Knowing Me, Knowing You: Reciprocal Self ...

1 downloads 2735 Views 108KB Size Report
tion could be answered in a scrolling text box measuring 175 cm in width and 45 .... JavaScript Web pages that make HTML forms for re- search on the Internet.
CYBERPSYCHOLOGY & BEHAVIOR Volume 4, Number 5, 2001 Mary Ann Liebert, Inc.

Knowing Me, Knowing You: Reciprocal Self-Disclosure in Internet-Based Surveys ADAM N. JOINSON, Ph.D.

ABSTRACT Candid self-disclosure is desirable for many behavioral science studies. Although there is ample evidence that self-disclosure is increased when people communicate or participate in research over the Internet, few studies have looked at ways of increasing this effect. In the present pilot study, participants were randomly allocated to either a condition in which they received self-disclosing information about the experimenter (and then moved on to the study) or were directed straight to the study. Participants completed six open response questions on the Internet. Participants who received the experimenter disclosure divulged a significantly higher quantity of information about themselves, but their answers were not scored as significantly more revealing or intimate than those participants who did not receive the experimenter disclosure. Implications for conducting research over the Internet are discussed.

INTRODUCTION

C

A N D I D S E L F - D I S C L O S U R E is crucial to many behavioral science studies—quite simply, without participants' willingness to disclose information about themselves, many researchers would be without data for analysis. Indeed, the rapid growth of the Internet as a medium for conducting research suggests that the worldwide web (WWW) is proving to be an invaluable tool for the completion of interviews, surveys, questionnaires, and psychometric measures, all of which require self-disclosure in some form or other. As well as sampling and cost advantages, conducting research via the Internet has the potential benefit of increasing candidness and reducing social desirability amongst participants. For instance, Joinson reports that scores on a standard measure of socially desirable responding were lower in par1

2

ticipants allocated to a WWW completion method compared to a pen and paper condition. A review of studies in which psychological measures were completed electronically concluded that electronic surveys, compared to pen and paper, reduce socially desirable responding. Studies of computer-mediated communication (CMC) have also found heightened self-disclosure during CMC compared to faceto-face participants. In many cases, it would be desirable to produce an online survey that maximizes candid self-disclosure by participants. In pen and paper surveys, this is usually done by stressing participants' anonymity and the confidentiality of their responses. Studies using the WWW replicate this technique by similarly stressing the confidential nature of participants responses, while also avoiding technologies (e.g., setting cookies to check for multiple submis3

4,5

Institute of Educational Technology, The Open University, Milton Keynes, United Kingdom.

587

588

JOINSON

sions) that would compromise this confidentiality. An intriguing technique for further increasing self-disclosure by participants was developed by Moon. Moon argued that people tend to respond to computers as social actors, and as such apply the same social rules to interaction with a computer as to interaction with a human. One such social rule is the principal of self-disclosure reciprocity. According to Moon, "There is substantial evidence that people will engage in intimate self-disclosure— even with relative strangers—if they first become the recipients of such disclosures from their conversational partners" (p. 324). Moon found that her participants disclosed more information, of greater depth, when the computer they were being interviewed by also disclosed information about itself. For instance, in Moon's study, the computer would state: "This computer has a Pentium II processor, so it's a very fast computer compared to most other models on the market today. It also has a DVD drive . . . And its hard drive is huge . . . 9 gigabytes." The computer would then ask the question "What characteristics of yourself are you most proud of?" Moon argues that, because people treat computers as social actors, they reciprocate the computers' self-disclosure with their own. However, the applicability of this technique to most behavioral science research is limited, in that the computer disclosed before each question. For traditional Likert-based questionnaires or surveys, such a methodology would be cumbersome, if not impossible. The research also was not conducted using the Internet, but instead relied on students turning up to sit in front of the experimental computer, so whether such a technique would work for WWW-based studies is unknown. However, it is plausible that such reciprocal self-disclosure could be triggered by the experimenter self-disclosing before participants completed an online survey. This has a number of potential advantages: Firstly, it is a methodology useable for all WWW-based surveys, as it simply requires a single page before the participant completes the measures. Secondly, it does not rely on the disclosure per question technique developed by Moon, and so 6

