Usage and user experience of communication before and during rendezvous

Abstract

This paper reports a field evaluation of the mobile phone as a ‘package’ of device and services. The evaluation compares 44 university students' usage and user experience of communication before and during rendezvous. During a rendezvous (en route), students rated many aspects of the experience of phone use less favourably than before a rendezvous (prior to departure). This impairment of experience is attributed to the cumulative effect of various adverse factors that occur more often during rendezvous—incomplete network coverage, environmental noise, multiple task performance, time pressure, conflict with social norms, and conflict with preferred life-paths. Also, during a rendezvous, students were more likely to use the telephone, less likely to use e-mail, but equally likely to use text messaging, compared to before a rendezvous. This change in usage is attributed to the need to exchange and ground information almost instantly during a rendezvous. Implications for the design of 3G phones are discussed.

Notes

¹The work reported here began in late 1999. Text only mobile access to the Internet via Wireless Application Protocol (WAP) was introduced in 2000, but usage by the UK public for shopping and information seeking has been modest. GPRS services were introduced in 2003.

²The purpose of the work is not just to suggest more usable forms of current services, but to support the design of packages of 3G devices and services including, for example, navigation, personal information management and information seeking.

³Time Geography is a school of Human Geography that emphasises the development and use of continuous models of human activity with respect to time and space. It argues that these models are a basic component of the understanding of spatial behaviour (Haggett 2002, Gregory 2000, Pred 1996, Carlstein et al . 1978, Hagerstrand 1975).

⁴ Figure 1 is based upon a life-path diagram drawn by Lenntorp (1978). This form of diagram clearly conveys the definition of contexts of use from a Time Geographic perspective, but will probably be unfamiliar to readers, so it is briefly explained here. In life-path diagrams, two dimensions of geographic space (longitude and latitude) are represented ‘horizontally’ as an apparent surface (in this case, the map of East London suburbs at the foot of figure 1). Time is represented along a vertical axis, and so successive snapshots of geographic space combine to form an apparent column of time–space. Locales within geographic space, such as Kate and Sonja's homes, and the railway stations, are represented as many-sided pillars within the main column of time–space. Indivisible entities, such as Kate and Sonja, are represented as continuous lines, or ‘life-paths’ running through time–space (the thick, black line, snaking its way across figure 1). A stationary entity, for example, Kate at home, has a vertical life-path. A moving entity, such as Kate on the train, has a life-path that projects through time–space, its slope indicating speed of movement, and its horizontal deflection indicating change in position. An activity bar adjacent to the time axis indicates the activity of each individual, such as transiting and shopping, and the times at which these activities start and stop.

⁵Since ‘mobile’ contexts are in part defined by the locales before and after travel, it is unclear that any single mobile context is representative of mobile contexts in general.

⁶Their construct was ‘attitude’.

⁷In the future, it may be possible to use mobile devices and wireless networks to directly observe and record the behaviour of mobile participants and their context of use. Meanwhile, a combination of diary and questionnaire is a useful substitute, even if the techniques are pushed to their limits (Palen and Salzman 2002).

⁸The youth market was the first to adopt text messaging en masse www.mobilemms.com., and continues to show interest in mobile applications (www.forrester.com).

⁹An alternative questionnaire could have returned a single set of user experience ratings for each communication, together with the context in which that communication occurred, and the service used. Such a questionnaire would have been more informative (it would have been possible to directly relate ratings to communication services). However, such a questionnaire does not scale to rendezvousing. It was clear from the pilot study that the average rendezvous involved approximately four communications, and often six or more, so one set of ratings per communication would have required diary-keepers to answer 50 questions on average per rendezvous, and 70 questions quite often! Increasing the demands of participation to this level was considered likely to distort the behaviour being studied, because, for example, completing the questionnaire might take twice as long as sending the message. This study posed 37 questions per rendezvous, which is already a high number.

¹⁰Manual checks revealed 49 recognition errors made by the Reading Machine, due to smudged responses, or incompletely-erased responses. Where possible, these recognition errors were corrected by referring to the original response sheets and/or the free text in the diaries. Manual checks also identified 33 out of the 248 rendezvous entries that were spoilt due to a common error interpreting the questionnaire. When no communication occurred during a rendezvous, diary-keepers should have left the response fields to Questions 32 – 37 blank, but on these 33 occasions, Questions 32 – 37 had been answered ‘1,1,1,1,1,1’. This response makes little sense, unless, as some participants commented, participants felt obliged to mark the sheet somewhere, and there was no response entitled ‘n/a’. Ratings of the experience of communication were only included in subsequent calculations, provided at least one communication of some type had occurred in that context. Repeating statistical calculations with these 33 ‘interpretation errors’ included in the data has little effect upon statistical significance, because the total number of rendezvous is large (248).

¹¹To conduct this t test (which compares mean scores), the mean number of communications ‘during’ a rendezvous was multiplied by 1.845, to compensate for the fact that users make fewer communications during a rendezvous than before (mean = 1.49 during, compared to 2.75 before).

¹²Satisfaction was rated 4 out of 5 in both contexts. Disruption increased during the rendezvous, from 1 to 2 out of 5.

¹³Increasingly, some networks/phones now provide acknowledgements of receipt; most used in this study did not. Many networks/phones used in this study did provide acknowledgement of sending.

¹⁴An unpublished study of 16 users of a mobile phone during homeward commutes indicates that around 10% of phone use is cut short, or completed later, because of the context of use.

¹⁵This is strongly implied by the factors responsible for the overall reduction in user experience ratings. Of course, the effect of a ‘mobile’ context of use on specific services was not tested here.