Copyright Sociological Research Online, 1997

 

Coomber, R. (1997) 'Using the Internet for Survey Research'
Sociological Research Online, vol. 2, no. 2, <http://www.socresonline.org.uk/socresonline/2/2/2.html>

To cite articles published in Sociological Research Online, please reference the above information and include paragraph numbers if necessary

Received: 3/1/97      Accepted: 19/6/97      Published: 30/6/97

Abstract

The Internet and electronic mail increasingly offer the research community opportunities that it did not previously have. Access to information has increased as has access to and discussion with those working in similar areas. One other aspect of 'cyberspace' which presents enormous possibilities to the research community, currently in its infancy, is the use of the Internet to reach individuals as research subjects. In particular, there may be significant research benefits to be gleaned where the group being researched is normally difficult to reach and/or the issues being researched are of a particularly sensitive nature. This paper outlines some recent survey research using the Internet as the interface between researcher and researched. The target group, illicit 'drug dealers', are difficult to access under normal conditions and contacting a spread of such individuals across international borders was previously prohibitive. A discussion of sampling issues is undertaken which concludes that the Internet can be a valuable source of indicative as opposed to easily generalizable data. A practical guide to undertaking research via the Internet is also included.

Keywords:
Drugs; Internet; Internet Research; Internet Sampling; Newsgroups Research; Survey Research; World Wide Web

Introduction

1.1
The existence of the Internet and the World Wide Web (WWW) clearly provides new horizons for the researcher. A potentially vast population of all kinds of individuals and groups may be more easily reached than ever before, across geographical borders, and even continents. This is particularly true in relation to comparative social survey research. However, whilst the Internet and WWW does offer new and exciting prospects for sociological research, in many respects the methodological issues which it raises are by and large not new. The key issue that any survey research conducted via the Internet will have to contend with, as with non-Internet based surveys, is that of sampling bias. At present this is more acute than it is likely to be in the future because contact can only be made with those who can and do use the Internet and the WWW, and all that implies in terms of background, education, gender and resources. As use of this technology becomes more general, as is already happening, and methods for collection improve, concerns around sampling bias will come to resemble more closely those which regularly affect conventional surveys. The relative exclusivity of current Internet use needs to be considered seriously but it does not preclude attempts to do useful and informative sociological research, particularly on population sub-groups. Depending, for example, on who, and what, is being researched, the issue of 'representativeness' and sample bias will be of greater or lesser importance in comparison to the indicative data that has been collected. Not insignificantly, at other times, users of advanced communication technology may be exactly the individuals that the research is aiming to reach. Other sampling concerns, such as understanding the differences in those who respond to surveys and those who do not, are similar for conventional surveys which are reliant upon respondents' replies to an invitation to contribute. Again, depending on what it is that is being researched, this will matter to a greater or lesser extent. Other problems relating to doing research via the Internet and the WWW, at this stage of its development and in relation to the sociological research community, revolve around how actually to go about doing it and what issues need to be considered when undertaking such research.

1.2
This article will briefly outline some recent research undertaken via the Internet and the WWW into the practices of drug dealers and their drug adulteration/dilution practices (the 'cutting' of drugs with other substances to increase profit margins on drugs sold). Discussion will focus primarily on the problem of sample bias in relation to this research and then be extended to consider this issue more broadly for other sociological research on the Internet and WWW. Following this, a detailed description of how this research was carried out will be combined with some guidance on how other researchers, new to this technology, may utilize it for their own research concerns.

Background to my Internet Based Research on Drug Dealers

2.1
Between 1994 and 1996 I carried out a review of the disparate forensic evidence concerning illicit drug purity (which partially includes the analysis of other substances found in them). Further to this, I also interviewed 31 drug dealers,[1] located in South East London. The combined pieces of research (Coomber, 1997a; 1997b) enabled me to make the assertion that the practice of dangerous drug adulteration/dilution[2] was neither common, nor for a variety of reasons logistically likely. This assertion was contrary to professional (drug agency workers), academic (drug field related), drug user, drug dealer, political, media and thus general public perception of what makes up street drugs. In fact, I further asserted (again on the basis of the forensic evidence and professed practice of the dealers interviewed) that the 'cutting agents' that are found in street drugs were there either 'to enhance' the drug, to bulk it out (but safely), or both. As such, common diluents found to be present in street drugs were lactose and glucose whilst common adulterants were paracetamol and caffeine. Finally, not only did I conclude dangerous adulteration/dilution was anomalous, but I also found that the actual practice of adulteration/dilution with any substance was not as prevalent as commonly believed. Primarily, this was due to the other means which were open to drug dealers to secure profit on their sales and that many dealers were not confident in tampering with the drugs they sold in this way. For example, 65% of those who sold heroin and 73% of those who had sold amphetamine said that they never adulterated or diluted the heroin/amphetamine that they sold (Coomber, 1997b). Most of the findings from my work were at odds with what was normally considered to occur in the trafficking and dealing of illicit drugs. In the conventional model it was assumed (as per Preble and Casey, 1969) that as drugs passed down the chain of distribution (from the importer through to the street dealer) they were 'stepped on' or 'cut' at each level to increase the profit margin of each purchaser. This essentially USA model, countered by my research, did not match the reality of the drug trade in the UK in the 1990s. The research question I then wanted to ask was whether it any longer fitted the reality of the drug trade in the USA or any other countries. This was not an area of research being pursued elsewhere however, and accessing drug dealing populations on an international scale is obviously prohibitive in all manner of ways in the absence of substantial funding. The Internet beckoned as a means for me to explore the question further.

