5.2 Statement of Authorship
5.3.4 Method
5.3.4.2 Measures
The questionnaire included demographic, psychological, and academic items and scales. Standardised scales consisted of items measured on Likert scales. The means of summed scales were the interval-level data necessary to meet the assumption of normal distributions for parametric tests, which are nevertheless robust for anomalies (Norman, 2010;
Wigley, 2013).
Demographic items. Participants gave their home country, or Australian home post- code, post-code of term-time residence, and indicated type of residence, sex, and age. A text box allowed comment on residence.
138 Novel 12-Item Integrated Support Scale (ISS; Straub, 2020).
The measure consisted of the following constructs.
Comfort (material): Participants indicated 1 (No), or 2 (Yes) to items related to comfort, that is, a four-item sub-scale measuring amount of domestic support, for example,
“Are most meals provided for you?”, and the following additional questions: “Would it be quite easy to get things fixed when not working properly or broken?” and “Are there spaces where you can meet with others? Or relax?”
Communication (ease): Participants indicated 1 (No) or 2 (Yes) to two items tapping ease of communicating with (a) those in control of the residence, and (b) with co-residents.
Commonality (purpose): Participants indicated 1 (No) or 2 (Yes) to the question “Like you, are other residents mostly students?” and three questions related to academic support, for example, “Is all the technology you need for your studies available?”
Care (interpersonal): Two questions related to the care and valuing of residents.
Participants indicated 1 (No) or 2 (Yes) to the question “Do more senior residents check on how the younger are doing or feeling?” Participants then indicated how many co-residents they would greet when out, ranging from 0 (None), to 8 (150+). The dichotomous and ordinal responses were summed.
The 12-item belonging sub-scale of the Interpersonal Support Evaluation List for College Students (ISEL; Cohen & Hoberman, 1983).
The measure included such items as, for example, “I don’t have friends who would comfort me by showing some physical affection, like giving me a hug.”, “I belong to a group
…. that meets regularly or does things together regularly”. Responses to the items of the sub- scale were modified from 1 (Probably true) or 2 (Probably false) to a five-point Likert scale ranging from 1 (Disagree strongly) to 5 (Agree strongly).
139 The Psychological Sense of Community Scale (PSCS; Jason et al., 2015).
The measure consisted of nine items. Five items related to the entity (macro-system), three to membership (micro-system), and one to both entity and membership. Items were modified to apply to both small and large communities in residence, for example, “For me, this community of people is a good fit” (Entity), “The people with whom I live can depend on each other” (Membership), “I feel good being involved in this community (Entity), and in helping the others (Membership)”. A five-point Likert scale measured responses from 1 (Strongly disagree) to 5 (Strongly agree). Participants who resided alone indicated 0 (N/A – I live alone).
To measure psychological sense of community at the universities, items of the PSCS were modified as follows. University was the entity, for example, “For me, this university is a good fit”, and students were the members of the entity, for example, “Students can depend on other members of this university.”
Academic outcomes. Estimated GPA and intention to persist defined short-term and intermediate-term academic success respectively. For GPA estimate, participants reported end-of-course grades which were allotted points according to the system of the university at which participants were enrolled, for example, High Distinction = 7 points, Distinction = 6 points, and so on. The total of each participants’ points was multiplied by the number of units allotted to each subject (for example, 4.5), and subsequently divided by 18, the total number of possible units.
For intention to persist, participants affirmed as personally true one of two statements of intention to persist with tertiary education, or one of three statements indicating the
contrary.
140 Well-being. The 14-item Warwick-Edinburgh Mental Well-Being Scale (WEMWBS;
Tennant et al., 2007).
The measure included such statements of well-being as, for example, “I’ve been feeling relaxed”, “I’ve been feeling close to other people”, “I’ve been interested in new things”. The scale was modified from a 7-point Likert scale to a 5-point scale 1 (Less than usual), 2 (Same as usual), and 3 (More than usual). In its development, the scale showed a significant moderate inverse correlation with the General Health Questionnaire-12 (GHQ-12) (Tennant et al., 2007).
Well-being, contact with distal co-residents, and contact with distal university peers defined psychosocial outcomes. For contact with distal co-residents and university peers, participants indicated 1 (No) or 2 (Yes) to the questions, “If you went home during the
pandemic, did you stay in regular contact with any friends in your accommodation?” and “….
at the university?”
