3. METHODOLOGY
3.4 Instruments
3.3.3 Other Demographics
Age of the sample participants was between 17 and 50 years of age. The majority (n =261;
75.4%) of the sample fell into the 17–21-year age category.
Home language: First-language English speakers comprised the largest language group (n=147;
42.5%), followed by first-language Zulu-speakers (n=143; 41.3%). Other languages reported included Xhosa, Afrikaans, and other African and European languages.
Occupation: Students made up the majority of the sample (n=335; 96.8%). Other occupations were reported mostly as temporary or part-time jobs over and above being students, including waiter, teacher, coach, sales clerk, librarian, and research assistant.
Course: The largest proportion (n=80; 23.1%) of the sample were BSocSci degree students, followed by students in the SFP (Science Foundation Programme) courses (n=79; 22.8%), and thereafter variously distributed between BA, BCom, Psychology, Law, BSc and BAgric students.
The majority of the participants were in first year (n=165; 47.7%) and unmarried (n=335;
96.8%).
(Benatar, 2002; Kent, 1996; Lindegger & Richter, 2000; Meisel & Roth, 1983), to date, very little work has been done on the principle of autonomy as it is applied in informed consent procedures in South Africa. As outlined in the literature review above, autonomy does not equate with voluntariness, that is, with competence, understanding, and freedom from coercion.
Perceptions and experiences of autonomy extend beyond these checklist criteria, and develop from a web of inter-related factors, among them, selfhood, culture, gender, social and historical influences. The measurement of autonomy, therefore, is not straightforward, and requires instrumentation that will assess as many of the aspects that constitute personal autonomy as possible.
An extensive search of the empirical literature on autonomy was conducted and a number of autonomy scales were found, using gender, self, agency, culture, morality, and ethics/informed consent as the major parameters. It was interesting to note that many of the more relational discussions and assessments of autonomy came from the nursing literature, while review of the literature revealed more frequent treatment of autonomy as voluntariness according to standard criteria of principled autonomy in informed consent. This could be because nurses work more in the divide between the principles of medical ethics and the practice of patient care, creating a need for a more relational form of ethical principles and practices.
The assessment of autonomy in the South African context is problematic, because no appropriate measures of autonomy were found for application to the South African population with adequate validity. In choosing instruments, the primary aim was to measure autonomy as it is understood and practiced in mainstream bioethics vs. relational autonomy as proposed by, among others,
feminist scholars. The main factors being measured were the selfhood dimension of autonomy – independent vs. relational self identity – and the care and justice orientations of autonomy, using gender as the independent variable. Autonomy- and self-related constructs from a number of empirically validated scales were reviewed in order to identify the most appropriate measures for the purposes of this study. An extensive review of the literature yielded several possible scales that could potentially be used in combination to measure the relevant aspects of autonomy, self, and moral orientation. From the 65 measures that were examined, the following 8 instruments were subsequently considered more closely for inclusion in this study: the Autonomy Scale (Bekker, 1993); Autonomy, the Caring Perspective (Boughn, 1995); the Relational-
Interdependent Self-Construal Scale (Cross et al., 2000); the Moral Justification Scale (Gump et al., 2000); the Relational Being Scale (McChrystal, 1994); the Relationship Self Inventory (Pearson et al., 1998); the Self-Construal Scale (Singelis, 1994); and the Moral Orientation Scale (Yacker & Weinberg, 1990).
3.4.1 Pilot Instruments
Ultimately, the final instrument had to include a measure of independent versus relational autonomy; a measure of independent versus relational self; and a measure of justice versus care orientations. Of these, the following were chosen to measure autonomy – Bekker (1993), Boughn (1995) – the following to measure relational / independent aspects of the self – Cross et al.
(2000), McChrystal (1994), Pearson et al. (1998), and Singelis (1994) – and the following to measure the justice and care moral orientations – Gump et al. (2000) and Yacker and Weinberg (1990). (See Appendix A for a comparison of these scales). The authors of each scale were
contacted to explain the purpose and goals of the research and to request permission to use their scales in this study. Every author responded positively and granted permission for their scales to be used in this research. Further correspondence was entered into with some of the authors concerning subsequent studies that they had conducted using their scales, and providing valuable insights or comments on the proposed research. After further examination the measures that were included in a preliminary pilot study were Bekker‟s (1993) Autonomy Scale; Cross et al.‟s (2000) Relational-Interdependent Self-Construal Scale; and Pearson et al.‟s (1998) Relationship Self Inventory.