2.2
As I shall discuss below, that I had some prior significant familiarity with the research area underpinned the efficacy of using the Internet as a means to carry out this piece of work. The questionnaire I used can be viewed online.

My Findings (Briefly)

3.1
Regarding the research I carried out via the Internet and the WWW, 80 responses (considered to be reliable[3]) were received from 14 countries and 4 continents. USA dominance of the Internet was reflected by the fact that 40% (32) of responses came from that country. The rest were from countries such as the UK (10), Canada (9), Australia (5) and others from as far afield as Finland, South Africa and New Zealand.

3.2
Consistent with the forensic evidence and with my earlier work (Coomber, 1997b), neither the extent of involvement in selling drugs (whether drugs sales represented total or supplemental income) or the point in the chain of distribution that they were located (importer, wholesaler, 'street-dealer') could enable the detection of the likelihood of someone being involved in adulteration/dilution practices. The majority of my respondents, (63), reported either 'never' (47) or 'rarely' (16), cutting the drugs that they sold. Again, this is consistent with my earlier work which shows that purity levels of drugs such as heroin seized at street level differ less than might be expected from samples seized at, or prior to, importation (Coomber, 1997a). Whilst most respondents believed that the cutting of drugs with dangerous substances was common, none reported using dangerous adulterants/diluents such as strychnine, brick-dust, domestic scouring powders such as Vim or Ajax ('Vim-in-the- Veins') themselves. This question was also accompanied by a question asking for first hand knowledge of such cutting (with details) - so that they could, if hesitant to admit to the practice personally, instead refer to some mythical other as the one responsible. Evidence of dangerous adulteration - as in Coomber (1997b) - was not forthcoming. The primary reasons given by the respondents as to why they did not use dangerous substances were of the humanitarian/ethical kind: 'I don't want to hurt anyone' and/or of the rational calculative 'the comeback', 'it's not good for business', the two often overlapping. Other important reasons involved the common practice of 'bagging' and 'wrapping' of larger quantities into smaller drug sales where the mark-up on smaller sales enabled reasonable profit, and the (slightly rarer) practice of 'skimming' where sales were sold slightly underweight, again increasing profit margins. These methods, often employed as the primary means to secure profit, also often made the dilution of drugs unnecessary. Overall, this research (Coomber, 1997a; 1997b) suggested that the picture of adulteration/dilution practices being built up for the UK in the 1990s was likely to be more broadly generalizable to other countries. Dangerous adulteration/dilution is in all likelihood a 'street myth', less adulteration/dilution than is commonly believed actually occurs, and that when it does it is with relatively harmless substances such as caffeine, glucose, lactose, paracetamol and other 'over the counter' drugs. Discussion of the methods and sample involved will be discussed more fully below[4].

My Sample

4.1
I would have liked to have had a representative one, but nobody knows what this looks like in relation to drug dealers. There is not even much in the way of hypothesizing (see Dorn et al, 1992 for some discussion on this). Drug dealers, by the very nature of their activities, are a very difficult group to access. Often the drug addict who would be willing to be visible for research purposes for research into drug addiction would not be willing to be visible as a drug dealer for research into drug dealing for obvious reasons. Given this, the main concern was to try to obtain a reasonable spread of individuals who had been involved in selling drugs at different levels of the chain of illicit drug distribution (user-dealer; 'street' dealer; wholesaler; friend-only dealer etc) and at different levels of involvement (received total income from drug sales; supplemented income; covered cost of drugs only) as this was considered to be primarily influential in adulteration/dilution practices. In my earlier work (Coomber, 1997b) reaching dealers at different points in the chain of distribution was largely achieved by access to a captive prison population. Those contacted who were not in prison all refused to provide me with contacts of other dealers with which they had connections. Using the 'snowballing' technique as a means to accessing either a broad population, or indeed any other drug dealers at all, was thus unsuccessful.