9,10

could be used for surveys and questionnaires of any length or format. Finally, it addresses some of the ethical concerns of collecting data from the WWW without the participant knowing who is collecting the data. In the present pilot study, participants were allocated to one of two conditions. In the first condition, they were presented with some basic demographic information about the experimenter, along with a photograph, e-mail address, and telephone number. In the second condition, participants were instead presented with no information about the author, but instead were directed straight to the survey. It is predicted that participants in the reciprocal self-disclosure condition will disclose more, of more depth, than participants in the nonreciprocal self-disclosure condition. In light of possible gender differences in self-disclosure, gender was included as a further factor in the analyses. MATERIALS AND METHODS Design A one-way, between-subjects design was used. Participants were randomly assigned to either the reciprocal or nonreciprocal condition using a Javascript on an introductory web page. Self-disclosure was measured for breadth (word count) and depth (using two trained raters). Participants Participants were 49 students of the Open University in the United Kingdom recruited via a posting on their computer conferencing system (34 female, 15 male). The Open University is a distance education provider. Participants' domain addresses were checked to ensure that there were no multiple submissions. Materials Experimenter disclosure. The experimenter disclosure took the form of a WWW page titled "Welcome to this Study." The information included the experimenter's name, a photograph, research interests, a brief career summary and

RECIPROCAL SELF-DISCLOSURE ON THE INTERNET

other biographic data (e.g., name of spouse, children, pets), the town he lived in, and contact information (telephone and e-mail address). (For a full copy of the experimenter disclosure, please contact the author.) Self-disclosure questions. Participants were asked to type answers to six questions: What do you most dislike about your appearance?, What recent event in your life has made you happy?, What do you most like about your appearance?, What has made you most proud with yourself?, What's the most embarrassing thing you've ever done?, and What recent event in your life has made you sad? Each question could be answered in a scrolling text box measuring 175 cm in width and 45 cm height. Procedure. Participants were recruited using a message sent to their student online discussions on 14 August 2000 and removed on 22 August 2000 with the subject line "Volunteers Needed for WWW Study." The message asked, "Could you help with a psychology/Internet study? The survey takes only 5 minutes to complete, and your help would be much appreciated." Below this message was a hyperlink to the survey start page. This page randomly allocated subjects to condition using a modified version of a Javascript developed by Birnbaum. In the reciprocal condition, the link took the participants to the experimenter disclosure page, and hence to the survey. In the nonreciprocal condition, participants were directed to the survey. On completing the survey, all participants went to the same debriefing page, where they were thanked and given an opportunity to contact the experimenter; those in the nonreciprocal condition were provided with a further link to view the experimenter's disclosure material. 7

RESULTS

Two analyses of the participants' responses were conducted: one to measure breadth of disclosure and one to measure depth of disclose. Breadth of disclosure was operationalized as the word count for each disclosure, with a sum calculated for each participant using the six

589

questions. Depth was scored using two trained raters who scored each answer on a 1-10 scale for the depth of intimacy. Each rater was given a definition of intimacy as "information that makes the discloser feel vulnerable in some way, for instance, emotionally vulnerable." The measure was calculated using the mean of the two raters' scores. The interrater reliability was 0.76. Breath of disclosure The word count for each participant (all six questions collated) was 76.63 (SD = 56.62). Differences in the breadth of disclosure were analysed using a two-way ANOVA (condition (reciprocal, nonreciprocal) X gender (male, female). Cohen's d was used as a measure of effect size. The main effect of condition was significant [F (1,45) = 5.16, p < 0.03, d = 0.59, means: 62.15 (SD = 30.16) and 93.00 (SD = 73.71)] for nonreciprocal and reciprocal respectively). The main effect of gender was nonsignificant [F (1, 45) = 1.45, p > 0.12, d = 0.41, means: 68.97 (SD = 44.47) and 94.00 (SD = 76.59)] for females and males respectively). The interaction between condition and gender was also nonsignificant [F (1, 45) = 1.25, p > 0.25, means: nonreciprocal, female = 59.83 (SD = 29.97]; nonreciprocal, male = 67.37 (SD = 31.99); reciprocal, female = 79.25 (SD = 55.85); reciprocal, male = 124.43 (SD = 102.32)]. 8

Depth of disclosure The mean intimacy rating participants' responses to each question was 4.01 (SD = 1.74). Differences in the depth of disclosure were analysed using a two-way ANOVA (condition [reciprocal, nonreciprocal] X gender [male, female]). The main effect of condition was nonsignificant [F (1, 45) = .50, p > 0.48, d = 0.10, means: 3.92 (SD = 1.47) and 4.10 (SD = 2.02) for nonreciprocal and reciprocal respectively]. The main effect of gender was nonsignificant [F (1, 45) = 1.06, p > 0.3, d = 0.28, means: 3.84 (SD = 1.49) and 4.37 (SD = 2.21) for females and males, respectively]. The interaction between condition and gen-

590

JOINSON

der was also nonsignificant [F (1, 45) = 0.88, p > 0.3, means: nonreciprocal, female = 3.90 (SD = 1.62); nonreciprocal, male = 3.95 (SD = 1.17); reciprocal, female = 3.78 (SD = 1.38); reciprocal, male = 4.85 (SD = 3.05)].

