Student Researchers: Sriram Kalyanaraman (PhD Student) and Carson B Wagner (PhD Student, University of Colorado).

Faculty Supervisor: Dr. Shyam Sundar Sethuraman

For a complete report of this research, see:

Sundar, S. S., Kalyanaraman, S., & Wagner, C. B. (2001, August). Titillation, frustration, or just plain orientation? Teasing out the "tease effect" of slow downloading. Paper presented to the Communication Theory and Methodology Division at the 84th annual conference of the Association for Education in Journalism and Mass Communication (AEJMC), Washington DC.

Introduction

The exponential growth in the number of Websites as well as Web users in recent years, and the resultant bottlenecks in online reception, have introduced a new variable to the literature of mediated communication: Download Speed. Different Websites download at different speeds, depending on their use of memory-heavy features. Different users receive the same Web content at different speeds, depending on their connection to the internet. Recent survey findings suggest that users perceive download speed of Web pages to be the single most important factor in determining their level of online use, and this has led to the emergence of download speed as a critical component of user-interface design, with an all-out emphasis on increasing speed and reducing download time. However, prior research (Sundar & Wagner, 1998) has shown that the slow-downloading version of a sexual image is more physiologically arousing than the fast-loading version of the same image. It is not clear however whether this is due to titillation, frustration, or orienting response. This paper explores these three theoretical mechanisms for explaining the so-called "Tease effect" with two experimental studies.

Study 1

Three possibilities--titillation, frustration, and orientating responses-were examined to explain the arousability differences between slow- and fast-loading images. The tittilation (or "tease effect") explanation describes the phenomenon by saying it is more arousing to receive something in stages, with the promise of more to come, rather than receiving it instantaneously and denuding the anticipation factor. Frustration, on the other hand, says that the wait necessitated by slow downloads induces emotions of annoyance or boredom if expectations of immediate sensory stimulation are not met by the computer, and in turn such feelings will result in expressions of physiological arousal. Lastly, interpreting results using orienting responses (ORs), one would conclude more simply that this automatic physiological and behavioral response was triggered by experiencing a slow download as a novel stimulus.

The logic is that, by using a non-sexual image, we can empirically determine whether download speed elicits a "tease" response (Sundar & Wagner, 1998) even in the absence of a sexually suggestive image. If it does, then we may conclude that the effect is NOT purely a "tease" effect, because there's nothing seductive about a non-sexual image appearing on the computer screen in stages, and thus we rule out the titillation explanation. Depending upon over-time physiological data, it could either be frustration (if we observed gradually rising skin conductance level, or SCL, over the downloading phase, with a greater slope in the slow download condition) or OR (if we notice a sudden rise in slow, but without a concomitant rise in fast, download condition sometime during the beginning or middle of the downloading phase of the experiment, most likely in response to a specific content element). However, if on the other hand, the slow download condition does not prove to be physiologically superior to the fast download condition, then we may accept Sundar and Wagner's (1998) contention that there is a "tease" effect operating, or that our present design suffers from some methodological flaw, which would necessitate further empirical exploration.

Method

Thirty individuals in a between-participants experiment were exposed to a Website featuring an image. For half the participants, the image downloaded on the screen slowly, in stages. For the other half, the image downloaded relatively quickly. Participants' arousal during the entire exposure to the image was measured via skin conductance. Subsequent to the image, all participants were linked automatically to a news Website, which loaded uniformly in both conditions. Physiological arousal was again measured during their exposure to the news site. In addition, an unobtrusive measure of their browsing activity was obtained by recording the number of hyperlinks selected/clicked by each participant.

Results

The arousal and excitation transfer measures failed to show a statistically significant differentiation as a function of download speed, although the direction of means indicated that the slow download condition elicited greater skin conductance levels than did the fast download condition. Mixed ANOVAs with arousal and transferred arousal as dependent variables failed to yield any significant effects.

