Power Analysis

For the ethics application a power analysis was conducted using a priori sample size calculator for multiple regression. The current study used an alpha level of .05, statistical power level of .8, and an anticipated effect size of .35. Due to the scarcity of previous research in this area, the current study could not be based on previous effect sizes. Following the calculations, a minimum sample size of 43 participants was aimed for. However, the final number of completed valid questionnaires fell slightly short of this with a total of 36 participants. Every effort was made to recruit as many participants as possible over a period spanning 2 years. However, a low number of participants is not unusual in TBI caregiver studies (Douglas & Spellacy, 2000; Hanks et al., 2007; Harris et al., 2001; Knight et al., 1998).

7.4.3 Data Collection

Once ethics approval was obtained, information packs were sent to the participating organisations. These packs included an information sheet outlining the study and the requirements for participation, a request to participate and a stamped, pre-addressed return envelope (see Appendix C). In total about 300 information packs were posted. Participants who responded by returning the request to participate were sent a self- report questionnaire with a stamped, pre-addressed return envelope. Of the 49 participants requesting a questionnaire, 41 returned it. Of those, 3 did not fit the criteria (hypoxia, near drowning, medical mishap), one was incomplete, and one was excluded because of language difficulties (participant was Indian). Of the eight questionnaires that were not returned, two potential participants said it was too stressful and one did not complete it because her husband with TBI had passed away the year before. Completed, valid questionnaires were received from 36 caregivers, yielding a response rate of 73%.

64 Participants were recruited from throughout New Zealand and all caregivers were contacted with the aid of a professional community service. Neither the agencies nor the participants received any financial compensation for their participation or assistance. Note that the participants are not a representative sample of caregivers of a family member with TBI and any subsequent results cannot be generalised to all caregivers in this population.

65 CHAPTER 8 – RESULTS

The aim of the study was to examine the relationships between coping strategies, anxiety and depression and resulting QoL in caregivers looking after a family member with TBI. This chapter presents the results of the data analyses for the eight stated research questions.

8.1 Missing data

Missing data was handled in the following way: No data was missing in the (a) general questionnaire, (b) BDI-II or (c) BAI. (d) WHOQOL-BREF: According to the scoring instructions for the WHOQOL-BREF, where more than 20% of data is missing from an assessment, the assessment should be discarded (Harper & Orley, 1996). This was not the case in the data collected for this study. The only missing item was question 21 “How satisfied are you with your sex life” (missing three times). According to the scoring instructions for the WHOQOL-BREF the missing items were assigned a value using horizontal mean substitution calculated from the mean of that domain (Murphy et al., 2000; Power et al., 1998) (e) WCQ: The only missing data for this instrument was that one participant did not complete it, stating that there were no difficult or troubling situations with the family member in the previous week. As this was comparable with another participant’s answer for scoring the complete questionnaire with zero (did not use any of the coping strategies) this questionnaire was also included with a scoring of zero on each item.

8.2 Data Management

All data was analysed using the computer-based statistical program SPSS (version 17). An alpha level of 0.05 was used to determine statistical significance. In this study, the analyses were conducted in three stages.

• First, descriptive data were generated to provide information on the characteristics of the participants and their family member (see Tables 1 and 2 in the method section). The sample is described using means and standard deviations for the continuous variables, while frequencies and percentages are used for the categorical variables.

• Then the means, standard deviations, frequencies, and percentages (where appropriate) were calculated for each questionnaire.

66 •The relationships between dependent and independent variables (coping strategies, depression, anxiety, QoL domains, and years since injury) were examined using correlational analyses. After inspection of data for linearity, Pearson’s correlation coefficient was used for all correlations unless stated otherwise. However, due to the small sample size, all correlations were also conducted using Spearman’s rho.

The presentation of the findings is arranged first by the individual parts of the questionnaire and then by the eight research questions.

8.3 Questionnaires

8.3.1 General questionnaire

The descriptive data for this questionnaire are shown in the method section.

8.3.2 Ways of Coping Questionnaire – Revised (WCQ) (Folkman, 2008; Folkman & Lazarus, 1985; Folkman, Lazarus, Dunkel-Schetter et al., 1986)

Raw scores were calculated, which represent the sum of the items. High raw scores indicate that a person frequently uses the behaviours described in that scale (Folkman & Lazarus, 1988b). Then, relative scores were calculated as described by Vitaliano, Maiuro, Russo, & Becker, (1987). The relative scoring method has the advantage of controlling for the unequal numbers of items within the coping scales and for individual differences in response rates. This method of scoring is supported by the developers of the measure (Folkman & Lazarus, 1988b; Folkman et al., 1987). The subscales were divided into two variables: problem-focused coping and emotion-focused coping. This grouping is consistent with the two dimensions empirically derived from the original Ways of Coping Checklist (WCC) (Folkman & Lazarus, 1980). The two variables were used to look for differences in coping. As the current study examined the proportion of problem-focused coping and emotion-focused coping rather than the frequency of the coping strategies used, any findings are reported using the relative scores or the percentage of coping strategies.

Overall participants used 53% emotion-focused coping and 42% problem-focused coping3. This indicates that the participants in this study used emotion-focused coping

3 _{Two questionnaires were scored as zero as the participants did not have any stressful situations with} their family member in the previous week and stated they did not use any of the coping strategies. In

67 strategies slightly more than problem-focused coping strategies. However, it needs to be considered that there are five emotion-focused coping subscales compared to three problem-focused coping subscales. When examining the individual subscales, two problem-focused coping subscales were used the most: planful problem solving (17%) and confrontive coping (14%), and the third (seeking social support 10%) was used more than three out of five emotion-focused coping subscales. This suggests that individually there was a higher use of problem-focused coping than emotion-focused coping.

Table 5 provides the raw and relative mean scores, standard deviations, and percentages for the participants on each domain of coping as well as the overall scores for problem-focused and emotion-focused coping as measured by the WCQ.

Table 5

Means and standard deviations for raw scores and relative scores on the WCQ

Raw Score Mean (SD)

Relative Score Percentage Problem-focused coping 18.39 (11.41) 42%

Seeking social support 5.00 (4.61) 10%

Planful problem solving 7.92 (5.28) 17%

Confrontive coping 5.47 (4.10) 14% Emotion-focused coping 26.14 (15.97) 53% Distancing 5.69 (3.62) 13% Self-controlling 7.67 (4.97) 13% Accepting responsibility 3.00 (2.74) 9% Escape-avoidance 5.64 (4.56) 9% Positive reappraisal 4.94 (4.65) 8%