4.2
So, whilst my Internet sample were a voluntaristic sample, they were also probably a broader sample (in terms of the population sought) than would be achieved by other convenience sampling methods, such as snowballing, as the primary problem with using this method regarding drug dealers would be transcending levels of distribution, where in the research I carried out on the Internet a spread of contacts across the chain of distribution was achieved. Of those responding to my Internet survey, 19 received the bulk of their income from selling drugs. The majority 'supplemented' their income through drugs sales, although a significant number of these individuals were dealing in large amounts of drugs. Those classified as 'street' dealers numbered around 17.[5] There were at least eight 'middlemen' at the wholesale level, two who responded that they were importers, and one had been a manufacturer (MDA, MDMA). The majority of those remaining were relatively less involved and reported selling mainly to friends and acquaintances. As such, the sample attained for this research provided a reasonable spread of individuals that have been involved across the levels of drug distribution and with different intensities of involvement. The findings are obviously indicative and should be read with the necessary caution (accepting the likely relative sample bias). However, given that the findings tend to support both the forensic evidence and the more localized, terrestrial research which preceded it, it seems reasonable to suggest that the picture of adulteration/dilution practices built up for the UK may be more generalizable on the international level. If the findings had contradicted the earlier research this also would have been a significant indicator demonstrating both that important differences in adulteration practices take place in different locations and that the Internet was a valuable resource for finding such things out.

4.3
Undoubtedly in this particular case there were benefits from using the Internet, and specifically that it enabled me some access to a group who are notoriously difficult to research and to glean some interesting indicative information on a sensitive subject. However, some important methodological issues remain. When I first considered the results of my Web survey, I concluded that, because the results corroborated my previous research, the results cannot have been excessively biased. However, on further reflection, it became clear that there are actually two 'unknowns' here: neither the amount of bias, nor whether the results are 'true', are (or can be) known. all that can be deduced from this research is that either the results are true and there is no significant bias, or the results are false and there is bias.

Internet-User Sample Bias: Having a Method which is Appropriate for Particular Types of Research

5.1
Clearly, when using the Internet for survey research there will be a bias in terms of who is responding and a relative lack of representation of those who do not have access to the Internet. Those responding will be users of advanced information technology, with all that this suggests in terms of class/stratification, education, personal and life resources. A number of surveys into the demographics of Internet users have consistently found that Internet users are more likely to be white, male, first world residents, relatively affluent and relatively well educated in comparison to any more general population (Nielsen & CommerceNet, 1995; Kehoe & Pitkow, 1996). This obviously makes generalizing about research findings from Internet users to the general population highly problematic. Importantly however, the demographic research suggests that significant changes are occurring which move the user group in the direction of greater representativeness: 'While Internet users still tend to be upscale, their overall characteristics are coming more in line with general population averages', and, 'Internet access and use are becoming increasingly mainstream' (CommerceNet/Nielsen, 1996), also see Fisher et al(1996); Boncheck et al (1996) and Kehoe and Pitkow (1996). Good news for the future perhaps but a range of difficulties remain in the mean-time. Moreover, doing research via the Internet also presents its own specific issues regarding sampling which go beyond the representativeness or otherwise of the aggregate user population.

5.2
Work on Internet demographics has demonstrated that the Internet can be used to sample effectively to a point, and particularly that it can be used to produce relatively informative and reliable data about Internet users. Moreover, technical methods to improve the reliability of such first level research are constantly being theorized and refined (Urken, 1996; Kehoe & Pitkow, 1996). However, once we go beyond this, to use the Internet as a means to investigate beyond the use of the Internet, problems regarding sampling are exacerbated. There has however been little published to date about this, in particular because of a lack of appropriate exposure to and expertise regarding the Internet in the social science research community. Where such attempts have been made, there is some useful discussion of the pros and cons of using the Internet as a medium for accessing research subjects.

5.3
Fisher et al (1996) for instance, whilst investigating 'how citizens are using the Internet to participate in civil life' quickly realized that getting a representative sample was, even from a sample where Internet use itself was a defining parameter, a 'virtual impossibility'. Despite the significant survey research experience of the team, consultation with colleagues at the University of Cincinnati and other institutions and with those at the Public Opinion Research mailing list, no comprehensive solutions to the sampling issue emerged. The problem, simply put, was this:
There is no comprehensive list of individuals who use the Internet, nor is there any certainty about how many users log on from any particular node. ... Complications stem not merely from individuals having multiple accounts at various nodes or multiple memberships in various Internet groups (something analogous to having multiple phone lines) but also from the ability of 'lurkers' to read and reply to messages posted for groups to which they may not formally be registered (Fisher et al, 1996: p. 16).

And so, rather than trying to sample individuals 'out there', they decided to target a range of USAENET newsgroups and LISTSERV (email) mailing lists. Within this, they further stratified their samples by selecting obviously political and non-political groups. By sampling from a large enough number of groups of each type (30 or more) they assumed, utilizing the central limit theorem, that their samples would be of a reasonable spread.

5.4
In light of their experience, Fisher et al (1996) have suggested a range of procedures to improve the representativeness of samples when surveying on the Internet. These include 'a combination of political and technical strategies' (p. 22), such as gaining the confidence of the managers of the lists that are to be posted to and thus (perhaps) 'official' approval of the project, and the use of screening techniques to improve feedback about where the responses originated, amongst others. Importantly, they concluded that, despite the problems relating to survey research via the Internet and the need to develop more sophisticated techniques (and on this point also see Urken, 1996; Pitkow and Kehoe, 1996) in relation to the collection of data which could test formal hypotheses or models the mailing of surveys to 'mailing lists and newsgroups can produce data suitable for exploratory analysis' (p. 22, my emphasis).

