Gender and affect: linguistic predictors of successful academic performance among economically disadvantaged first year college students

Abstract

Contemporary college students face a range of challenges to academic success, including inadequate preparation, anxiety and depression, social heterogeneity, and an uncertain future (Harackiewicz et al., 2002; Hunt & Eisenberg, 2010; Krieg, 2013; Richardson et al., 2012; Simon, 2017). As economic disparity grows, many students start college at a disadvantage. Education researchers have reported, for example, that students from economically impoverished backgrounds come to campus less prepared academically, with less economic and social capital, and experiencing greater stress than their more well-off peers (Beegle, 2000; Borrego, 2008; Brock, 2010; Cabrera & La Nasa, 2001; Duffy, 2007; Goldrick-Rab, 2006; Green, 2006; Howard, 2001; Kahlenberg, 2004; Lareau, 1987; Stinebrickner & Stinebrickner, 2002; Ting, 1998; Walpole, 2003; Yoder, 2005). As completion of college has been shown to provide social and economic advantages for financially disadvantaged students (Hout, 2012), it is important that we understand what contributes to success in this group of students. We should note that “impoverished” background is not limited to money; obviously, there are major disadvantages associated with living in violent neighborhoods, having a seriously dysfunctional family, exposure to substance abuse, etc. The tricky part for research is untangling causality as there is a high empirical correlation among these variables.In terms of college and in this article, however, economic poverty is the primary focus. Not surprisingly in the face of these obstacles, data suggest that students from poor backgrounds take longer to graduate, have lower grades, and drop out more often than others. It is important to note, however, that this is only part of the story, based on a nomothetic approach. That is, the studies cited above all compare poor students as a group to non-poor students as a group. When studies have examined how and why economically disadvantaged students do well (the implicit assumption being that this is a surprise), it is generally attributed to “resilience,” an often ambiguous term reflecting the fact that students have done well who were not expected to do so (Bonanno & Mancini, 2008; Fergus & Zimmerman, 2005; Infurna & Gress, 2010; Luecken & Gress, 2010; Luther et al., 2000; Waxman et al., 2003). This report is from an on-going, longitudinal study of economically disadvantaged students pursuing academic degrees at a large, public, four-year research university in the mid-South. We are examining a broad range of potential predictors of college performance, including high school GPA and standardized test scores, previous life experiences, students’ perceived academic success at admission, history of stress prior to college admission, and affective and cognitive styles. The goal is to examine variables that distinguish economically disadvantaged students who do well from those who struggle. In contrast to programs addressing the needs of economically disadvantaged students that approach the problem as a “one size fits all” institutional support requisite (Tinto, 2012), our broader goal is to ask what strengths allow otherwise disadvantaged students to perform well and how those strengths might be supported and encouraged. In this analysis, we focus on the predictive utility of students’ linguistic characteristics reflected in college application essays using the Linguistic Inquiry and Word Count (LIWC). The LIWC is a computerized, formal analysis of word usage based on a fixed dictionary of words that have been assigned to categories covering cognition, affect, motivation, and social characteristics (Pennebaker & King, 1999; Pennebaker & Graybeal, 2001; Pennebaker et al., 2003). With the exception of Total Word Count (TWC), each characteristic is expressed as a percent of total words.Various linguistically expressed characteristics have been correlated with a wide range of health, personality, and functional behaviors (see Pennebaker et al. 1999; 2001; 2003; 2014; 2015). Of particular relevance to our study, LIWC analysis has been used with college writing samples to predict academic performance (grade point average, GPA) with some success. Robinson et al. (2013) examined the utility of predicting term GPA using the LIWC to analyze written samples from students taking the same course.Students in a personality course were asked to write a brief “self-introduction” with the resulting LIWC scores correlated with final course GPA. Four variables correlated significantly with course GPA: Total Word Count (.11), Certainty (.11), Negative Emotion (.11), and Present Tense (-.11). TWC has been interpreted as reflecting “... intellectual engagement with the world... “ (Robinson et al, 2013, p. 8) and as reflecting complexity of thinking (Graybeal et al, 2002; Vercellonne-Smith et al, Gender and Affect: Linguistic Predictors of Successful Academic Performance among Economically Disadvantaged First Year College Students