« Previous | Contents | Next »
Listen
ANNEX 4 LOG LINEAR ANALYSES
Chapter 3, Table 3.4
For this section of analysis, it was decided to use a hierarchical log-linear approach because this method is suitable for multiple categorical variables. Loglinear Analysis is a multivariate extension of Chi Square used to detect the varying associations and interactions between the variables and provide a systematic approach to the analysis of complex multidimensional tables. The log linear analyses used in this study were hierarchical and carried out using SPSS13.0 and the associations and interactions are discussed thereafter. A log linear analysis was applied to the frequency data using granted or refused EM bail (2), age group (4), number of presenting offences (4), previous offending history (4), in which court the application was made (3) and the type of court where the application was made (2). This relationship is illustrated in Table 3.4 below.
Tests that K-way effects are zero
K | DF | L.R. Chisq | Prob. |
|---|
1 | 13 | 113.017 | .0000 |
2 | 68 | 159.919 | .0000 |
3 | 182 | 252.203 | .0004 |
4 | 261 | 165.816 | 1.0000 |
5 | 189 | 4.437 | 1.0000 |
6 | 54 | .420 | 1.0000 |
Tests of PARTIAL associations
Effect Name | DF | Partial Chisq. | Prob. |
|---|
Granted EM bail by number of presenting offences by age group | 9 | 22.320 | 0.0079 |
|---|
Table 3.4 Number of presenting offences by age
Granted EM Bail | Refused EM bail |
|---|
Age groups | No of presenting offences | Age groups | No of presenting offences |
|---|
One | Two | Three | Four+ | Total | One | Two | Three | Four+ | Total |
|---|
Under 21 yrs | 16
41% | 13
33% | 8
21% | 2
5% | 39
100% | Under 20 yrs | 16
31% | 13
25% | 11
21% | 12
23% | 52
100% |
|---|
21-30yrs | 7
35% | 7
35% | 1
5% | 5
25% | 20
100% | 21-30 yrs | 18
29% | 14
23% | 10
16% | 20
32% | 62
100% |
|---|
31-40 yrs | 3
17% | 6
33% | 1
6% | 8
44% | 18
100% | 31-40 yrs | 15
43% | 9
26% | 6
17% | 5
14% | 35
100% |
|---|
41+ yrs | 5
45% | 3
27% | 1
9% | 2
18% | 11
100% | 41+ yrs | 5
42% | 3
25% | 2
17% | 2
17% | 12
100% |
|---|
Total | 31 | 29 | 11 | 17 | 88 | Total | 54 | 39 | 29 | 39 | 161 |
|---|
Chapter 3, Table 3.8
A log linear analysis was applied to the frequency data using granted EM bail, refused EM bail after reports or refused EM bail outright (3), age group (4), number of presenting offences (4), previous offending history (4), in which court the application was made (3) and type of court (2). Table 3.8 shows this relationship.
Tests that K-way effects are zero
K | DF | L.R. Chisq | Prob. |
|---|
1 | 14 | 117.234 | .0000 |
2 | 80 | 200.908 | .0000 |
3 | 238 | 319.225 | .0003 |
4 | 387 | 195.470 | 1.0000 |
5 | 324 | 4.704 | 1.0000 |
6 | 108 | .204 | 1.0000 |
Tests of PARTIAL associations
Effect Name | DF | Partial Chisq. | Prob. |
|---|
EM bail granted, refused after reports and refused outright | 4 | 23.806 | 0.0001 |
|---|
Table 3.8 Applications refused outright versus suitability reports requested
| Application refused outright | EM bail refused after reports | EM bail granted | Total |
|---|
Glasgow | 18 (17%) | 39 (37%) | 48 (46%) | 105 |
|---|
Kilmarnock | 8 (9%) | 50 (54%) | 35 (38%) | 93 |
|---|
Stirling | 49 (45%) | 26 (24%) | 33 (31%) | 108 |
|---|
Total | 75 (25%) | 115 (38%) | 116 (38%) | 306 |
|---|
Chapter 5, Table 5.5
A log linear analysis was applied to the frequency data in the comparison sample using categories of age group (4), no of presenting offences (4), previous offending history (4) and outcome of first hearing. This log linear analysis found an interaction between number of previous offences and the outcome of the first hearing where the sheriff decided if the accused should be remanded, bailed or ordained.
Tests that K-way effects are zero:
K | DF | L.R. Chisq | Prob. |
|---|
1 | 11 | 148.987 | .0000 |
2 | 45 | 133.124 | .0000 |
3 | 81 | 77.485 | .5901 |
4 | 54 | 16.761 | 1.0000 |
Tests of PARTIAL associations:
Effect Name | DF | Partial Chisq | Prob. |
|---|
No of previous offences by outcome of first hearing | 6 | 16.573 | 0.0110 |
|---|
Table 5.5 Comparison courts: Number of previous offences by outcome of first hearing
Number of previous offences | Outcome of first hearing | Total |
|---|
Remanded | Bailed | Ordained |
None | 2 (11%) | 13 (72%) | 3 (17%) | 18 |
|---|
1 - 5 | 10 (13%) | 47 (59%) | 23 (29%) | 80 |
|---|
6 - 10 | 12 (31%) | 18 (46%) | 9 (23%) | 39 |
|---|
10+ | 21 (42%) | 19 (38%) | 10 (20%) | 50 |
|---|
Total | 45 (24%) | 97 (52%) | 45 (24%) | 187 |
|---|
5.43 Table 5.5 shows that in the comparison sample those accused who have an offending history of 6+ previous offences are more likely to be remanded in custody (73%) pending trial than those who present with 1-5 previous offences (13%) or no previous offences (11%). This difference is statistically significant ( X_ = 18.879, p<0.01).
« Previous | Contents | Next »