CCITT COMMISSION D'ETUDES I Contribution tardive) ET SES GROUPES DE TRAVAIL Delayed Contribution) Contribucion tardia ) STUDY GROUP I AND ITS WORKING PARTIES Texte disponible seulement en) COMISION DE ESTUDIO I Text available only in Y SUS GRUPOS DE TRABAJO Texto disponible solamente en) Geneve, 28 mai - 7 juin 1991 Question: 28/I SOURCE: AT&T TITLE: Performance, Preference, and Usage for Q and Z on 10-Digit Keypads 1. Introduction Ten-key pads are used world-wide for telephones, Automatic Teller Banking Machines (ATMs), and Point of Sale terminals (POSs). Both CCITT and ISO subcommittees are currently considering standardization of the placement of Latin letters on 10-key pads. To date, CCITT SGI, ISO/IEC/JTC1/SC17, and ISO/IEC/JTC1/SC18/WG9 have agreed that a common international standard should affirm the placement of the 24 letters as they currently appear on telephones, ATMs, and POSs. This telephone layout is illustrated in Figure 1a. However, the letters Q and Z do not currently appear on telephones, and there has been much recent discussion of where they should appear on 10-key pads in an International Standard. The two possibilities that have received the most consideration to date have been 1) placement of Q and Z on the 1 key (Q/Z=1) and 2) placement of Q on the 7 key and Z on the 9 key in alphabetic order. These assignments are illustrated in Figures 1b and 1c. This contribution summarizes results of two studies conducted to determine if there are usage, preference, or performance differences that would support one of these arrangements over the other: A laboratory study examined performance and preference. A national preference and usage study among active ATM and calling card users collected - 2 - information on PIN usage and on preferences for the two layouts. The following summarizes the design of the studies and their main results. 2. Laboratory_Study Subjects keyed in letter strings on DTMF telephone keypads using each of the two mapping schemes. Three keypads were used in the study, as illustrated in Figures 1a-c. Half of the subjects used standard phones without Q and Z marked on the keys. The others used telephones modified to include Q and Z on the keypad, half the time using the Q/Z=1 arrangement, half the time the Q=7/Z=9 arrangement. A total of 64 people participated in this study. Each subject dialed names, 800 "vanity" numbers, and four-letter passwords. There were 32 strings of each type, and half of the strings of each type contained the letters Q or Z. The letter strings to be keyed in were presented one at a time in large letters on a computer monitor. Presentation of lists and mapping schemes were appropriately counterbalanced. 2.1 Results Both performance and preference data were collected. Performance was measured by the number of errors subjects made in keying a letter string and by the amount of time it took to enter the letters. Preference ratings were collected after subjects used each mapping scheme, as well as after the experiment. Errors An error was recorded whenever the digits dialed did not exactly correspond to the letter string presented to the subject. In addition, any time a subject paused longer than 5 seconds between digits, that trial was recorded as an error. Table 1 shows the errors for the two mapping schemes for marked and unmarked phones. The results show no statistically significant differences in errors made with the two mapping schemes.1 - 3 - Table 1. Number and Percentages of Errors ___________________________________ | Marked Phones| Unmarked Phone| |______|_________|_______|_________| | Q/Z=1| Q=7/Z=9| Q/Z=1| Q=7/Z=9| |______|_________|_______|_________| | 134| 133 | 156| 130 | | 17%| 17% | 20%| 17% | |______|_________|_______|_________| Keying Time Total keying time was calculated by adding the times from first to last keypress and the time from stimulus presentation to first keypress. Only correctly keyed strings were included in the timing data. Table 2 shows the mean keying time in seconds for the two QZ mapping schemes for the marked and unmarked telephones. Users keyed slightly faster on marked phones with the Q=7/Z=9 mapping, but on the unmarked phones using the Q/Z=1 mapping. The differences found here are small (on the order of half a second or less) compared to the average time it takes to key in a string (11.7 seconds). Table 2. Means and SDs for Total Keying Time (in Seconds) __________________________________________ | Marked Phones | Unmarked Phones| |_______________________|_________________| | Q/Z=1 Q=7/Z=9| Q/Z=1 Q=7/Z=9| |_____|_________________|_________________| | Mean| 12.0 11.3 | 11.5 12.0 | | SD | 2.8 2.2 | 1.6 2.7 | |_____|_________________|_________________| Preference Ratings At the end of the experiment, subjects were asked which of the QZ mapping schemes they preferred. Table 3 shows the number of subjects preferring each arrangement. Overall, significantly more of the subjects preferred the Q=7/Z=9 arrangement (61% vs 39%).2 __________ 1. By a chi square test, X squared = 0.88, p = .35. 