During the pilot study (N=52), it became clear that the Autonomy Scale (Bekker, 1993) and the Relationship Self Inventory (Pearson et al., 1998) were not the most suitable measures for inclusion in a final questionnaire. Participants in the pilot reported that the statements of the Autonomy Scale were vague and confusing, possibly as a result of the translation of this scale from Dutch to English. The results generated by this scale were also unsatisfactory as they were inconsistent and their reliability and validity questionable. It was also found that the Relationship Self Inventory was too long (60 items) and reportedly tedious to answer; it was thus not included in the final questionnaire given the time constraints in asking participant students to complete the instrument in an allocated amount of time.
Based on the feedback and results from the pilot study, available instruments were reconsidered.
Length of the scale was an important consideration, as was simplicity of language. In the final elimination, three self-report questionnaires were included to assess autonomy, self-other orientation, and moral orientation: McChrystal‟s (1994) Relational Being Scale; Cross et al.‟s
(2000) Relational-Interdependent Self-Construal Scale; and Yacker and Weinburg‟s (1990) Moral Orientation Scale. These instruments are discussed in further detail below.
3.4.2 Relational Being Scale
3.4.2.1 Description
The Relational Being Scale (McChrystal, 1994) is based on the Stone Center‟s self-in-relation theories – that relational beings develop in and through a matrix of relationships with, rather than through separation from, other people (McChrystal, 1994). The Relational Being Scale (RBS) (see Appendix C) is a self-report, visual analogue scale that was developed to quantitatively measure the qualities of relatedness and autonomy as defined by Gilligan (1982), Miller (1986, 1990), Surrey (1991) and their colleagues at the Stone Center. Comprising 28 items in total, the RBS has two subscales: the Autonomy subscale (A) with 13 items, and the Relational subscale (R) with 15 items. R subscale items were devised using key concepts from Relational Being theory – “the maintenance of relationships over adherence to abstract concepts of justice;
definition of self; the theory of human development; the concept of the ideal person; the capacity for empathy; psychopathology and psychotherapy practice” (McChrystal, 1994, p. 5). These concepts were formulated into statements which required participants to consider their opinions of themselves in their responses. Items for the A subscale were inferred from the concepts generated from the work on relatedness.
3.4.2.2 Reliability and validity coefficients
In the original study, total scale alphas and item total correlations were 0.68 for the A subscale and 0.77 for the R subscale. The interscale correlation (-0.18) indicated no correlation between the two scales (McChrystal, 1994). No reliability data were given by the authors.
3.4.2.3 Administration and scoring
The RBS is a visual analogue scale: each statement is followed by a 9cm line, with „very
accurate‟ at the left end of the line and „very inaccurate‟ at the right end of the line. Respondents are asked to make a cross at the point which most accurately reflects the accuracy of the
statement as it applies to them. A ruler is used to score each item – it is placed along the line and the centimeter measurement (of 1-10) where the participant has marked his /her cross is given as the score. The lines are measured from right to left.
3.4.3 Relational-Interdependent Self-Construal Scale
3.4.3.1 Description
The Relational-Interdependent Self-Construal (RISC) Scale (Cross et al., 2000) developed out of the growing concern that Western, individualistic assumptions of personhood dominate much of the research on the self, while the connection of the self to others has largely been ignored (Cross
& Madson, 1997). More recently researchers have begun to recognize the importance of others in the self-identities of many people, particularly women, while individual differences in the self- construal have been shown to explain some of the differences women and men‟s behaviours (Markus & Oyserman, 1989; Surrey, 1991). The RISC Scale (see Appendix C) was developed by
Cross et al. (2000) to measure the tendency to define self in relation to others, identified as the relational-interdependent self-construal. Items were generated from concepts related to relational forms of the interdependent self-construal as defined in the literature, as well as from
modifications of conceptually related measures. It was found that individuals with high RISC Scale scores were more committed to and placed greater importance on their close relationships, and were more likely to take the needs and opinions of others into account when making
decisions (Cross et al., 2000). The RISC Scale was subsequently examined in three separate studies: a validation study; a study of the role of relationship considerations in decision-making;
and an investigation into the association between the relational-interdependent self-construal and relationship development strategies (Cross et al., 2000). In the original studies, women
consistently scored higher on the RISC scale than men did.
