. In Gackenback, J. (1998) Psychology and the Internet: Intrapersonal, Interpersonal, and Transpersonal Implications. California. Academic Press. (Harcourt, Brace & Company).

Siegal, Dubrovsky, Kiesler and McGuire (1986) in Joinson, A (1998). Causes and Implications of Disinhibited Behaviour on the Net

Sproull, L., & Kiesler, S. (1986) Reducing social context cues: Electronic mail in organizational communication. Management Science, 32, 1492-1512.

Straus, S. G., & McGrath, J. E. (1994). Does the medium matter? The interaction of task type and technology on group performance and member reactions. Journal of Applied Psychology, 79, 87-97.

Suler, J.R. and Phillips, W. (1998). The Bad Boys of Cyberspace: Deviant Behaviour in Multimedia Chat Communities. CyberPsychology and Behaviour, 1, 275-294.

Turner, R. H. (1974) Collective Behaviour in R.E.L. Farris (ed.), Handbook of modern sociology (pp. 382-425). Chicago, IL: Rand-McNally.

Williams, K.D., Cheung, K.T., & Choi, W. (2000). Cyberostracism: Effects of Being Ignored Over the Internet. Journal of Personality and Social Psychology, 79(5), 748-762.

Yoon, S. (2001) Culture, Computer Literacy, and the Media in Creating Public Attitudes toward CMC in Japan and Korea in Community informatics : shaping computer-mediated social relations / edited by Leigh Keeble and Brian D. Loader. New York : Routledge, 2002