The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
925 
projected images. According to Donner and Cunningham (1984) 
trials with limited participants in a study the throw accuracy 
values can be mixed for statistical analysis. Therefore the 
values were separated into two groups: 1) accurate throw for 
the top 40%; and, 2) inaccurate throw for the bottom 40%. The 
remaining 20% of the middle values were discarded. The 
results based on the grouping are presented in Table 1. 
Throw Distance from target (m) 
Condition 
Accurate group 
(n=25) 
Inaccurate group 
(n=25) 
Mean 
Mean 
At the 
front 
0.530 
1.091 
At the 
middle 
0.744 
2.784 
At the back 
1.611 
8.441 
Table 5. Accuracy scores for all throws collected, across all 
throw conditions. 
3.2.2 Lineout thrower’s motion 
Details of the results and analyses of all movements measured 
can be found in Croft (2007). However, some significant 
findings are provided below. 
i) Front foot step distance 
Figure 11 shows the scatter plot of data spread for the front 
foot step distance over a short throw (jumper at the front 
position) condition for the accurate and inaccurate groups. It is 
apparent that there were more outliers coming from the 
inaccurate group. Also it is evident that on the average accurate 
throwers have longer front foot step distances for the long 
distance throw (Figure 12). A Kolmogorov-Smimov test for 
normality and a square root function to normalise the front foot 
step distance residuals were carried out. The displacement that 
the thrower’s front foot travelled during the first two phases of 
the throw was then compared with a two-way ANOVA for 
accurate and inaccurate groups over the three distance 
conditions (Figure 12). 
The clinical significance test found a 100% probability of a 
clinically positive result (Hopkins 2001). The interaction 
between the accuracy and distance independent variables was 
also significant (F=5.18, P<0.007). This interaction can be 
observed in Figure 12. The figure clearly shows that accuracy 
is also affected by an increase in distance and distance of throw 
(the column length between accurate and inaccurate group in 
the ‘back’). 
Frmf fcfiiMlf* Rartr 
Figure 12. Mean and standard error of the front foot step 
distance for all throw distance conditions. 
ii) Trunk flexion 
Figure 13 shows a scatter plot of the data spread for trunk 
flexion over short throw conditions for the accurate and the 
inaccurate groups. The throw accuracy for the first group 
(between 0.25 and 0.64 m) is significantly higher than the 
second group (0.85 and 1.5 m). However, the majority of the 
groups show a trunk flexion of 4.5 to 7.5 degrees. Figure 14 
shows the mean scores and the standard error the mean scores 
of the angle of trunk flexion for all throw distance conditions. 
For the front throw the size of the angle of trunk flexion 
between the groups was insignificant but the value differs by 
more than one degree for the middle and the back throw 
conditions. The absolute mean angle was compared using a 
two-way ANOVA for accurate and inaccurate groups over the 
three distance conditions. There was a statistically significant 
difference (F/=6.69, PO.Oll) between accurate (5.9 ± 0.2°) 
and inaccurate groups (6.7 ± 0.2°) across all three throw 
distances (Figures 13 and 14). A clinical or practical 
significance test presented mixed results with a 51% 
probability of a clinically trivial result, with 40% for clinically 
positive and 9% for clinically negative results. 
♦ Aceraie 
♦ mm 
©■© 9<* % 
Iiiiw* »! ««•» tmt 
Figure 11. Scatter plot of data spread for front foot step 
distance over a short throw (at the front position) condition for 
accurate and inaccurate groups. 
The accurate group had a significantly (F/=21.5, PO.OOO) 
higher score (177 ± 14 mm) than the inaccurate group (82.5 ± 
13.7 mm) across all three throw distances. Bonferroni /-tests 
(P<0.0167) were run showing significant differences at the 
middle (7=4.17, PO.OOO) and long (7=4.06, PO.OOO) throw 
conditions between the accuracy groups. 
c 
I 
> 
***♦*♦ 
* Amwali? 
© W ©A ©4 ©# t It M I* 
ttozutw# ffcw© tetpet mit 
Figure 13. Scatter plot of data spread for trunk flexion over 
short throw condition for accurate and inaccurate groups. 
Although there was a significant result for the accuracy 
condition the Bonferroni /-tests (PO.0167) indicated no 
significant differences between the accuracy groups at different 
throw conditions (Figure 14). This was possibly because the P- 
value for significance was reduced from PO.05 to PO.0167 to 
avoid Type I errors.