5.5
Similarly, undertook a survey of those receiving political information direct from the White House (to email addresses and from the WWW) in order to assess how this affected the 'flow of political information' to civil society given that the media and other established institutions of mass communication were being partially by-passed. Again, their research had a particular population in mind, those who received political information via the White House, and these were contacted either direct to the email lists or through their visitation to the White House Web-Site. Responses to their survey suggest that there is evidence that those who utilize the WWW to access political information unmediated by the news media are 'changing the way that people feel about and participate in politics' (Boncheck et al, 1996: p. 6). Concern over representation, particularly in regards to voluntary response and non-response, was highlighted by Boncheck et al as a problem and discussed by them. The demographic status of their respondents, however, closely resembled those that derived from demographic surveys of Internet users in general, including no bias towards heavy users, leading them to conclude that: 'the demographic profile for individual users approximates the profile of that produced by a random sampling of users, if that were possible' (p. 4). For Boncheck et al, the data collected by means of the Internet can be considered useful indicative data upon which further research may build and information regarding the changing nature of political communication considered.

5.6
Surveys of particular Internet users raise somewhat different issues. The many USAENET newsgroups offer the opportunity to access populations with a wide range of interests. Some of these groups may be explicitly for individuals who have an interest, have taken part, or are taking part, in particular behaviours (eg. newsgroups on kite-flying, prostitution or heraldry), or who have a particular orientation towards certain ways of thinking and/or behaving (Alcoholics Anonymous; entrepreneurship; republicanism). Although there is a paucity of information regarding the surveying of particular newsgroups via the Internet, it is clear that many of the problems outlined earlier for more general surveys also apply here although in a different form. More specifically, a questionnaire posted onto one newsgroup may be posted onto numerous other newsgroups (as happened to Fisher et al, 1996), and this can cause immense problems unless some means of identifying where the response came from exists, and which is not always easy to ensure. Moreover, depending on the research population, this may be ethically prohibitive: for instance in my own research outlined earlier on drug dealers, the tracing or monitoring of respondents to any degree might have jeopardised the research itself and/or render the respondents open to prosecution.

5.7
If the concern of the survey is to contact members of a particular newsgroup only, then the occurrence of this could be extremely problematic, as the sample would be difficult or impossible to control. If, however, the primary concern is to contact individuals who are, say, kite-flyers, as opposed to members of the kite-flyers newsgroup, then it may not be. Kite-flyers may be expected to cluster around the kite-flyers newsgroup and thus a strategy of posting there is bona fide, but might not wish to be exclusive of kite-flyers from elsewhere. When looking for general group opinions, beliefs or other characteristics of populations who are members of specific USAENET groups, then concern over the difference between those who do respond and do not respond may be very important. Those who do not respond may predominately hold different opinions, beliefs or other characteristics from those who volunteered their responses, and this of course would significantly affect the research. It is, however, primarily the responsibility of the designer of the research to be aware of the issues that might produce a non-response bias (Shipman, 1988) and weight the responses appropriately. If, for example, we wanted to know whether box kites or some other popular design tended to be favoured by male or by female kite-enthusiasts and for what reasons, the Kite-flyers newsgroup could well be the source of a good basic data set. Non-response in this instance might not be as important than if the same group of people were targeted about say, the politics of kite-flying, although this would need to be considered carefully in the research design.

5.8
Doing research on Internet populations or with Internet populations can, then, provide certain research opportunities, but as Fisher et al (1996: p. 23) have stated in relation to sampling problems 'the Internet is not as open ... as a site for survey research ... as it appears to be at first flush'. We should not be too downhearted about this however, for there are other opportunities where the concerns of general surveys are lessened for sociologists and where the potential of the Internet remains real, as I will go on to discuss. Whilst researching particular types of Internet users as Internet users modify the sampling problem away from a concern with the representativeness of general populations, research not concerned with Internet users as a group involves other issues. For instance, sample bias may be of less concern to a researcher when they are interested in particular (especially 'deviant' or hidden) types of behaviour, as distinct from looking for representative or generalizable behaviour. 'Deviant' or 'rare element' sampling where the target population is difficult to access is in fact rarely able to aim for representativeness in any case (Smith, 1975). Thus, we might wonder, how much more data and how many more contacts Zinberg (1984) might have had for his studies on how people consciously act to control their drug use if he had been able to use a resource such as the Internet, rather than adverts in newspapers and college bulletin boards. Prior to Zinberg's work, the assumed representativeness of heroin users was that they were addicted users, while his research pointed to the 'invisible' population of heroin 'chippers' (occasional users) now deemed to predominate in heroin use, and the Internet, would most probably have been able to demonstrate the existence of large numbers of such users.

