Academia.eduAcademia.edu

Figure 1 – uploaded by Miguel Nacenta

See full PDFdownloadDownload figure
Figure 1. Hybrid configuration for information seeking. less interaction across opportunistically available displays.

Figure 1 Hybrid configuration for information seeking. less interaction across opportunistically available displays.

Related Figures (12)

In large data trials, the markers were spread out across the full map, which was the same size as the large screen (1920x1080), and was accessible from the mobile device through standard 2D panning. In the hybrid condition, a rectangle on the large screen dynamically represented the viewport currently shown on the mobile display (Figure 2). Figure 2. Map search task on mobile and large display (hybrid configuration). Devices not to scale.
In large data trials, the markers were spread out across the full map, which was the same size as the large screen (1920x1080), and was accessible from the mobile device through standard 2D panning. In the hybrid condition, a rectangle on the large screen dynamically represented the viewport currently shown on the mobile display (Figure 2). Figure 2. Map search task on mobile and large display (hybrid configuration). Devices not to scale.
To investigate the relationship between gaze shifts and completior time, we ran a regression test for the data in the hybrid configura. tion (the only one with gaze shifts). Trials from small and larg data were analyzed separately to avoid causing the regression tc appear significant due to the differences in task requirements. For small data, there is a linear relationship between the number of gaze shifts (#GS) and completion time (CT(seconds) =1.8*#GS+ 8.21) as shown in Figure 5. This regres- sion is statistically significant (F\;5= 6.55, p < 0.05) and explains 21% of the completion time variance (R’=0.214). For large data, the regression is very similar (CT=1.8*#GS+9.53), and also sig- nificant (F),25 = 15.01, p < 0.01), although it explains a larger por- tion of the CT variance (R7=0.385, 38%).
To investigate the relationship between gaze shifts and completior time, we ran a regression test for the data in the hybrid configura. tion (the only one with gaze shifts). Trials from small and larg data were analyzed separately to avoid causing the regression tc appear significant due to the differences in task requirements. For small data, there is a linear relationship between the number of gaze shifts (#GS) and completion time (CT(seconds) =1.8*#GS+ 8.21) as shown in Figure 5. This regres- sion is statistically significant (F\;5= 6.55, p < 0.05) and explains 21% of the completion time variance (R’=0.214). For large data, the regression is very similar (CT=1.8*#GS+9.53), and also sig- nificant (F),25 = 15.01, p < 0.01), although it explains a larger por- tion of the CT variance (R7=0.385, 38%).
(which showed low interaction), the gaze shifts in this configura- tion still mattered, making hybrid and mobile equivalently slow for different reasons.
(which showed low interaction), the gaze shifts in this configura- tion still mattered, making hybrid and mobile equivalently slow for different reasons.
Figure 6. Text search task shown on mobile and large display
Figure 6. Text search task shown on mobile and large display
Table 3. Text search: average ratings in the TLX questions. Table 4. Text search: configuration preference rankings.
Table 3. Text search: average ratings in the TLX questions. Table 4. Text search: configuration preference rankings.
An omnibus ANOVA of length of interaction with configuration and data size as main factors and participant as random factor yielded main effects of data size (F'1,21 = 25.31, p < 0.001, Ny 0.55), configuration (F2,42 = 23.24, p < 0.001, = = 0.52), as well as an interaction between the two (F2,42 = 16.39, p < 0.001, tp = 0.44). A post-hoc analysis (Tukey’s HSD) showed statistical dif- ferences in the interaction length between large display (Mgmau = 2947px, Miarge = 24815px) and the other configurations (mobile: Momall = 16px, Muarge = 192 1px; hybrid: Momalt = 12px, Muarge = 1973px), but not between mobile and hybrid. Figure 8 shows the length of interaction with different configurations and data sizes for text search.
An omnibus ANOVA of length of interaction with configuration and data size as main factors and participant as random factor yielded main effects of data size (F'1,21 = 25.31, p < 0.001, Ny 0.55), configuration (F2,42 = 23.24, p < 0.001, = = 0.52), as well as an interaction between the two (F2,42 = 16.39, p < 0.001, tp = 0.44). A post-hoc analysis (Tukey’s HSD) showed statistical dif- ferences in the interaction length between large display (Mgmau = 2947px, Miarge = 24815px) and the other configurations (mobile: Momall = 16px, Muarge = 192 1px; hybrid: Momalt = 12px, Muarge = 1973px), but not between mobile and hybrid. Figure 8 shows the length of interaction with different configurations and data sizes for text search.
Table 5. Photo search: average ratings in the TLX questions. Table 6. Photo search: configuration preference rankings.
Table 5. Photo search: average ratings in the TLX questions. Table 6. Photo search: configuration preference rankings.
answers among three 3x6 grids. Paging and input were identical to those in Experiment 2.
answers among three 3x6 grids. Paging and input were identical to those in Experiment 2.
An ANOVA test on the small data trials showed that configura- tion had no significant effect on completion time (2,38 = 2.91, p > 0.05, iis = 0.13). For the small data condition hybrid was the fast- est configuration (M=8.89s), followed by large display (M=10.35s, 16% slower) and mobile (M=12.01s, 35% slower).
An ANOVA test on the small data trials showed that configura- tion had no significant effect on completion time (2,38 = 2.91, p > 0.05, iis = 0.13). For the small data condition hybrid was the fast- est configuration (M=8.89s), followed by large display (M=10.35s, 16% slower) and mobile (M=12.01s, 35% slower).
a 9 a, Y An omnibus ANOVA of /ength of interaction with configuration and data size as main factors and participant as random factor yielded effects of data size (F1,19 = 101.49, P < 0.001, t= 0.84), configuration (F2,38 = 28.51, p < 0.001, t= = 0.60), as well as an interaction between the two (Po 38 = 21.47, p < 0.001, = 0.53). A post-hoc analysis (Tukey’s HSD) showed statistical differences in interaction length between large display (Mgmat = 872px, Miarge = 4217px) and the other configurations (mobile: Mgnay = 11px, Marge = 1115px; hybrid: Mgma = 14px, Miarge = 1132px), but not between mobile and hybrid. Figure 11 shows the mean length of interaction with small and large data. To investigate the relationship between gaze shifts and completion time, we ran a linear regression test for the data in the hybrid con- figuration which was not significant in either data size (F'1,19 = 2.34, p > 0.05; F'i,19= 1.59, p > 0.05).
a 9 a, Y An omnibus ANOVA of /ength of interaction with configuration and data size as main factors and participant as random factor yielded effects of data size (F1,19 = 101.49, P < 0.001, t= 0.84), configuration (F2,38 = 28.51, p < 0.001, t= = 0.60), as well as an interaction between the two (Po 38 = 21.47, p < 0.001, = 0.53). A post-hoc analysis (Tukey’s HSD) showed statistical differences in interaction length between large display (Mgmat = 872px, Miarge = 4217px) and the other configurations (mobile: Mgnay = 11px, Marge = 1115px; hybrid: Mgma = 14px, Miarge = 1132px), but not between mobile and hybrid. Figure 11 shows the mean length of interaction with small and large data. To investigate the relationship between gaze shifts and completion time, we ran a linear regression test for the data in the hybrid con- figuration which was not significant in either data size (F'1,19 = 2.34, p > 0.05; F'i,19= 1.59, p > 0.05).
Related topics:
Computer GraphicsHuman Computer InteractionGame DesignMusic Information RetrievalMulti-Display EnvironmentsGraphical User InterfacesMultimedia Technologies
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved