The term “online research” refers to two different concepts which are often confused: (1) applying online methods in social research, and (2) social research of online phenomena. The first part of the article sketches the most important empirical methods and their online version. The second part briefly illustrates how selected online phenomena or research questions can be methodically addressed.
Online Methods In Social Research
Many traditional methods in communication studies have been successfully transferred and adapted to the Internet as a research tool. Only a few online methods are completely new developments (e.g., Internet data mining with robots or search engines).
Quantitative Surveys
The traditional methods of quantitative surveys are face-to-face interviews, (computerassisted) telephone interviews, and self-administered questionnaires (paper and pencil), either disseminated via mail or administered in groups that are present (e.g., students on university courses). Online surveys apply self-administered questionnaires, which are usually presented as a series of web pages including questions, items, and answering options. Three strategies are commonly used to establish contact between a questionnaire and potential participants: email invitations, hyperlinks on websites announcing a survey, and unexpected “pop-up” windows introducing a survey.
Web questionnaires have technical, socio-psychological, and economic advantages: all kinds of multimedia (photos, illustrations, animations, films, spoken word, music, etc.) can be displayed. Participants’ answers can be dynamically evaluated, commented, rejected, or corrected, which enhances data quality and reduces the need for manual data correction. The order of questions and items can be rotated, systematically or at random. Questions and answers can be filtered and modified in real time (for example, if a participant indicates that the “Tonight Show with Jay Leno” is his or her favorite television show, the program title can be directly inserted into the follow-up question: “what are your motives for watching the ‘Tonight Show with Jay Leno’?”). The interview situation is completely anonymous; thus, online surveys can deal with sensitive social or psychological issues (loneliness, exposure to pornography, drug abuse, etc.), where all traditional survey modes would imply grave effects of social desirability. As no personal interviewers are involved, interviewer effects (sympathy/antipathy, accentuation or variation of questions and wording, etc.) can be ignored. The execution is almost free: online surveys can be run on a simple Internet server (with database server software), no interviewers are needed, no postal or telephone charges (as in postal surveys) have to be paid, and no manual data entry is needed. Consequently, the sample sizes of some online surveys can be gigantic (up to hundreds of thousands).
On the other hand, there are crucial disadvantages, many of them related to sampling procedure and sample quality. As there is still a significant number of people who are not online, online surveys cannot replace telephone or face-to-face interviews if a representative sample of a country’s population is aimed at. The total population of Internet users is unknown and there is no complete email directory. Hence, no valid random sampling technique is available, and no evaluation of a sample’s representativeness, as long as a survey aims at the whole population. If a restricted and known population is under analysis (e.g., visitors to a website, a company’s employees, students at a university), appropriate random sampling techniques permit representative samples (e.g., drawing email addresses from a list, presenting a pop-up questionnaire to every nth visitor to a web page). There is no safe method of barring individuals from completing an online questionnaire several times. This problem especially applies to polls yielding results of some public or personal relevance, where participants might try to achieve a higher weight for their personal opinion by repeatedly answering the questionnaire. As in most other surveys applying self-administered questionnaires, there is no control of the questionnaire completion situation in online surveys, i.e., participants can collaborate, look up an encyclopedia, or visit other websites during a test. As text reading on a computer screen is laborious, and Internet users are often impatient, online questionnaires should be comparatively short and easy to answer.
Qualitative Interviews And Focus-Group Discussions
Qualitative personal or group interviews demand personal contact between interviewer and interviewee. Only in a face-to-face setting can a confident communication situation be established, and paraverbal or nonverbal signals (e.g., emphasis, facial expressions, sweating) be properly registered. Nonetheless, high-quality audio and/or video conferencing techniques offering real-time and sensory-rich signals can be pragmatic and affordable tools for qualitative interviews. In in-depth interviews, collaborative software, which enables interviewer and interviewee to edit and comment texts, charts, or pictures together, may be a help. These and other tools may be applied in qualitative research if there are great geographical distances between interviewer and interviewee (e.g., in international studies) or if the interviewee prefers not being in personal contact.
Online Experiments
Conventional laboratory experiments are laborious. Subjects have to be acquired individually or in groups; they have to show up at the laboratory at an arranged point of time; if computers are needed for stimulus presentation and/or subject inputs, each participant needs access to one, etc. Many experimental settings involving stimuli and measurements that can be displayed on a computer can also be conducted as virtual experiments on a website.
The basic advantages are: (1) no investigator is needed; (2) participants can use their own computer at home or in the office; and (3) they can perform the experiments whenever and wherever they want to. These characteristics in sum result in a comparatively natural exposure situation. Finally, virtual experiments often exhibit extraordinarily large samples. There are major disadvantages as well: as indicated for online surveys, the sample quality may be poor and experimental groups may differ significantly. Participants have to administer the experiment on their own, and the researcher has poor control over the execution. If the prepared experimental procedure does not work at some stage or if the participant does not understand the instructions and requirements, the execution may be canceled and/or the received data may be corrupt.
