Chapter 4: Detailed description of the dataset used in the analyses

Statistical analysis

All data entry and analyses were conducted using Statistica (version 6.1) and ClinTools Version 4.1 (Devilly, 2007). Rather than simply considering resilience in terms of PTSD symptomatology, this study will inspect the concept as comprising three life domains. An individual will be considered resilient only if they show no change or some improvement across all three domains.

The first of the domains, Health and Wellbeing, will encompass factors associated with general health and wellbeing. Measures of general health, affective distress (depression, anxiety and stress), drug and alcohol use, and relationship satisfaction will be used for assessment. Resilience in this domain is equal to either no change or improvement across time, for all four of these measures.

The second of the domains, Reactivity to Trauma, will reflect the amount of self-reported PTSD symptomatology experienced by the individual following exposure to potentially traumatic policing events. A measure of PTSD symptomatology will be used to assess this domain, and resilience will equal a score on this measure that is below a clinical cut-off point.

The third of the domains, Workplace Functioning, will reflect the individual’s ability to work and function in a stressful occupation. A measure of burnout will be used to assess this domain, and in addition the individual’s use of both police and community mental health services will be measured. Resilience in this domain will equal a score on the burnout measure that is below a clinical cut-off point.

Prior to analysis all independent variables were examined for accuracy of data entry, missing values, presence of univariate and multivariate outliers and the distribution of data. Cases with a small number of missing values (3 or less) were retained, and missing values were replaced with the individual’s mean score for the subscale of the measure where the missing value was located (when the measure contained subscales), or with the individual’s mean score for the corresponding measure. Cases that were missing a significant number of values on any one measure (more than 3) were deleted from analyses involving this measure.

Both graphic and statistical methods were used to determine whether the data was normally distributed. For variables that were not normally distributed, transformations were applied.

One outlier was detected for substance involvement at six-month follow-up, and this case was therefore removed from analyses involving this measure. A case with high scores for affective distress at six-month follow-up and two cases with high trauma symptomatology scores at six-month follow-up were also identified. Upon closer inspection, the scores for these cases were considered to be believable and were therefore retained. However, due to the low condition means for affective distress and trauma symptomatology, these scores are considered outliers. As a result, analyses involving six-month follow-up affective distress and six-month follow-up trauma symptomatology will be conducted using non-parametric tests. No multivariate outliers were identified.

Attrition, session attendance, questionnaire completion and training integrity adherence

Sample retention was high over the course of the training program, with the majority of recruits attending all five training sessions. Of the 311 participants who completed the pre-program assessment 10 failed to complete their operational training at the academy. As a result they failed to complete the resilience/control training and were excluded from the study. This represented 3 per cent of the initial sample. Of the 301 recruits who completed the training program, 280 (93%) completed all five sessions, and 21 (7%) missed one of the five sessions. Recruits who missed a session did so for reasons unrelated to the study (eg physical illness).

The completion rate of the post-program (training satisfaction) assessment was high: 287 out of the 301 recruits who completed the training program also completed the post-program assessment (95%). The completion rate of the six-month follow-up assessment was also relatively high: 281 of the 301 recruits who completed the training program also completed the six-month follow-up assessment (93%). The attrition rate from the pre-program assessment to six-month follow-up was relatively low, at 10 per cent. Only one-third of the sample was asked to complete the 12-month follow-up assessment, so the attrition rate was not calculated for this assessment time point.

Overall, for the 18 different squads that participated in the two different training programs, 60 separate training sessions were run. Of these 34 (57%) were attended by an independent assessor who rated the trainer’s adherence to the training manual. The mean training adherence rating was high, at 5.50 (SD = 0.51).

Descriptive statistics

Of the 281 participants, over half were aged under 29 years (see Table 5.1). With respect to relationship status, 57 per cent of the sample were single, while 43 per cent were in a relationship. Participants identified with several religions: 30 per cent were Catholic, 19 per cent were of other Christian denominations, 1 per cent were Muslim and 5 per cent described themselves as ‘Other’. Overall, 59 per cent of participants had some type of religious belief, compared to 41 per cent who had no religious belief.

The majority of the sample (92%) identified themselves as feeling that they were Australian, while 2 per cent reported feeling that they were Northern or Western European. Two people (0.7%) reported feeling Southern or Eastern European, two (0.7%) described themselves as Asian, one (0.4%) described themself as African, one (0.4%) as described themself as North American, one described themself as South American (0.4%) and 3 per cent of the group described themselves as ‘Other’.

The majority of participants reported having completed Year 12 or its equivalent (37%) as their highest academic attainment; 22 per cent reported having completed an undergraduate degree and 5 per cent a postgraduate degree. Twenty-four per cent of people reported having seen somebody for emotional problems before joining the police service. All descriptive statistics are detailed below (see Table 5.2).