5.3.4.3 Procedure.
The cross-sectional study was conducted online by questionnaire survey (see Appendix F). Data were collected following the first semester examination period August - November 2020 (N = 122). Participants had remained in South Australia
A priori, the desired alpha level (0.05), power level (0.95), and effect size (0.3) were entered into the G*Power 3 analysis program (Faul et al., 2009) to determine the sample size for correlational and regression analyses. Output showed that a sample of 111 participants was sufficient.
Access. Participants accessed the survey (a) using information printed on notices and flyers distributed at all campuses of the three universities (except one regional campus), (b) activating a hyperlink embedded in a flyer emailed by staff of student
141 villages and colleges to all first-year residents; or using a Research Participation System to receive .5 hour of course credit as did first-year psychology students at one university.
Instrument. The introductory letter stated (a) title of the study, (b) University of Adelaide School of Psychology human research ethics’ code of clearance (H-2020-20/61) and details for contacting the committee, (c) eligibility and ineligibility criteria for participation, (d) estimated 12 minutes to complete the questionnaire, (e) voluntary nature of participation, (f) assurance of anonymity in results, (g) confidentiality, (h) benefits of the research, (i) list of counselling services, and (j) telephone numbers of researcher, principal supervisor, and co- supervisor. Thus informed, participants indicated consent to proceed.
Participants, other than Research Participation System psychology students, were invited on completion of the survey to enter a draw for cinema vouchers and/or have a summary of the research results sent to them. Both options required an email address seen only by the researcher.
Missing data. Missing data were managed by excluding cases pairwise in the final analyses (N = 122). Missing data accounted for small fluctuations in sample sizes (N and n) across analyses, except for contact with distal co-residents and university peers which each had N = 100 (22 missing cases). The 22 missing cases for the contact variables were those of participants who stayed, that is, did not return home in Semester 1, 2020.
Recoding. Two categorical variables were recoded after obtaining frequencies and summary statistics. Type of residence was recoded from six to four categories. The categories “Flat” and “Share house” were retained. Organisationally similar residences were combined, that is, collegiate Hall (n = 6) and Colleges (n = 36), and commercial (n = 12) and university “Village” (n = 8). Statistical analysis proceeded using four types of residence.
Intention to persist was dichotomised, (No/unsure) and 1 (Yes), for analysis by logistic
142 regression. Similarly, binary values of Contact co-residents and Contact university peers were recoded to 0 (No) and 1 (Yes).
Scale Reliability. Table 20, shows obtained Cronbach’s alpha for each of the four scales, and the Belonging sub-scale of ISEL. The alphas indicate internal reliability of good (.60 - .70) to very good (> .80). Alpha for Residence community (R-community) indicates some redundancy (> .95) (Hulin et al., 2001). A factor analysis confirmed the internal reliability of the novel ISS (see Appendix J).
Table 20
Obtained Reliability Coefficients for Scales and Belonging Support Sub-scale
N No. of Items Cronbach’s Alpha
Integrated support 121 14 .74
Belonging (ISEL sub-scale) 119 12 .91
Residence community 122 9 .96
University community 119 9 .87
Well-being 121 14 .94
5.3.4.4 Statistical analysis.
Collected data were compiled and analysed using IBM SPSS Statistics (Version 27) predictive analytics software. All continuous variables were scored in the direction of low to high. Frequencies and descriptive statistics were obtained.
One-way between-groups analyses of variance (ANOVAs) found the effect size of each type of residence (Flat, Share house, College, “Village”) on levels of Integrated support, perceived available Belonging, sense of Residence community (R-community), and sense of University community (U-community).
Parametric tests were conducted where assumptions were met and non-parametric
143 alternatives where not. Pearson’s product-moment correlation coefficients and Spearman’s rank correlation coefficients were categorised small ≥ 0.1, medium ≥ 0.3, large ≥ 0.5 (Cohen, 1988, pp. 79-81). Multiple and logistic regressions found sizes of effect on relationships between continuous factors and outcomes. Effect sizes (η2 or r2 change) were categorised small ≥ 0.01, medium ≥ 0.06, and large ≥ 0.14 (Cohen, 1988, pp. 79-81) as was magnitude of difference between means (Cohen, 1988, pp. 284-287).
The alpha level to determine significance was .05 for all statistical tests.