1) Researchers should identify themselves, their affiliation, and provide contact co-ordinates whereby potential participants can verify the legitimacy of the study.  Inquiries from participants', along with remarks and/or observations concerning the study format should be welcomed.  The latter may provide valuable experience in the design of follow-up studies and in the development of further guidelines.

2) Potential participants should be assured of the confidential treatment of personal information.  Research has shown that data privacy is a key concern for Internet users, who value the anonymous nature of the medium (GVU's 7th WWW User Survey).  Ideally, participants should have the opportunity to remain semi-anonymous (with identification occurring via their e-mail address, rather than their name or postal address) whenever possible.
 
3) Participants’ should be made aware that anonymous use of statistical data will occur in the normal course of the study.  Quotes should only be used with consent.

4) Participants’ consent for participation should be obtained.  This task is more difficult on the Internet than in laboratory studies as participants' signatures cannot be readily obtained.  Alternatively, the researcher may caution the subject that the completion of the study instrument signifies their consent to participate, a method which is legally similar to the means used by credit card and computer software companies.  Such companies warn consumers that the use of the credit card (or the breaking of the software-package seal) legally signifies the acceptance of the terms and conditions associated with the use of their product.  However, participants should be reminded that their participation in the study is totally voluntary and that they are free to withdraw at any time if data collection proceeds on multiple occasions.

5) Information regarding the purpose of the study, criteria for participation, research procedure, potential risks, use of data and method of response should be fully and clearly presented.  The use of ambiguous words or sentences should be avoided.  The most widely cited reason for not registering or providing personal information at Web sites is that 'the terms and conditions of how the collected information is going to be used is not clearly specified' (GVU's 7th WWW User Survey).

6) If fully disclosing the nature and aims of the study would invalidate the research, it may be necessary to withhold some information from the participant.  As with traditional research, however, such a step is only justifiable if a) incomplete disclosure is absolutely necessary  b) there are no undisclosed risks to the subject which present a greater than minimal risk, or greater than that encountered in other forms of ethical research, and  c) adequate plans for debriefing participants have been made.

7) Adequate pilot testing is particularly important in Internet research.  Each stage of the study should be well piloted before beginning data collection, particularly the technical methodology.  A friendly, non-pressuring and captivating text should be developed and piloted inside or outside of the Internet.  An appropriate, catchy title or subject line for the advertisement should be selected to seize the attention of potential participants.  Whilst the researcher cannot be expected to predict any major system crashes or failures, it is advisable to check with system administrators about any potential system changes which might occur during the course of the study.

8) Incentives which may encourage subject participation should be introduced.  For example, the promise to disseminate research findings or papers to the group on completion may be viewed as an incentive for participation in a study.

9) An answer reliability-control method should be embedded in the questions.  Many questionnaires in the social sciences contain items which are reversely rated.  The identical rating of oppositional items may indicate that deception on the subject's part is occurring.  Other rating scales have built in lie or conformity scales.  Alternatively, the researcher could utilise programs such as Internet Relay Chat (IRC), which allow users to convene and talk electronically in synchronous time, to follow-up a small percentage of participants in greater detail and check the reliability and validity of their answers under interview conditions.

In addition to using reliability-checking methods, the researcher should compare the internal consistency of the employed measure observed for the Internet sample to that reported for the instrument in traditional research environments.  In doing so, the researcher should take into consideration the comparability of the Internet sample with the sample upon which the original internal consistency value was based, and set the acceptable discrepancy between the two values accordingly.

10) The use of forced choice answers (i.e., closed answer questionnaires or rating scales) over open-ended questions aides ease of transcription, data entry and reliability-checking, and discourages respondents from recounting irrelevant information (Saris, 1991).  However, the Internet can be an ideal sphere for the collection of qualitative, machine-readable data which is ready for analysis without the added labour of transcription.  IRC programmes could again be used in conjunction with methodologies such as discourse analysis, allowing the researcher to conduct real-time interviews in a setting and at a time which is convenient to both interviewer and interviewee.

11) The amount of information sought from the subject should be reasonable.  Posting a study to a newsgroup is similar to publishing it in a newsletter or magazine.  Any questions should be concise, clear and germane to the group in question in an attempt to promote co-operation.  Unduly long questionnaires may be split into separate parts which may be submitted at the user's convenience.

Web sites, however, may be seen as a more private medium, and are usually only accessed by users already interested in the Web page's subject matter.  Given this, it may be more acceptable to administer lengthy questionnaires or tests (i.e. requiring between 10 and 30 minutes of response time) on specifically designed Web sites.  Such considerations give rise to important questions, such as whether homogeneity exists between newsgroup respondents, who come across the posted studies 'accidentally', and Web site respondents, who make a  conscious decision to visit a specific site in order to participate in the study.  The answer to such questions remains to be determined in future studies.

12) Advertisements for studies or e-mail based questionnaires should be posted to newsgroups at a reasonable frequency.  This posting rate should be decided on the basis of the size of the targeted group, which may be estimated by the number of messages posted to the group per day, and the average posted message/answer ratio.  For example, a newsgroup may be considered large if there are more than 100 messages posted every day, medium-sized if the number of postings is between 30-100, and small if the number of messages is below 30.  It should be noted, however, that the size of a given newsgroup is not always proportional to the number of answers which may be expected.  Potential response rate may be better estimated by the number of answers received after the first posting.  From the authors' experience, it is reasonable to assume a 20-30% decline in the answer rate with each posting within a discussion group.