Thus, virtual experiments only work if they can be taken easily and rapidly, and if participants are highly involved and keep to the rules. Experiments involving measures of personality may serve as an example here: many personality tests comprise simple stimuli and only a few questions and item batteries, and most individuals are highly interested in learning about their own personality (e.g., “how good am I in coping with crises?”). Another area of application uses tests of websites and other online media (usability, user selection and navigation, etc.). After random allocation, site visitors are presented with different stimulus versions of media content. Navigation behavior can be measured by log-file analysis and afterwards an online questionnaire can be administered.
Observations
Online research provides a rough, yet completely unobtrusive, method of observing human behavior – as far as it refers to the Internet: log-file analysis measures which page or document website visitors opened at what time, what part of the screen they looked at, whether they scrolled up and down the page, and other kinds of overt behavior that cannot be observed. Log-file analysis is the online variant of TV metering. The difference between the techniques is that people-meter systems are restricted to a sample while log-file analysis measures all visitors’ actions on a website. With the help of cookies (little files stored on a visitor’s computer by a web server), separate visits to a website can be recognized and connected to other data sources (e.g., online questionnaires answered on the same computer). There also exists (client-based) software that protocols all email and web activities of a person. Like TV metering, this method requires the participants’ permission and active software installation on their computers.
Data Mining And Collaborative Work
The Internet is a worldwide network of computers containing all kinds of mass media content, personal messages, interpersonal communication, group communication, and other information interconnected by hyperlinks and retrievable by search engines. Accordingly, it can be used as an endless source of social data and documents, which can be gathered unobtrusively with the help of search engines, robots, and other kinds of automatic data mining. An example may illustrate the potential: public opinion toward a politician, as publicly communicated on the net, can be sampled by searching for the politician’s name with a standard search engine, saving all web pages, contributions on discussion forums, and comments on weblogs, and finally running a conventional content analysis. The Internet not only offers new capabilities for gathering empirical data, but also facilitates and enhances traditional methods of literature research, expanding the availability of research literature beyond countries, disciplines, and scientific communities. This kind of online data mining can yield research data for extended meta-analyses, which would not have been possible in former times. The same applies to Internet-based tools for collaborative work.
Social Research Of Online Phenomena
The Internet is a virtual world in which almost all real-world phenomena exist too. This especially applies to social and communication phenomena. Some real-world phenomena have changed dramatically in the virtual condition (e.g., accessibility of pornographic and extremely violent material to children; anonymous online communities); others still are quite comparable in both “worlds” (e.g., mainstream media content o journalism). As it is impossible to draw a complete picture of all analyzed online phenomena, some examples might give an impression of the field.
Social research on the Internet begins with descriptive demographics of the Internet population (i.e., the number and structure of online users). Another “classical” realm of online research is website evaluation, including usability tests (combining retrieval tasks and think-aloud technique in the usability lab; additionally applying eye tracking, screen recording, and video observation of participants), user surveys, clickstream analyses (logfile analysis), and focus-group discussions. In advertising research, content and effects analyses of banners and other forms of mass media and direct advertising on the Internet are of relevance. Research on online journalism and public relations (PR) looks at communicators’ characteristics on the Internet and new forms of interactions between journalism and PR. Research in political communication addresses political participation on the Internet, e-government as Internet-based organization of policymaking and administration, the impact of politicians’, parties’, and other institutions’ online communication on public opinion, etc. Research in health communication deals with the effectiveness of online campaigns, and recipients’ use of and trust in online medical information. Audience research discusses uses and gratifications of different media channels and audience acceptance of new media products. Psychology and sociology analyze processes of interpersonal and group interaction, and the emergence or development of online communities, while rhetorical studies deal with discourse strategies, codes, and styles on the Internet. One might easily figure out how this enumeration is to be continued.
It is important to note that social research of online phenomena often, but not always, applies online methods. The demographics of the Internet population, for example, cannot be addressed with an online survey. Such data inevitably have to be based on a random sample of a country’s total population, which, in turn, calls for a traditional telephone or face-to-face survey. It is hard to forecast the future of online research: online methods, as well as online phenomena, change at such a dramatic pace, that today’s detailed descriptions of the field will be at best of historical value tomorrow. But hopefully the basic methods and ideas that this article has sketched will remain of relevance in the nearer future.
References:
- Alreck, P., & Settle, R. (2003). Survey research handbook, 3rd edn. Boston, MA: McGraw-Hill.
- Batinic, B., Reips, U.-D., & Bosnjak, M. (2002). Online social sciences. Göttingen: Hogrefe.
- Dillman, D. A. (2006). Mail and Internet surveys: The tailored design method. New York: John Wiley.
- Mann, C., & Steward, F. (2000). Internet communication and qualitative research: A handbook for researching online. London: Sage.