5.9
It is important to acknowledge that there is not always reliable information as to what a 'representative population' is. Prior assumption about what defines a given research population (derived from say treatment or criminal justice samples) and thus assumption (by default) of what constitutes a non- representative sample, has considerable dangers, not least that it can be lead to the dismissal of new information. Thus, Robins' (1974) methodologically tight follow-up research into the re-occurrence of opiate addiction amongst Vietnam veterans who had been addicts whilst in Vietnam was at first disregarded by experts, for it showed an extremely low (around 5% after one year) rate of relapse, and this finding was disregarded because it did not fit with the pre-existing theoretical stance on heroin addiction. However, twenty-five years later it stands as an informative and indeed classic piece of research.

5.10
Some groups, due to the existence of the Internet may well be more easily reached now than ever before. It is up to the sociologist, as a researcher to explore how useful the Internet might be to their particular research concerns. Whilst it is proper, indeed essential, to point out the problems with data derived from restricted sampling, it is also worth remembering that such data can also lead research in new and exciting directions.

Targeting a Sample, Preserving Anonymity and Ensuring a Response

6.1
The rest of this paper will outline how I carried out my research into drug adulteration/dilution practices via the Internet. For an alternative account of how survey research utilizing the Internet may be approached, see Smith (1997), and for those interested in how ethnographic approaches may be applied to cyberspace communities, see Paccagnella (1997).

6.2
There are a number of ways of attracting people who use the Internet to your research and some will be more successful than others. Simply having a Web page, for example (a 'location' on the Internet at which you can put information, a questionnaire or a link to other sites), will not be sufficient. This is analogous to waiting for people to come to you, and while some will (maybe), many will not. In my research, I decided to exploit those areas where it was self evident that those interested in drugs, and those who use or have used drugs, spend some of their cyberspace time - the drug related newsgroups. A rough search of drug related newsgroups (English language) turned up 23 main groups. These ranged from the alt.drugs.hard newsgroup where those interested in the so- called 'hard' drugs post questions, pose queries, answer questions and provide information, to rec.drugs.misc where issues around the so-called 'smart' drugs, stimulants and sedatives and other miscellaneous items are aired. Whilst it is true that drug dealers are potentially anywhere (and therefore posting to the thousands of different interest related newsgroups was a possibility), it made more sense (and conformed more closely to Net-etiquette - of which more later) to concentrate my effort around these groups. As such, my request to take part in the research was posted to each of the many drug related groups. Getting people to take part was less straight forward.

6.3
As previously discussed, accessing illicit drug sellers is a difficult business. This is made more problematic when the means used to receive responses - communication from one registered user to another - means that the sender is potentially (and normally) traceable: Email, for example, provides a sender's unique address. Concerns about anonymity were paramount in the minds of many of those I was trying to contact. One USA lawyer who had noticed my posting emailed me that, as the USA and the UK had arrangements for sharing of information involving criminal activity, I was unwittingly laying myself open to being subpoenaed to present the identities and/or electronic mail addresses of those contributing to my survey. Other 'surfers'[6] also expressed their concerns as to how I could be sure that the USA Drug Enforcement Administration (DEA) or other such organizations were not 'tapping' my line and thus tracing back information to where it came from, leaving me open to accusations of unwittingly entrapping others. These are serious issues and may concern others wishing to undertake sensitive research on the Internet (Lee, 1993; Frankfort-Nachmias & Nachmias, 1992). Such problems, however, are largely resolvable.

6.4
There are a number of ways that a researcher can protect their respondents and themselves from potential prosecution/subpoena. By effectively protecting our respondents and demonstrating to them that we can protect them, we are also more likely (logic suggests) to attract a greater response from what may be (and was in this case) a very vulnerable group.

6.5
Ideally, especially if you are expecting a potentially large number of responses you will have a questionnaire (for example) located on a Web Page which respondents can 'fill-in' on-line at a terminal. Included in the posting will be a 'click-on' address for the Web Page which will whisk them to the questionnaire which is set up there waiting to be filled in. 'Behind' this questionnaire there will be a programme (preferably on a secure server[7]) that will store the data being sent through. This may be a database or an appropriate statistical package[8]. This will have the double benefit of (a) providing inputted data ready for analysis, and (b) as these packages will store only the fields specified, information on where the data came from is not stored nor available. Thus, even if a researcher was to be asked for such data, they would have no information on who had sent it[9]. These facts should be (briefly) explained to potential respondents.

6.6
For the benefit of those potential respondents who may feel that such procedures (being invisible to them and needing to be taken on trust) are insufficient, the following two options should also be suggested: first, that they use an anonymous terminal (eg. in a public library, university laboratory, cyber-cafe) where the response cannot be traced to an individual. Second, that they print off the questionnaire (of which a text copy version is to be included at the bottom of the newsgroup posting) and return it by post (colloquially known by email users as 'snail-mail').