The number of links visited/clicking activity was examined in a t-test with download speed as the independent variable, and the result was statistically significant. Participants in the slow-loading condition clicked on a significantly higher number of hyperlinks than their counterparts in the fast-loading condition.

Study 2

Study 1 was designed to identify specific theoretical formulations that could explain the relationship between download speed and the dependent measures of not only immediate arousal after stimulus exposure but also subsequent transferred arousal and behavioral activity. If the negative relationship between download speed and arousal is replicable with a non-sexual image (namely a flower with a grasshopper in it), then Sundar and Wagner's (1998) contention that slow-downloading is arousing because it mimics a striptease routine is unlikely to be the most compelling explanation. We however found a negative relationship but it was not statistically significant.

When the content element is conceptualized in terms of arousability (or arousal potential), as opposed to its sexual/non-sexual nature, it becomes important for us to test whether download speed differentially affects high vs. low arousing images that are both non-sexual. Our comparison with Sundar & Wagner (1998) seems to differ as a function of content arousability, given the sexual nature of their stimulus material. That is, their stimulus content had high arousal potential and was sexual in nature. Ours was decidedly non-sexual, and it's not clear what the arousal potential of our image is. It may be argued that, based on classical research, nature shots like the one we used are distinctly low in arousal potential. On the other hand, our stimulus did not just show a flower in isolation but also included a grasshopper in the center of the flower, and some recent research has shown that insects do have arousal potential. all things considered, our stimulus material can perhaps be characterized as an image with "medium-arousal potential." Therefore, we proceeded to create a low-arousal version of the image used in Study 1, with the insect removed. Thus, the only difference between the two studies is this subtle change in the stimulus image; while Study 1 employed a flower with a grasshopper in the center, Study 2 employed the exact same image with the only difference being that the grasshopper was cropped off.

Method

Twenty-six participants took part in Study 2, which was similar to Study 1 in every respect except for the change in stimulus material. We also included an extra set of dependent measures to determine participants' affective reactions to the image as well as their reactions to the download, on a paper-and-pencil questionnaire administered after their browsing activity.

Results

We showed a significant effect for download speed, but the direction of means were in the exact opposite direction of what was found in Study 1 as well as in Sundar & Wagner (1998). Participants in the fast-download condition showed significantly higher arousal, on average, compared to participants in the slow-download condition during the downloading phase of the experiment This difference was carried over to the CNN-browsing phase wherein participants showed significantly higher transferred arousal in the fast-download condition than in the slow-download condition.

A mixed ANOVA with arousal as the dependent variable echoed the significant main effect for download speed and also showed a main-effect for time-segment, such that the arousal for the downloading flower image was highest during the first five seconds and declined steadily for the three time-segments thereafter. The same ANOVA with transferred arousal as dependent variable did not show a similar decay in arousal during the CNN-browsing phase. The mean arousal (expressed as a percentage of baseline) held steady around .20 throughout the 20-second news browsing period.

None of the adjectival measures relating to participants' perceptions of the flower image showed significant variation as a function of download speed.

Conclusion

If we take into account that the Sundar and Wagner (1998) study employed a high-arousal image, whereas the two studies reported here employed medium- and low- arousal images respectively, the above findings can be understood quite clearly by forwarding a transverse interaction explanation, accounting for the combined effect of content arousability and download speed on SCL. That is, when an image has high arousability (such as the sexual image used in Sundar & Wagner), it evokes significantly greater SCL when downloading slowly rather than quickly. This distinction between download speeds seems to be nullified when considering a medium-arousal image (like the one we used in Study 1). However, when considering a low-arousal image (Study 2), the effects are diametrically opposite to that of the high arousal image such that fast downloading image evokes greater SCL than the slow downloading image. More research is needed with wider samples of stimuli that are differentially arousing and vary systematically on those content characteristics (such as sex) that have particular implications for arousal generation. Even though this line of research is devoted to the study of a technological variable, content appears to play a crucial interactive role and is therefore of central concern in helping us develop a fuller understanding of the psychological effects of download speed in the Web medium.