Abstract
This article reports on the development and validation of the Distance from Privilege (DFP) measures. Items for the measures were developed on theoretical grounds. The primary sample consisted of 292 undergraduates with various majors from a Midwestern University and a Historically Black University in the south. One of the measures is the DFP-Resources scale which consists of 11 items with two factors described as social capital and economic resources. The other measure is the DFP-Status scale that has 10 items and one factor. Results revealed that the DFP consists of two measures with good internal consistency and construct validity. Each measure accounts for a sufficiently high proportion of the overall variance and has moderately high stability over time.
Although gifted girls and women have achieved great progress in most spheres of intellectual activity, research has shown that the pathway for them to STEM (Science, Technology, Engineering, Math) careers in the United States has milestones and danger zones (National Science Foundation [NSF], 2004). Ability alone does not predict which gifted women will go on to major in STEM and to persist to graduation. Almost one third of all male freshmen (29%), compared with only 15% of all female freshmen, planned to major in a STEM field in 2006 (NSF, 2009). More than 20% of male freshmen planned to major in engineering, computer science, or the physical sciences, compared with only about 5% of female freshmen. Only 32% of the White women who major in STEM persist to the doctorate; only 4.4 % of Asian and Pacific Islander Americans, 3.3% of Hispanic American women, 3% of African American, and .3% of Native American women persist to the PhD (NSF, 2009).
Women who enter STEM majors in college tend to be well qualified. Female and male first-year STEM majors are equally likely to have taken and earned high grades in the prerequisite math and science classes in high school and to have confidence in their math and science abilities (Brainard & Carlin, 1998; U.S. Department of Education, National Center for Education Statistics, 2000; Vogt et al., 2007). Nevertheless, many of these academically talented women leave STEM majors early in their college careers, and the rate of persistence is particularly low among minority women (NSF, 2009; Seymour & Hewitt, 1997).
Minority women are much less likely to choose STEM majors, to persist in STEM fields, and to graduate in STEM than nonminority women or men (NCES, 2009). Minority women who are high achieving in STEM fields may not only lack the mentors, networks, and social support of nonminority women and men; they are more likely to have barriers related to socioeconomic status, citizenship status, language status, and geographic status (lack of mobility and access) as well as those related to race and ethnicity (Kurpius, Kerr, & Harkins, 2005).
Understanding why bright, capable women such as these leave STEM majors requires a complex and broad knowledge of the variables that predict persistence. Too often, research on equity issues has focused on only a few predictors, or only on intrapsychic predictors, such as self-efficacy.
One model of the talent development of gifted women created by Noble, Subotnik, and Arnold (1999) takes into account a wide variety of predictors and may be very relevant to understanding bright women’s persistence in STEM careers. In addition, their conceptualization addresses context by focusing on a unique construct in the literature of giftedness: talented women’s relative distance from privilege (DFP), that is, the mainstream or core of their societies’ power centers. Noble and her colleagues consider the mainstream to be the cluster of traditions, values, and practices that constitute what has been termed the dominant culture within a given society. That the context of a woman’s life deeply affects her talent development is an underlying premise of this model.
Their model posits that predictions of potential based on ability should include the capacity to overcome certain barriers created by the individual’s distance from the center of privilege and power. Thus, it would predict that a poor, rural Navajo girl with above-average math achievement scores who has managed to graduate high school and enter a prestigious college may have as much potential for persistence in STEM majors as a wealthy European American boy with very high math-achievement scores at that same college.
The construct of privilege has long been examined and investigated in the literature of sociology and psychology (Fussell, 1989), but assessments of privilege have been hampered by the use of objective data such as socioeconomic status. Traditional indices of social class such as parental occupation, education, or income may not help in assessing the psychological impact of belonging to a particular status (e.g., Fouad & Brown, 2000). As Liu (2011) and his colleagues have argued, exploring social class as a subjective experience allows psychologists to shift beyond the strict social-class hierarchies (i.e., lower, middle class), which classify individuals based on income, education, and occupation. This classification scheme, says Liu, assumes everyone within a “class” views the world similarly.
Health studies have shown that for women, objective measures are less predictive of health variables than perceived privilege (Adler, Epel, Castellazzo, & Ickovics, 2000) Adler and colleagues have shown that higher objective SES is related only to less pessimism and passive coping with health problems, whereas subjective SES is significantly related to all of the psychological variables studied: Chronic stress, subjective stress, negative affect, pessimism, passive coping, perceived control over life, and active coping. In the area of career development, vocational psychologists in particular have attempted to discern the psychological effects of social class on individuals’ career decision-making processes and outcome (Blustein et al., 2002). Thomsen and Subich (2006), using a multidimensional measure of social status (Brown et al., 2002), found that students who self-reported greater economic resources, social power, and social prestige also reported greater confidence in their abilities to complete career decision-making tasks. That is, students who perceive themselves as having greater economic resources, social power, and social prestige than their peers report more certainty in and comfort with their career decision.