6.7
These precautions are somewhat onerous but when dealing with an (understandably) vulnerable and thus relatively more suspicious group, they are also essential.

6.8
But of course, any research posting must be interesting enough to get noticed and secure responses. My research experience suggests that a description of the research should be provided which will catch the eye of those being targeted. In the case of my research the Subject Heading used to advertise the posting on the newsgroup was: Have You Ever Sold Powdered Drugs? If So, I Would Like Your Help. If people's curiosity was raised by this heading then a simple 'click' on it would take them to the posting which explained in more detail what the research was and why it was being carried out. If still interested, a click on the posting then whisked people to my Web Page and questionnaire.

6.9
Because newsgroup postings are removed after a period of time a researcher using the Internet will benefit by re-posting the message regularly. New visitors are accessing the site all the time. As a posting moves down the list it may fall off the end (some people only look at the first 50 postings for example, and many newsgroups have literally thousands of postings). 'Re- post the message on a weekly basis' is the best advice here.

6.10
Having concentrated on the newsgroup which is pertinent to a piece of research, the researcher may then, depending on their resources (essentially, time), target more closely. This may be done by mailing to the individuals who have posted onto the newsgroup. Postings automatically record the email address of the postee unless they purposely do not provide it, although the vast majority do. For example, a hypothetical research project wanting to access individuals who have experience of using psychedelic drugs can by targeting more specifically individuals who have readily admitted (to a world wide audience) to have used such substances in their discussion postings. Newsgroups such as rec.drugs.psychedelic, alt.drugs.psychedelic or even plain old alt.drugs have numerous discussions on such issues. Emailing these individuals directly and politely asking them if they would be willing to contribute to the research is not unreasonable - they have already declared publicly their willingness to discuss issues around drug use and have provided an address for personal contact. Depending on the nature of the research and concerns around anonymity, many of these users (as was the case for my research) will be willing to contribute in a more in-depth way.

6.11
Not everyone who accesses newsgroups has what is known as a graphical interface (eg. Netscape), on their computer, and therefore these individuals will not be able to access easily[10] or fill in a Web Page based questionnaire. This problem will of course diminish over time, but at present it affects a large group of people world-wide. Accessing newsgroups on the Internet, however, only requires the ability to use email. Thus having a text copy of the questionnaire on the posting enables these individuals to take part if they wish to do so.

6.12
Whilst the Internet is overwhelmingly English speaking (and USA dominated) there are numerous foreign language newsgroups covering a vast number of countries. Here all that is required is a translation of the research posting into the appropriate language, and a posting to the appropriate newsgroups. The vast number of (often narrowly) differentiated newsgroups that are found in the English speaking groups may not be replicated in the non-English groups. Thus a newsgroup dedicated to drug issues does not appear under the French language newsgroups but a small number do in the German language newsgroups. Of course a researcher can duplicate the whole process, having a French, German, Spanish or whatever language questionnaire located on a Web Page with a storage package behind it but unless a great many responses from these newsgroups are expected a translation of the posting with the questionnaire added at the bottom will suffice and the translated data (posted, or emailed to you directly) can later be input into the Web based questionnaire by the researcher. Little more than a translation of the message is required, as posting is simple and quick.

6.13
My research included a posting (in German) to German newsgroups that were dedicated to drugs, such as de.alt.drogen or that were closely drug related or where drug issues had been debated, and also to French newsgroups (in French) where drugs issues had actually been discussed. The French newsgroups varied in their nature and tended to be more broad in their ambit than most English language newsgroups (mainly due to there being less of them), such as fr.misc.divers (a french newsgroup dealing with miscellaneous issues of a diverse nature) and soc.culture.french (a french newsgroup dealing with issues of society and culture). Again, depending on the nature of the research in mind this opens up new possibilities in terms of the numbers of potential participants involved in the research and, potentially, also in terms of cross- societal/cultural comparisons. For example, a broad based research project on 'identity' could potentially do very well on the various soc.culture (soc.culture.spanish; soc.culture.french; soc.culture.hungarian etc) news groups, whereas a project concerned with sexual abuse may struggle.

Net-Etiquette

7.1
The internet has its own general operating guidelines on how to use and not abuse it. This is colloquially known as 'netiquette'. Most newsgroups have their own variation of netiquette. It is important not to breach netiquette whilst undertaking research as the result will be few responses and a barrage of mail informing a researcher of their non-compliance. Breaching netiquette is deemed, simply, as rude and not to be encouraged or tolerated. Most basic netiquette is in fact fairly commonsensical, (for an introduction and general guidelines to netiquette see <http://rs6000.adm.fau.edu /rinaldi/netiquette.html>) but two basic practices should be avoided which may be tempting to the researcher.

