Development and validation of an instrument to assess the self- confidence of students enrolled in the advanced pharmacy practice experience

American Journal of Pharmaceutical Education, Spring 2002 by Wongwiwatthananukit, Supakit, Newton, Gail D, Popovich, Nicholas G

Data collected from the students were obtained on a crosssectional basis. The collection dates, however, were varied slightly because the time of each clerkship among various colleges/schools might be different. To minimize the history and maturation effects, the investigators would attempt to coordinate with each school/college's experiential coordinators an appropriate time to collect student data to ensure that students respond at relatively similar stages of their APPE clerkships. For this study, a selected time for data collection was at the end of three weeks of clerkship for all schools/colleges of pharmacy.

Statistical Analyses. Data were analyzed and managed using the Statistical Analysis System (SAS) software system V 8.1(22). Descriptive statistics of students' demographic variables were computed. The level of significance for any statistical tests was established at a = 0.05. Pearson correlation coefficients were calculated for scores on each of the 74 items for the developed instrument with the total scores on the MCSDS to eliminate items that might contribute to a possible biasing effect of social desirability before performing exploratory factor analysis (EFA).

EFA (i.e., using PROC FACTOR in SAS) was used to explore the four tentative subscales within the group of items. In this study, the term "factor" was used interchangeably with "subscale". To determine factorability of the data (ie., the appropriateness of factor analysis), correlation matrices [ie., observed, partial (anti-image)], the Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMOMSA) for all of items and each item, and squared multiple correlation were determined. A principal factor analysis was used as a method of factor extraction with an oblique (i.e., Promax) rotation. The squared multiple correlations (SMC) were used as an initial communality estimate. An eigenvalue (Kaiser's criterion) cutoff of one, the Cattell's scree test, the proportion of the common variance accounted for a factor and the factor solution, and residual correlation matrix were used to determine the number of factors to be retained. An item was retained on a given factor/subscale if the factor loading was > 0.60 for that factor. The simple structure, interpretability criteria, and at least three to four items per factor were used to interpret a factor solution. Item analysis (i.e., Cronbach's coefficient alpha, corrected item-subscale correlation) was also performed after the factor analysis procedure.

Objective One Results. Of 260 student instruments mailed, 137 student instruments (i.e., 52.7 percent response rates) were returned and determined usable. Fifty-five male students (40.1 percent) and 82 female students (59.9 percent) responded. The mean and SD age of students (N=137) was 27.325.16. The median and mode of age were 25 and 24. The range of age was 23-53. The majority of students were white (N=100, 73 percent), 16.8 percent (N=23) were Asian-American, 5.1 percent (N=7) were Hispanic-American, 0.7 percent (N=1) were African-American, 0.7 percent (N=I) were Native-American, 0.7 percent (N=1) were International (Non-U.S.), and 2.9 percent (N=4) were other ethnicity/race. One hundred and eleven of the student respondents (81 percent) were enrolled in a traditional PharmD program, 23 (16.8 percent) were enrolled in a postbaccalaureate PharmD program, and 3 (2.2 percent) were enrolled in a nontraditional PharmD program. The mean and SD GPA (N=128) was 3.270.32. The median and mode were 3.25 and 3.00. The range of GPA was 2.10-4.00.

Pearson correlation coefficients demonstrated that 19 items correlated significantly (P

A three-factor solution appeared to be the best approximate simple structure and conceptual meaning of the factor underlying a set of items. The proportion of the common variance accounted for three factors still accounted for 72.3 percent of the total variance. Factors one, two, and three accounted for 56.6 percent, 10.8 percent, and 4.9 percent of the total variance, respectively. The coefficients of the residual correlation matrix demonstrated small and negative values, which supported the three-factor solution fit the data