As this study incorporates people who come from a wide range of backgrounds, a series of factorial, between-subjects multivariate analyses were conducted to determine whether these differences within the sample had any impact on the measures of interest throughout the analyses. MANOVAs were used, with Health and Wellbeing (ie general health, affective distress, substance involvement, and relationship satisfaction), Reactivity to Trauma (ie trauma symptomatology), and Workplace Functioning (ie burnout, police services accessed and external services accessed) as dependent variables. Independent variables in each analysis were (a) relationship status (single vs in relationship), (b) station location, (c) religious belief and (d) ethnicity (Australian (n=292) vs Other (n=17) due to small sample sizes of specific ethnic groups).

In each case Pillai’s criterion was used to determine the significance of the main effect of condition as it is considered more robust than other commonly used statistics and is recommended when unequal group sizes among the independent variables are being used (Tabachnik & Fidell, 2001, p.401). The combined dependent variables were found to not be significantly influenced by(a) relationship status (single vs in relationship) (Pillai’s(10,207) = 0.06, ns), (b) station location (Pillai’s (10,206) = 0.08, ns), (c) religious belief (Pillai’s (10,209) = 0.04, ns) or (d) ethnicity (Australian vs Other (F(10, 211) = 0.06, ns). Subsequently, it was considered valid to use a sample with mixed backgrounds.

Table 5.1: Age ranges of participants
Age range (years) N %
18–28 183 65.4
29–38 69 24.6
39–48 24 8.5
49–58 4 1.4
59–68 1 0.4
Table 5.2: Baseline characteristics of the sample
All conditions
Mean SD Mean SD Mean SD
Sex (m:f) 68:73 76:64 144:137
Relationship status (single:in a relationship) 78:63 82:58 160:121
Religious belief (yes:no) 76:65 90:49 166:114
Ethnicity (Australian:Other) 132:8 126:13 258:21
Highest education attained Postgraduate: 4 Postgraduate: 10 Postgraduate: 14
Undergraduate: 39 Undergraduate: 23 Undergraduate: 62
TAFE: 41 TAFE: 35 TAFE: 76
Year 12: 45 Year 12: 59 Year 12: 104
Did not complete Year 12: 10 Did not complete Year 12: 13 Did not complete Year 12: 26
Number who have seen someone
for emotional problems prior to
joining the police
38(27%) 29(21%) 21(24%)
Number of times that a professional was seen for emotional problems 3.13 4.62 3.53 5.83 3.33 5.25
Location of participants’ police station (metropolitan: rural) 120:21 119:20 239:41
Emotion-focused coping 2.45 1.31 2.43 1.28 2.44 1.29
Problem-focused coping 1.82 .64 1.88 0.75 1.85 0.70
History of traumatic events—number of personal trauma events where distress was ‘3’ or greater, at the time of the event 0.78 1.31 .78 1.38 0.78 1.34
ISEL-12 Total (‘interpersonal support’) 38.76 3.40 38.66 3.86 38.71 3.63
ISEL-12- Belonging 11.98 1.21 11.81 1.46 11.89 1.34
ISEL-12- Tangible Support 14.66 1.65 14.66 1.83 14.66 1.34
ISEL-12- Appraisal 12.16 1.59 12.12 1.43 12.14 1.51
ADAS (relationship satisfaction) 25.89 4.62 26.34 5.19 26.11 4.91
STAXI (trait anger subscale only) 16.06 3.58 15.89 3.46 15.98 3.51
LOT-R (optimism) 16.50 3.47 16.46 3.33 16.48 3.40
CD-RISC (resilience) 76.56 10.15 78.00 10.61 77.28 10.39
TIPI- Extraversion 10.23 2.65 10.16 2.63 10.19 2.63
TIPI- Emotional Stability 11.20 2.39 11.54 2.10 11.37 2.25
TIPI- Conscientiousness 11.77 2.07 11.93 2.01 11.85 2.04
TIPI- Agreeableness 10.92 1.82 10.91 2.10 10.91 1.96
TIPI- Openness to Experience 11.06 1.92 10.98 2.13 11.02 2.03
DASS total score 12.10 12.66 11.22 11.69 11.67 12.18
ASSIST total score 18.83 16.42 18.14 16.19 18.49 16.28
SF-36 total score 84.27 10.96 86.39 8.73 85.33 9.95
Group size, number of participants—M (range) 15.7 (11–21) 15.6 (12–21) 14.83 (12–18)

No significant differences were found between the conditions for any of the main measures at baseline. Thus, the two conditions were considered to be equal at the pre-program assessment time point.

The internal reliability of each of the measures used in the study across each time point was calculated using Cronbach’s Alpha. The majority of the coefficients were within acceptable parameters and were above the accepted level of 0.70 (Tabachnik & Fidell, 2001).