7.2
First, 'spamming', which is the colloquial term used to describe the seemingly arbitrary posting of messages to multiple newsgroups, sometimes with the intent to cause some annoyance, but, for a researcher, in order to contact as many potential respondents as is possible. For example, the posting of puritan propaganda to drugs newsgroups (as well as many others) but with no real intention of involving or starting/making a constructive and useful contribution to that area, would be seen as a spam. If a researcher's target group is likely to be disparately located, they can of course avoid the accusation (and of course the 'corrective' mail) of spamming simply by posting separately to the different newsgroups. This is because the posting will show which newsgroups they have posted to. A posting with 100 different and differential newsgroups will tend to look like a spam. When an approach includes a range of closely matched groups, as with some (but not all) drug related research, then multiple posting to these groups is of course not unreasonable. It should be self-evident to those reading the newsgroup that such a posting is not inappropriately placed.

7.3
The second practice to be avoided is that colloquially known as 'flaming'. Essentially, this means resisting the urge to correct too dismissively (with inappropriate language or style of language) and postings which themselves 'flame' or rubbish (often with rude or dismissive language) an original request. Being 'flamed' to a personal email address is also something a researcher might experience if they are considered to have be breached netiquette.

Conclusion

8.1
Using the Internet as a tool for survey research offers exciting new possibilities to the research. However, whilst it is important that the potential of the Internet is grasped it is equally important that its limitations on research are understood. Using the Internet as a means to accessing samples in some way representative of general populations is currently prevented by who has access to it and who is using it. Moreover, even when the desired sample is of Internet users themselves significant technical and operational problems remain in terms of how to ensure the population targeted is in fact the population which responds. Given this, doing survey research on the Internet will, for the time-being, continue to present a certain amount of unknowns regarding sample bias. Despite these problems it is has been the purpose of this paper to refer the reader to a number of pieces of survey research undertaken via the Internet where the indicative data was deemed to be useful and the research worthwhile. For while the Internet poses methodological problems of one kind it opens up possibilities of others: access to hard to reach populations on sensitive topics, for example, and, as with my research on the adulteration/dilution practices of drug dealers, across national borders and even continents. Researchers that are aware of the problems presented by doing survey research via the Internet and who apply themselves appropriately, can, and I am sure will, increasingly carry out important research via this medium.

Notes

1 31 drug dealers were contacted through a variety of methods, including 13 who had been convicted of supplying drugs and were interviewed in a South London prison. The 31 included mainly 'street' dealers (an incorrect term as almost no dealing in the UK takes place on the 'street') but also two 'importers'. They ranged from those who had sold drugs over many years in (relatively) large volumes to those who had sold (sometimes intermittently) merely to support their own drug habit. Significant differences in the practices of adulteration/dilution of the drugs being sold were not evident regardless of the type and level of involvement in the drug distribution network.

2 The term adulterant is used in this paper to refer to substances added to illicit drugs in the process of selling and distribution. Adulterants proper, are in fact other psychoactive drugs (like caffeine, or paracetamol) which are much cheaper than the main substance, have a similar or complimentary effect when mixed with it, and therefore help hide the fact that the substance has been diluted. Substances which are not psychoactive, such as glucose and lactose, are more formally known as 'diluents'. These are added to a drug to increase the amount of drug available to be sold. It should be noted however that some substances which are found in street drugs will be the result of the particular manufacturing process used to make the drug. In this sense those substances might be more properly referred to as 'impurities'. 'Excipients' found in drugs (primarily pills/tablets) are the products used to bind the drug together. Common excipients are starch, gelatin or other gums (ISDD, 1994).

3 Five responses were self-evidently 'spoof' ones. Characteristically, they tended not to finish the questionnaire, apparently getting fed up half-way through and did not attempt to answer the questions sensibly. If any of the 80 responses considered reliable were false responses then these respondents tended to answer the questionnaire in full, with apparently consistent, informed and non- sensational answers. The researchers knowledge about common adulterants and diluents and other aspects of the dealing/cutting process aided in the assessment of how reliable the responses were. All survey research however suffers from the possibility that some returns will be disingenuous. In this case it is felt by this researcher that the 80 considered were reliable. This is partially supported by the similarity of returns to the 31 dealers interviewed in Coomber (1997b) where all were known to be drug dealers.

4 For those interested in a more complete account of this research and the findings see Coomber (1997c) and for a historical account of how and why such beliefs have emerged and are perpetuated, Coomber (1997d).

5 I state 'around' 17 because it is clear from a perusal of the data that many of those who said that they sold only to friends and acquaintances are not reasonably put in the friend-dealer category. For example, it is clear that 'acquaintances' was interpreted very broadly and often essentially meant that drugs were sold to individuals they trusted. Thus whilst these respondents were not selling to anyone who asked them they were also not only selling to friends as it is normally understood.

6 'Surfing the internet' is a colloquial term for people that ride over the electronic waves of the Internet from one destination to another. Basically, it is a way of describing a user of the Internet who is using it fairly indiscriminately. For example, wishing to know more about a particular topic, you can key in a search, be given the option of viewing a number of related Web Sites from which the search may then extend to where you know not useful and useless information being gained on the way as you 'surf'.

