This study created a measure of the perceived learning process during graduate internship, the Scale of Internship Learning (ScIL). The rising number of internships and importance of assessing learning makes measurement of internship learning imperative. Criticized for lack of theoretical framework and theory development (Parks, Onwuegbuzie, & Cash, 2001), few internship studies have been published on the learning process (Linn, 2004) or consider influence of co-workers.
Guided by DeVellis (2003) and Creswell and Clark (2006), mixed method measure creation and revision utilized three phases of data collection: phenomenological interviews (Creswell, 1998), expert review, and field test. In addition to phenomenological data, literature from multiple fields were used to identify themes for 71 preliminary items, including Communities of Practice (Lave & Wenger, 1991); co-participation (Billett, 2002); expansive learning (Fuller & Unwin, 2007); self-directed learning (Guglielmino, 1978); and workplace learning (Coetzer, 2007; Tannenbaum, 1997). In follow-up interviews, qualitative participants completed a draft survey. Cognitive interviewing probes (Dillman et al., 2009; Groves, et al., 2009; Oskenberg, et al., 1991) were used to facilitate item discussion. Three experts subsequently evaluated the item pool, rating item relevance, clarity, concision and identified other aspects of the construct (DeVellis, 2003).
During the field test, 381 graduate students from a Northeastern private university completed the online instrument. KMO sampling adequacy was excellent (.941). Exploratory Factor Analysis Maximum Likelihood extraction with Promax oblique rotation was used to identify emergent dimensions of the perceived internship learning process and evaluate construct validity. Parallel analysis (Horn, 1965) utilizing O'Connor's (2000) SPSS syntax identified number of factors to retain.
The scale was optimized (a=.967), removing items contributing least to reliability. Five interpretable dimensions emerged, explaining 53.5% of variance and yielding strong alpha coefficients: Proximal Co-worker Relationship (.947), Proximal Supervisor Relationship (.951), Mode of Learning (.826), Intern Initiative (.839), and Intern Participation (.878). Goodness of fit of the model was significant (X2=.000) and factors correlated in a predictable manner. Intern Participation positively correlated to Proximal Co-worker Relationship (.643) and Proximal Supervisor Relationship (.591).
Empirical and theoretical analysis of factors and correlations suggest strong support for construct validity, recommending future testing with other populations.