TPC Journal V8, Issue 1 - FULL ISSUE

14 The Professional Counselor | Volume 8, Issue 1 from 12 to 84, with higher scores representing higher levels of perceived social support. For the purposes of this study, we used the Family Relationships subscale and Relationships with Friends subscale. The 4-item Family Relationships subscale includes items such as “My family really tries to help me” and “I get the emotional help and support I need from my family.” The 4-item Relationships with Friends subscale includes items such as “My friends really try to help me” and “I can count on my friends when things go wrong.” Zimet et al. (1988) reported high Cronbach’s alpha coefficients for scores on MSPSS subscales ranging from .85–.91, indicating good internal consistency. The reliability of the total scale for the initial sample was .88. Additionally, our sample reported coefficients ranging from .81 for the Family Relationships subscale and .88 for the Relationships with Friends subscale. College Self-Efficacy Inventory. The College Self-Efficacy Inventory (CSEI; Solberg, O’Brien, Villareal, Kennel, & Davis, 1993) was developed to assess a student’s confidence in their ability to successfully complete college-related tasks. Originally developed to measure college self-efficacy in Hispanic college students, CSEI data has established reliability beyond the initial norming population to also include ethnically diverse college students (Gore Jr., Leuwerke, & Turley, 2005). The CSEI is a 20-item instrument using a 10-point scale to assess a participant’s confidence in their ability to successfully complete a task from 1 ( not at all confident ) to 10 ( extremely confident ). Scores can range from 20 to 200, with higher scores indicating higher levels of confidence in one’s ability to successfully complete college-related tasks. The 20-item scale includes items such as “Make new friends at college,” “Talk to university staff,” and “Take good class notes” (Barry & Finney, 2009). Gore et al. (2006) reported Cronbach’s alpha coefficients for scores on the CSEI subscales ranging from .62–.89. The reliability of the CSEI for the initial sample was .93 (Solberg et al., 1993). Additionally, we observed a Cronbach’s alpha coefficient of .88 for our sample. Procedure After ethical review board approval, students registered in classes at one large university in Ghana were asked to participate in this study. A survey administrator, who was not the course instructor, shared the opportunity to participate in this study with students and disseminated an information sheet explaining the purpose, processes, and voluntary nature of the study. After having time to review the information sheet, the students choosing to participate in the study were given a packet including a demographic questionnaire, the INCA, the MSPSS, and the CSEI. All measures except for the demographic questionnaire were counter-balanced in an effort to control for random responding, order effect, and fatigue. Participants filled out hard copy surveys in class and turned them in to the survey administrator, who supplied them to the authors. Participant answers to the survey packet were entered into an SPSS spreadsheet. After all data was documented, the original hard copy surveys were securely destroyed. Data Analysis Statistical power analysis. We conducted a power analysis to determine the suitability of our sample size for identifying model fit using the criteria outlined by Stevens (2009): n/p ≥ 30. Using this standard, our largest scale (Belief in Self), consisting of eight items, would necessitate a sample size of at least 240. With a sample size of 696 (i.e., 87 participants per item), we considered our sample size sufficient for making statistical inferences about model fit. We also acknowledge that this model is over-powered for hypothesis testing and may lead to type I error. Therefore, when interpreting analyses, a greater emphasis was placed on model fit indices over p- values for χ 2 tests. Preliminary data analysis. The dataset was analyzed for missing values prior to performing statistical analyses. A small percentage of missing values (684 out of 71,100; .009%) was detected, but