Faculty Researcher: Dr. Shyam Sundar Sethuraman
For a complete report of this research, see:
Sundar, S. S.(2004). Loyalty to computer terminals: Is it anthropomorphism or consistency? Behaviour and Information Technology, 23 (2), 107-118.
Click here
for the press coverage of this paper.
Introduction
Most, if not all, school and college campuses house a wide array of computer labs to accommodate a variety of student needs. Students regularly visit computer labs and sometimes even stand in line to use these public computer terminals. The rapid growth of the internet, coupled with the increasing reliance on computer-based activities in institutions of higher education has made it incumbent on University students to regularly visit computer labs on campus. Such a heightened level of communicative engagement with a machine is unlikely to be without its psychosocial consequences. The psychological tendency to behave socially with a computer is quite well documented in the literature. The short-term socialness of human-computer interaction is said to be automatic and mindless. However, what appears to be less understood is the tendency, if any, for computer users to form long-term social relationships with machines. Is the social nature of human-computer interaction replicable in human-computer relationship? In particular, do we show loyalty to particular computer terminals over a period of time? This paper attempts to answer these questions through two studies.
Study 1
Hypotheses
Based on prior research on human-computer interaction, we proposed the following hypothesis:
H1: Students using computer labs on campus will tend to repeatedly frequent the same computer terminal(s) during subsequent visits.
Method
An electronic observation method was used to gather data to test our hypothesis. With the help of staff from the computer center, the use of 18 computer labs was monitored for a period of 30 days. These 18 labs were chosen because they were all general-purpose computer labs where students can perform a variety of computer-related task. All computers in all the labs under observation require that students log in with their unique identifier before using the terminal. This helped us record the visits by various students, discriminate between different students, and also track their terminal choice during subsequent visits. In addition to login ID, the electronic observation data included the IP address of the computer logged into, identification of the lab where the computer was located, and the date and time of log on.
Results
During the period under study, 42,845 users visited 871 different terminals in the 18 sampled computer labs for a total of 210,283 times. The overall average number of visitations per user was 4.9. This provides us with a reasonable rate of repeat visitations to enable a test of our hypothesis.
H1: Supported. When t-tests were performed to compare the mean number of visits to mean number of "different" computer terminals used, a statistically significant pattern emerged for each of the 18 labs. Users, however, did not use a different computer terminal during each visit. In fact, they tended to use a significantly lesser number of "different" terminals, thereby exhibiting a preference for certain computer terminals.
Study 2
Although we found support for our hypothesis that users of campus computer labs tend to be relatively loyal to certain computer terminals, this behavioral pattern does not necessarily ratify the rationale behind our hypothesis. We derived our hypothesis from the literature on psychological anthropomorphism, but our electronic observation method does not permit us to examine the motivations for user loyalty toward computers. The tendency for repeat patronage of certain terminals may not only be explained by the natural human drive to anthropomorphize computer terminals but also by our desire for cognitive and/or behavioral consistency. Therefore, a face-to-face survey was conducted on the same campus with users of campus computer labs for understanding the theoretical mechanisms governing the behavioral tendency observed in Study 1. The survey in Study 2 was conducted under the assumption that while behavioral loyalty may be latent and not consciously realized by users, the attitudinal components of loyalty can be reliably elicited by self-reports.
Method
In particular, users' preference for consistency as well as their tendency to anthropomorphize computers were measured, along with their self-reported patterns of computer use.
In addition to preliminary questions regarding demographics and some masking items, the questionnaire asked respondents if there was a particular terminal that they preferred each time they visited the lab. This was a dichotomous question with two response options: yes and no. [This questionnaire item will henceforth be referred to as the hardcore-loyalty question].
Respondents who answered in the negative were directed to a question asking them how likely they were to choose one among a set of preferred terminals if at least one of those terminals is available during their next visit to the lab. This was a continuous variable with ten response options anchored between "not at all likely" and "very likely." [This questionnaire item will henceforth be referred to as the reinforcing-loyalty question].
The questionnaire comprising these and other items described above were administered in person to a probability sample of one hundred and ninety-one (N=191) students on the same campus where Study 1 was conducted.
Results
On the hardcore-loyalty question, 163 out of 191 respondents provided an answer. Of these, only 42 (25.8%) of the respondents said that they preferred one particular terminal during each visit to the lab.
On the reinforcing-loyalty question, when asked how likely they would be to choose one among a set of preferred terminals during their next visit, they scored, on average, 7.56 on the 10-point likelihood scale.
Over 180 respondents filled out each of 12 items in the anthropomorphism scale, with means for items ranging from 3.85 to 7.40 on a 10-point scale. Also, over 180 respondents filled out each of the 18 items in the consistency scale, with means for items ranging from 3.18 to 5.95 on a 7-point scale of agreement. The dichotomous hardcore-loyalty variable (wherein the "yes" value was coded as 1, while the "no" answer was coded as 0) was first entered as a dependent variable in a logistic regression with both the anthropomorphism and consistency indices serving as continuous regressors. Both indices were positive predictors of hardcore loyalty, such that higher scores on either anthropomorphism or consistency independently increased the odds of expressing hardcore loyalty toward computer terminals.
In order to compare anthropomorphism and consistency means between those who scored 1 and those who scored 0 on hardcore loyalty, a set of t-tests was performed as follow-up analysis. The results revealed that respondents who openly professed hardcore (HC) loyalty toward one particular terminal were likely to score significantly higher on anthropomorphism compared to other respondents.
In addition, respondents who openly professed hardcore loyalty toward one particular terminal were likely to score significantly higher on consistency compared to other respondents.
Further analyses were conducted by classifying respondents as belonging to either "high," "medium" or "low" anthropomorphism and consistency conditions, respectively. When reinforcing (RC) loyalty was used as a dependent measure, respondents in the "high" anthropomorphism condition scored significantly higher than did respondents in the "medium" anthropomorphism condition.
Similarly, respondents in the "medium"
consistency condition scored significantly higher than did respondents in
the "low" consistency condition.
Conclusions
Findings from the studies reported here suggest
that the variance in people's tendency to enter into rich human-computer
relationships is partly accounted for by their tendency to anthropomorphize
and partly accounted for by their innate need for consistency in orienting
toward the object of their relationships. The study demonstrates that we
can elicit long-term social responses toward computers in a natural setting,
but more importantly, without embedding human-like cues in the interface.
It also forebodes increasing acceptance of computers in a variety of personal
roles hitherto performed exclusively by humans, and offers two theoretical
mechanisms to explain why certain individuals might be more comfortable
than others with computers performing certain roles in society.
For more details/information about this study, please contact:
Shyam Sundar Sethuraman
Associate Professor and Director
Media Effects Research Laboratory
College of Communications
Penn State University
University Park, PA 16802
Ph: (814) 865 2173
E-mail: sss12@psu.edu
Back to Research Projects Page
Media Effects Research Laboratory |
 |