3.4.3.2 Reliability and validity coefficients
The original validation study found the RISC to be a relatively stable self-report measure of individual differences in the relational-interdependent self-construal construct (Cross et al., 2000). Factor analysis revealed that the scale is underscored by a single factor, while reliability tests showed the scale to have high internal consistency, convergent, discriminant, and criterion validity, and good test-retest reliability. Coefficient alpha for the original study (averaged across the sub-samples) was 0.88.
3.4.3.3 Administration and scoring
The RISC Scale consists of 11 questions which require subjects to rate self-other attitudes according to a „strongly agree‟ (score = 7) to a „strongly disagree‟ (score = 1) forced choice
format (Likert-type scale). Total scores range from 11 to 77. Two of the eleven items are negatively phrased and their scoring is reversed in the data analysis.
3.4.4 Moral Orientation Scale
3.4.4.1 Description
The Moral Orientation Scale Using Childhood Dilemmas (hereafter referred to as the MOS) is an objective test developed by Yacker and Weinberg (1990) to measure two distinct moral
orientations as outlined in the work of Kohlberg and Gilligan. Concepts underlying the MOS were based on the hypothesis that individuals showing a stronger care orientation or ethic place greater emphasis on responsibility towards others and the preservation of relationships; those showing a greater tendency towards a justice orientation emphasize individual rights over
relationships (Yacker & Weinberg, 1990). The MOS consists of 12 moral dilemmas that children (aged 8-10) typically face in their daily lives (see Appendix C). Although the scale was designed to measure adult moral orientation, childhood dilemmas were used in the assessment as they are relatively simple and universal, as opposed to the moral dilemmas that adults might face (Yacker
& Weinberg, 1990). The childhood moral dilemmas were formulated in consultation with child development specialists, and were based on published and unpublished materials including curricula, moral judgement interviews, popular child-rearing texts, and interviews with parents (Yacker & Weinberg, 1990).
The MOS consists of two subscales: a Justice subscale (J) and a Care subscale (C). As hypothesized in the original validation study, there was a significant gender difference on the
scores of the MOS, with men showing a stronger tendency towards a justice orientation and women showing a stronger tendency towards a care orientation.
3.4.4.2 Reliability and validity coefficients
The MOS was found to provide a valid assessment of preferred mode of moral thinking (Yacker
& Weinberg, 1990). Initial validation showed the scale to have stable discriminant validity and good test-retest reliability (0.71). No other reliability data was provided in the validation study.
3.4.4.3 Administration and scoring
As mentioned above, the MOS consists of 12 childhood dilemmas that require respondents to imagine that they are helping an 8-to 10-year-old child decide what to do in each situation. Each dilemma is followed by four choice alternatives that respondents must rank from 1 to 4,
according to their preferences for choosing each consideration in helping a child decide what to do. Without being identified as such, two of the four choices presented with each dilemma are defined within the justice mode of moral reasoning, and two are framed within the care mode (Yacker & Weinberg, 1990). Only the first choice given for each dilemma is scored in the final analysis: a respondent‟s total score on the Care subscale is calculated by adding the number of care responses selected as first choices; the number of justice responses selected as first choices are added to obtain the total score on the Justice subscale. Scores may therefore vary from 0 to 12, with higher scores on the Justice and Care subscales indicating a stronger orientation towards justice and care respectively. In order to avoid falsely dichotomizing moral thinking, the authors of the MOS did not designate cut-off scores for the scale, in line with Gilligan‟s own findings
that individuals do not exhibit one or the other type of moral orientation, but rather stronger tendencies towards a care or justice orientation (Yacker & Weinberg, 1990).
3.4.5 Demographic Questionnaire
A demographic questionnaire (see Appendix C) was included in order to obtain information about participants‟ age, gender, race, familial, and demographic details. This data provides another source of information against which significant findings can be compared.