DISCUSSION As predicted, experimenter self-disclosure led to a greater breadth of self-disclosure amongst participants. However, contrary to predictions, this greater breadth of self-disclosure did not differ in depth according to either condition or gender. To take the first finding that experimenter self-disclosure is reciprocated by participants disclosing a greater breadth of information about themselves, this finding would be expected if the experimental encounter on the WWW is in some senses treated as a social interaction by the participant. Self-disclosure tends to be reciprocated, and in social interactions, lack of reciprocation is reacted to negatively by the original discloser. However, an alternative explanation is that participants were simply matching the speech patterns of their conversational partner (i.e., the experimenter). In the reciprocal condition participants are addressed with a higher number of words than in the nonreciprocal condition. Further research that involves the same number of words in the reciprocal and nonreciprocal conditions would be needed to differentiate between these effects. The second finding, that the depth of disclose was not influenced by condition is explicable if we assume that participants had little or no degree of freedom in their answers. For instance, if the thing you most dislike about your appearance is your nose, then you are unlikely to score that highly on intimacy—a nose is a nose. There is some latitude for explaining the problem—that your nose might be too big, too small, and so on, but the basic depth of your disclosure remains the same. As none of the questions asked about potentially hidden or deep-rooted intimate secrets, perhaps it is to be expected that breadth, not depth, was affected. A more accurate test of this notion would be to ask questions that do require a level of intimacy

to respond to, for instance, "what are you most afraid of?" or "what sexual fantasies do you have?" An alternative explanation is that in the present pilot research, the cell sizes were too small—increasing the likelihood of a type II error. More detailed research with a larger sample would confirm this explanation. Implications for behavioral science research on the WWW To be sure, not all behavioral science research requires candid self-disclosure. But, the results of this pilot study imply that by understanding the social nature of the experimenter—participant interaction on the WWW—it may be possible to encourage more candid responses to research measures on the Internet. Moreover, it could be argued that too many social science experiments are posted on the Internet for passing participants without enough information about the source of the study. Thus, not only does the method outlined in the present research increase self-disclosure, but it may well address ethical concerns associated with Internet-based data collection. When one is gathering data from students, the students obviously know the source of the experimental material (or at least the faculty/university). The results of this research suggest that providing Internet-based participants with similar information about the experimenter has definite benefits for the breadth of information they disclose. Further studies using different materials, matched word count, and a larger sample will be needed to check whether or not the effect extends to the depth of self-disclosure as well. The potential interaction between the experimenters' disclosures and the reactions of the participants is also worth investigating. As a starting point, however, the experimenterdisclosure approach suggests possible benefits for both the participant and the researcher.

REFERENCES 1. Birnbaum, M . H . (Ed.). (2000). Psychological experiments on the internet. San Diego: Academic Press. 2. Joinson, A . N . (1999). Social desirability, anonymity

RECIPROCAL SELF-DISCLOSURE ON THE INTERNET

3.

4.

5.

6.

7.

and Internet-based questionnaires. Behavior Research Methods, Instruments and Computers, 31:433-438. Weisband, S., & Kiesler, S. (1986). Self-disclosure on computer forms: meta-analysis and implications [Online]. C H I 96 Electronic Proceedings. Available: h t t p : / / w w w . a c m . o r g / s i g c h i /chi96/proceedings/ papers/Weisband / s w txt.htm. Coleman, L . H . , Paternite, C . E . , & Sherman, R . C . (1999). A reexamination of deindividuation in synchronous computer-mediated communication. Computers in Human Behaviour, 15:51-65. Joinson, A . N . (2001). Self-disclosure in computer-mediated communication: the role of self-awareness and visual anonymity. European journal of Social Psychology, 31:177-192. Moon, Y. (2000). Intimate exchanges: using computers to elicit self-disclosure from consumers, journal of Consumer Research, 27:323-339. Birnbaum, M . H . (2000). SurveyWiz and FactorWiz: JavaScript Web pages that make H T M L forms for re-

591

search on the Internet. Behavior Research Methods, Instruments, and Computers, 32:339-346. 8. Cohen, J. (1977). Statistical power analysis for the behavioral sciences (rev. ed). New York: Academic Press. 9. Archer, R . L . (1976). Role of personality and the social situation. In: Chelune, G.J. (Ed.), Self-disclosure:origins, patterns and implications of openness in interpersonal relationships. San Francisco: Jossey-Bass, pp. 28-58. 10. Chaikin, A . L . , & Derlaga, V.J. (1974). Self-disclosure. Morristown, NJ: General Learning Press.

Address reprint requests to: Adam Joinson, Ph.D. Institute of Educational Technology The Open University Walton Hall Milton Keynes, United Kingdom MK7 6AA E-mail: [email protected]