Table 11: Observed and expected frequencies for Log-linear model of unsafe acts
Skill-based error
|
Decision error
|
Perceptual error
|
Violation
|
Observed count
|
Observed per cent
|
Expected count
|
Expected per cent
|
Residuals
|
Standardised Residuals
|
Yes
|
Yes
|
Yes
|
Yes
|
3
|
0.10
|
3.11
|
0.10
|
-0.11
|
-0.06
|
|
|
|
No
|
10
|
0.20
|
13.73
|
0.30
|
-3.73
|
-1.01
|
|
|
No
|
Yes
|
34
|
0.80
|
37.80
|
0.90
|
-3.80
|
-0.62
|
|
|
|
No
|
240
|
5.80
|
232.37
|
5.60
|
7.63
|
0.50
|
|
No
|
Yes
|
Yes
|
1
|
0.00
|
0.88
|
0.00
|
0.12
|
0.13
|
|
|
|
No
|
26
|
0.60
|
22.33
|
0.50
|
3.67
|
0.78
|
|
|
No
|
Yes
|
32
|
0.80
|
27.96
|
0.70
|
4.04
|
0.76
|
|
|
|
No
|
981
|
23.60
|
988.79
|
23.80
|
-7.79
|
-0.25
|
No
|
Yes
|
Yes
|
Yes
|
4
|
0.10
|
4.12
|
0.10
|
-0.12
|
-0.06
|
|
|
|
No
|
10
|
0.20
|
6.04
|
0.10
|
3.97
|
1.61
|
|
|
No
|
Yes
|
15
|
0.40
|
11.00
|
0.30
|
4.00
|
1.21
|
|
|
|
No
|
172
|
4.10
|
179.84
|
4.30
|
-7.84
|
-0.58
|
|
No
|
Yes
|
Yes
|
4
|
0.10
|
3.90
|
0.10
|
0.10
|
0.05
|
|
|
|
No
|
29
|
0.70
|
32.88
|
0.80
|
-3.88
|
-0.68
|
|
|
No
|
Yes
|
23
|
0.60
|
27.24
|
0.70
|
-4.24
|
-0.81
|
|
|
|
No
|
2571
|
61.90
|
2563.01
|
61.70
|
7.99
|
0.16
|
The associations can be dissected by examining cross tabulations and odds-ratios.
The model fit was acceptable and the residuals were consistent throughout the entire table (Table 11). Aside from the cases where no unsafe acts occurred, the most common unsafe acts were skill-based errors alone (981), decision errors alone (172), and combined skill-based and decision errors (240). It is also of interest that violations frequently co-occurred with skill-based errors, decision errors, or a combination of both. Of the 93 violations that co-occurred with an error (skill, decision or perceptual), 81 of these violations co-occurred with a decision error, skill-based error or both.
Table 12: Chi-square analysis of inadequate supervision and supervisory violations
|
Supervisory violations
|
No supervisory violations
|
Total
|
Chi-square
|
Significance
|
Inadequate supervision
|
2
|
85
|
87
|
22.8
|
<0.001
|
No inadequate supervision
|
6
|
4462
|
4468
|
Total
|
8
|
4547
|
4555
|
|
|
The relationship between perceptual errors and skill-based errors varied with the presence or absence of a violation (Table 13). When violations were absent, the presence of a perceptual error increased the probability of a skill-based error by two times. When a violation was present, a perceptual error reduced the chance of a skill-based error occurring.
Table 13: Three-way association of unsafe acts: skill-based errors, perceptual errors and violations
-
|
|
|
Perceptual error
|
Total
|
Odds ratio
|
|
|
|
Yes
|
No
|
|
|
Violation
|
Skill-based error
|
Yes
|
4
|
66
|
70
|
0.3
|
No
|
8
|
38
|
46
|
Total
|
12
|
104
|
116
|
|
No violation
|
Skill-based error
|
Yes
|
36
|
1221
|
1257
|
2.1
|
No
|
39
|
2743
|
2782
|
Total
|
75
|
3964
|
4039
|
|
There were also two 2-way interactions: decision errors by violations, and skill-based errors by decision errors (Table 14). Violations had 7.8 times higher odds of occurring if there was a decisional error than if there wasn’t a decision error. Skill-based errors had 3.6 times higher odds of occurring when there was a decisional error than when it was absent.
Table 14: Odds ratio of decision error and violations and decision error and skill-based error
|
Decision error
|
No decision error
|
Total
|
Odds ratio
|
Violation
|
56
|
60
|
116
|
7.8
|
No violation
|
432
|
3607
|
4039
|
Total
|
488
|
3667
|
4155
|
|
Skill-based error
|
287
|
1040
|
1327
|
3.6
|
No skill-based error
|
201
|
2627
|
2828
|
Total
|
488
|
3667
|
4155
|
|
Share with your friends: |