7 You could set the storage programme up on the hard disk of desk-top machines but this is of course far less secure (in a range of ways - from disk/file corruption to theft) than having it on a secure server which is regularly backed up as a matter of course. For those who have access, an arrangement with the manager of an institutional server linked to your Web Site would be preferable.

8 It is not essential to set up a questionnaire in this way but it is useful. The software needed to do this is increasingly user-friendly and will continue to become so. At this moment in time, access to help (eg. Computer Services in academic institutions) on setting up the questionnaire and database behind it is recommended.

9 Although this is true, institutions which provide Internet services (including the University of Greenwich server I had the data sent to), do log the address of the host machine although this information is rarely accessed, or used. The trick is to send it via a 'public' host machine, and thus make it impossible to be traced to an individual.

10 I say easily because it can be done through text based WWW client systems like Lynx but this is not terribly straight forward and most users of the Internet are increasingly using Web Browsers like Netscape and Microsoft Internet Explorer.

References

BONCHECK, M. S., HURWITZ, R. and MALLERY, J. (1996) 'Will the Web Democratize or Polarize the Political Process?', World Wide Web Journal, vol 1, issue 3, <http://www.w3 .org/pub/WWW/Journal/3/s3.bonchek.html>.

COMMERCENET and NIELSEN MEDIA RESEARCH (1996) 'Internet Hyper-Growth Continues, Fuelled by New, Mainstream Users', press release, <http://www.co mmerce.net/work/pilot/nielsen_96/press.html>.

COOMBER, R. (1997a) 'Vim in the Veins - Fantasy or Fact: The Adulteration of Illicit Drugs', Addiction Research, (accepted and forthcoming).

COOMBER, R. (1997b) 'The Adulteration of Drugs - What Dealers Do What Dealers Think', Addiction Research, (accepted and forthcoming).

COOMBER, R. (1997c) 'Dangerous Drug Adulteration - An International Survey of Drug Dealers Using the Internet and the World Wide Web (WWW)', International Journal of Drug Policy, (accepted and forthcoming).

COOMBER, R. (1997d) 'Adulteration of Drugs - The Discovery of a "Myth"', Contemporary Drug Problems, (accepted and forthcoming).

DORN, N., MURJI, K. and SOUTH, N. (1992) Traffickers: Drug Markets and Law Enforcement. London: Routledge.

FISHER, B., MARGOLIS, M. and RESNIK, D. (1996) 'Surveying the Internet: Democratic Theory and Civic Life in Cyberspace', Southeastern Political Review, vol. 24, no. 3.

FRANKFORT-NACHMIAS, C. and NACHMIAS, D. (1992) Research Methods in the Social Sciences (4th Edition). London: Edward Arnold.

INSTITUTE FOR THE STUDY OF DRUG DEPENDENCE (1994) 'What's in a Drug?', Druglink Factsheet 10, in Druglink, Nov/Dec.

KEHOE, C. M. and PITKOW, J. E. (1996) 'Surveying the Territory: GVU's Five WWW User Surveys', World Wide Web Journal, vol 1, issue 3, <http://www.w3.o rg/pub/WWW/Journal/3/s3.kehoe.html>.

LEE, R. (1993) Doing Research on Sensitive Topics. London: Sage.

NIELSEN MEDIA RESEARCH and COMMERCENET (1995) 'The CommerceNet Nielsen Internet Demographics Survey', <http://www.nielsenmedia.com/>.

PACCAGNELLA, L. (1997) 'Getting the Seats of Your Pants Dirty: Strategies for Ethnographic Research on Virtual Communities', Journal of Computer Mediated Communication, vol. 3, no. 1, <http://207.201.1 61.120/jcmc/vol3/issue1/paccagnella.html>.

PREBLE, E. and Casey, J. J. (1969) 'Taking Care of Business - The Heroin User's Life on the Street', International Journal of the Addictions, vol. 4, pp.1 - 24.

ROBINS, L. N. (1974) 'The Vietnam User Returns', Special action office monograph, series A, no. 2. Washington, DC: U.S. Government Printing Office, May.

SHIPMAN, M. (1988) The Limitations of Social Research, (3rd Edition). London: Longman.

SMITH, H. W. (1975) Strategies of Social Research: The Methodological Imagination. London: Prentice Hall.

SMITH, C. B. (1997) 'Casting the Net: Surveying and Internet Population', Journal of Computer Mediated Communication, vol 3, no. 1, <http://207.201.161.12 0/jcmc/vol3/issue1/smith.html>.

URKEN, A. B. (1996) 'Polls, Survey, and Choice Processor Technology on the World Wide Web', World Wide Web Journal, vol. 1, no. 3, <http://www.w3.or g/pub/WWW/Journal/3/s3.urken.html>.

ZINBERG, N. (1984) Drug, Set and Setting: The Basis for Controlled Intoxicant Use. London: Yale University Press.

Copyright Sociological Research Online, 1997