On a Web site, the instrument should be made available for a period of time which is in accord with the sample size required to achieve the statistical power needed for the analyses of the research question.

13) With reference to point 12, however, it should be remembered that researchers should not bombard newsgroups with frequent repeated postings.  Whilst timely, appropriate postings to newsgroups may be acceptable to readers, excessive or inappropriate postings may be met with accusations of ‘spamming’ (flooding the groups with unwanted messages) and reprimands.

14) The location of the Web site, or the Web page address should be well advertised in different media, such as short messages posted to newsgroups, registration with Internet search engines such as 'Lycos', 'Alta Vista' or 'Excite' and through Internet or other relevant magazines and newspapers.  The target Web site may also be 'linked' to frequently visited home pages.  For example, most universities and research centres in Europe, the United States and other developed countries have a home page which may be linked to other related Web pages.  The extent and frequency of any advertisements should be estimated upon the basis of the response rate obtained during pilot projects.

15) If cross-cultural mixing in the targeted subject group is not desired, postings may be aimed towards specific newsgroups and specific geographic locations (i.e. North America).  This may be helpful if the researcher is examining sociological issues where cultural factors could influence the data, although by no means will it ensure that a culturally homogenous group of respondents will be obtained.  Whilst it is reasonable to ask participants that they be fluent in English, or the language in which the study is posted, no selection on the basis of ethnicity, religion, income, occupation or sexual preference should be attempted unless necessary to target a particular sample population (see point 18).

16) Whether the data is gathered via e-mail or submitted via a Web page, a reply in the form of a thank you note should acknowledge the receipt of the completed questionnaire and help avoid duplicate submissions.  A thank you sentence should also be included within the original recruitment message.
 
17) Volunteer participants should be reminded that if they have any doubt regarding the meaning of a question or an instruction, they should feel free to contact the researcher in charge for an explanation.

18) No unnecessary questions should be asked regarding participants' specific ethnic, political, religious, class or sexual orientations unless the research being conducting is directly addressing such issues.  Sensitivity and discretion should be used at all times when requesting such personal and potentially discriminatory information.

19) The solicitation of demographic data should be kept at minimum.  Whilst it may be necessary to identify participants in some way, users tend to be understandably reticent when providing personal information.  Ideally, participants should only be identified by their e-mail address (which is often necessary for follow-up research and dissemination of results) and a pseudonym.  Physical address, telephone numbers or the like should not be requested unless marked as optional.  Studies which require more detailed demographic or personal information might be better conducted in a more traditional research sphere at this time.

20) The researcher should not engage in arguments with newsgroup users if critical 'counter-messages' are posted in relation to the research.  If such messages are unfounded, the researcher's point of view may be politely explained through private channels, although in some cases such messages are best simply ignored.  However, if a counter-message contains an instructive criticism (which may often be the case), the issue raised should be addressed.

21) The researcher should carefully consider the time frame of the planned research, taking into account predicted response rate, (see point 9) and the sample size required for adequate statistical power.  Participants should be made aware of any deadlines for participation and a time lag between the data collection and data analysis periods should be incorporated into the planned time frame, as previous research (Michalak, 1997) has shown that responses may continue to be submitted to research Web pages some time after deadline announcements.

22) In preparing a Web-based study, the researcher should attempt to standardise the research instrument across most available presentation platforms (i.e. browsers such as Netscape Navigator).

23) When advertising a Web page within newsgroups, the researcher should remain aware that whilst all newsgroup members will have access to e-mail, some users will not have access to the WWW.  The study instrument should therefore also be available in a text-based version which may be e-mailed directly to participants.

24) One disadvantage of conducting research, particularly empirical research, in cyberspace exists in that it is impossible to standardise the environment in which the study instrument is administered.  For example, we know that arousal rates vary according to a number of factors, including biochemical influences and natural circadian rhythms.  In traditional research the study instrument may be administered to participants at a given time of day to avoid contamination, or the researcher may attempt to control participants’ drug use before study onset.  Such variables may be more difficult to control for in Internet research.  However, the researcher can ensure that as much relevant information is collected as possible, such as recording the time of day the study instrument was completed.

25) The researcher should remain conscious of the fact that (s)he may be accessing a highly geographically diverse subject pool, and make any necessary accommodations in the study instrument.  For example, when examining seasonal fluctuations in an international subject sample, Southern hemisphere participants will be experiencing their winter when Northern hemisphere participants are experiencing their summer.

26) When using a moderated newsgroup or listservers, it is advisable to contact the moderator (the person responsible for ‘screening’ the messages posted to a newsgroup) with details of the research before posting occurs.  Messages posted 'cold' to groups will often simply be rejected by the group’s moderator.  It also considered courteous for the researcher to subscribe to any listservers being used for research purposes.  When utilising a listserver, participants should be reminded to respond to the individual researcher, and not to the group by mistake, or private information may be mistakenly posted to the group as a whole.

27) It is feasible that members of specific disorder-related newsgroups (e.g. soc.support.depression.seasonal) may be more aware of the nature and characteristics of certain disorders than other users, and may feel the need to manipulate their responses to correspond with any preconceived knowledge regarding the syndrome.

28) Despite the fact that most published questionnaires are available for public use, it may be courteous for the researcher to request permission from those who developed the questionnaire before adapting it for use in cyberspace.

29) If the research has received external funding or sponsorship, the origins of this support should be clearly identified.

30) Researchers should specify, both when conducting research on-line and reporting results, what guidelines, ethical principles, and/or laws have been observed when conceptualising and performing the research.