1. Binomial test, p =.05. - 4 - Table 3. User Preference for QZ Mapping Scheme ___________________________________ | Marked Phones| Unmarked Phone| |______|_________|_______|_________| | Q/Z=1| Q=7/Z=9| Q/Z=1| Q=7/Z=9| |______|_________|_______|_________| | 12 | 20 | 13 | 19 | | 38%| 63% | 41%| 59% | |______|_________|_______|_________| 3. Field_Study The objectives of this study were to: o Assess peoples' preferences for the two Q/Z mapping schemes and the reasons for their preferences. o Assess the incidence of choosing personal identification numbers (PINs) based on letters and numbers, and of mapping assigned numeric PINs to letters. o Assess the incidence of dialing letters on telephones and ATMs as well as awareness of various situations involving alphabetic entry from 10-key pads. o Understand how people currently choose PINs. 3.1 Design Mall intercept interviews were conducted with 400 people in four national regions. Two hundred interviews were conducted with ATM cardholders and the same number with telephone calling card users. To ensure familiarity with PINs and calling card use, only respondents who reported at least weekly use of either card were included in the study. Interview protocols began with basic screening questions. Respondents were then shown one of two otherwise identical telephones that differed only in the placement of the letters Q and Z and asked to dial a small set of names, PINs and alphabetical telephone numbers using the dialpad. They then rated the telephone overall and on several specific dimensions. Similar use and ratings of a telephone with the other Q/Z placement, and then ratings comparing the two telephone layouts, followed. One-half of the participants saw each of the two layouts first. Further questions assessed card usage, PIN selection, awareness of alphabetic dialing situations, actual alphabetic dialing usage, and demographic variables. - 5 - 3.2 Results Preferred Mapping Scheme Overall, respondents preferred the Q=7/Z=9 letter placement over the QZ=1 placement, by 83% to 17%. This preference for the alphabetical order mapping scheme was stable across individual ratings for each telephone; comparative ratings after experience with both telephone alphabetical layouts; component ratings of the two layouts for ease of letter entry, speed of dialing letters, and accuracy of dialing letters; and for simple direct preference judgments. Reasons for Preference and Comments Of those respondents who preferred the alphabetical order layout (N=331), 81% mentioned basing this preference on something to do with alphabetical order, particularly that Q and Z are easier to find when they appear in their logically ordered positions. Of those who preferred the Q/Z on 1 layout (N=67), legibility and ease of finding Q and Z with them on the 1 key were the two most common reasons for preference. Spontaneous comments by respondents on the Q/Z=1 layout indicated that 25% found it hard to find Q and Z and 18% couldn't find Q at all. Incidence of Dialing Letters Awareness and usage of dialing letters on telephone dials and ATM pads was high for this sample of respondents. Two-thirds of the sample said they had ever dialed letters on a telephone, and 62% do so at least once a month. One-third of the ATM users report ever entering letters from an ATM 10-key pad. Between one-half and one-quarter of the respondents were familiar with alphabetical entry for 800 telephone numbers, bank card PINs, or calling card PINs. Incidence of Choosing/Assigning PINs with Letters Respondents were asked to provide generic information about their current ATM and/or calling card PINs. Approximately 20% of calling card owners have chosen their own PIN; nearly 64% of ATM card owners have chosen their own PIN. One-third of the ATM card owners who have chosen their own PINs chose an alphabetic PIN. This means that about 20% of ATM card owners chose aphabetical PINs. For calling card owners, this number is about 6%. Interestingly, approximately 13% of those who use an assigned numeric PIN use a word or letters to remember it. Very few people would be affected by either placement of Q/Z on 10-key pads in terms of the embedded base of alphabetical PINs. One-percent of the entire sample of active card users reports having a Q or Z in their PIN. - 6 - 4. Summary The purpose of the studies reported here was to determine if people's performance in keying in strings of letters and/or their current usage or preferences provide a basis for choosing between the Q/Z=1 and Q=7/Z=9 10-digit keypad letter-mapping arrangements. The performance results (errors and keying time) did not provide a basis for choosing one arrangement over the other. Preference results, however, support the choice of the Q=7/Z=9 arrangement. Performance measures - errors and keying time - showed only small differences between the two alternatives.