The studies above also point again to the problem with studies of privilege—the tendency to consider only a few variables, particularly class, race, and acculturation. Clearly, these variables interact in subtle and complex ways with a wide variety of other aspects of identity that make one feel more or less privileged. Noble et al.’s (1999) model addresses more than the obvious issues of nationality, race, gender, and class. Geography, they claim, not only affects a woman’s opportunities but, as importantly, her perceptions of possibilities for achievement (Astin & Leland, 1991). Language spoken and citizenship status have particular relevance for Latina students. For women, perceptions of age issues, perceptions of one’s abilities compared to others—often conceptualized as self-efficacy—and perceptions of one’s attractiveness compared to others—conceptualized as body image—may be importantly related to perceived privilege. It seems important that all aspects of identity that might be related to perceived privilege should be included in research and assessment. Hay’s (2001) ADDRESSING model of identity may provide a more comprehensive way of linking identity to perceived privilege. Hay’s model integrates all major variables that multicultural researchers have found to be important in shaping attitudes and behaviors: (a) age, (b) disability, (c) religion, (d) ethnicity, (e) social status, (f) sexual orientation, (g) indigenous heritage, (h) national origin, and (i) gender.
In addition, assessments created for the studies above were comprised of mostly middle-class European American students. Liu, Soleck, Hopps, Dunston, and Pickett (2004) claims that the lack of inclusion of more individuals who self-identify as belonging to either very high or very low social class limits response variability and perhaps the strength of the construct relations; a broader sample is needed to assess fully the implications of belonging to a particular social status.
Nevertheless, the studies that have been done so far have created a valuable foundation for the development of a DFP scale. A “ladder” format that allows participants to place themselves on a ladder of rungs from lowest to highest privilege, an inclusion of aspects of privilege that are related to social and cultural capital, and the inclusion of a measure of protective factors that might mitigate the impact of reduced privilege have all been advances in assessment of privilege.
The purpose of this study was to develop instruments to measure DFP that used the strengths and addressed the limitations of previous instruments. First, instruments were needed that subjectively assessed DFP. Second, instruments were needed that improved on previous multidimensional scales by adding more aspects of women’s identity that may be associated with perceptions of privilege. Finally, it was important to develop an instrument using a large, heterogeneous sample. Theoretically based and psychometrically sound measures of DFP could be used to test a comprehensive model of talent development for women in STEM.
Research in gender equity can benefit from a valid, reliable measure of DFP. In this field, particularly in STEM studies, researchers have struggled to understand issues of retention of minority, low SES, and differently abled women in the STEM pipeline. Often, studies are restricted to specific, narrowly defined populations, such as Hispanic women or, alternatively, have simply used broad, ambiguous definitions of diversity. A standardized instrument measuring all important facets of privilege may make it possible for STEM studies to both describe more accurately the subjective experience of girls and women as well as to identify those who could benefit from specialized interventions to enhance persistence.
The development of a DFP instrument also has the potential to contribute to the field of gifted education. Recent reviews of the research in gifted education (Dai, Swanson, & Chen, 2011) have established several major gaps in scholarly literature. First, all but 5.5% of research studies in the field of gifted education are descriptive, correlational, or qualitative. Empirical contributions using well-constructed measures are rare. Second, Jolly and Kettler (2008), tracing the history of research in gifted education note that in the last two decades, the only source of federal funding for research in gifted education, the Jacob J. Javits Program, (U.S. Department of Education, 2009) has required that the topic of investigation be related to underserved populations. A great deal of effort, therefore, has gone into adapting tests of cognitive abilities, or the development of alternative assessments of giftedness, but little effort has been made to operationalize the construct of “underserved.” A research study funded by the Javits program is likely to involve one of many definitions of giftedness and a sample of one or two of many groups perceived to be underserved. As a result, scholarly work has varied so widely both in definitions of giftedness and definitions of underserved populations, that generalization is difficult. An assessment such as the DFP instrument has the potential to provide clarity and consistency in research concerning populations of students who differ from majority, privileged individuals in important ways.
In both cases, generalization is difficult. A model of talent development that quantifies DFP can go a long way toward understanding how this variable interacts with ability, achievement, and environmental variables to affect the development of talent.
Method
Construction of the DFP Measures
Following a thorough literature review and based on results of the above studies, three separate domains were determined: (a) individual’s access to resources, (b) perceived privileged status in major demographic areas (e.g., SES, sexual orientation, disability status, race, and so on), and (c) protective factors. Each domain area was unique and was, therefore, comprised of different types of items.
The first domain, Access to Resources, measures the individual’s access, or lack thereof, to resources that coincide with privileged status. Initially 73 items were developed. Some examples include “I am at least a second generation member of my sorority/fraternity,” “I have not traveled abroad,” and “I can introduce myself properly to people of authority.” Responses range from 1 (strongly disagree) to 6 (strongly agree). A six-point scale was chosen to avoid the neutral response made possible by odd-numbered scales and to allow for sufficient variability in response. These items were randomly arranged and reverse scored to reduce the possibility of response bias.
The second domain is a measure of the individual’s Perceived Privilege Status across several descriptive categories, which were based on both Noble et al.’s model and Hay’s ADDRESSING model (Hays, 2001). The scale was constructed to represent cultural and personal characteristics of an individual that could be perceived as providing or prohibiting status in the individual’s current environment. These 10 areas were geography, gender, race/ethnicity, ability and disability status, citizenship status, intelligence, sexual orientation, religion, social class, and attractiveness. Perceived Privilege was measured by a Ladder scale, which had participants rank themselves from 1 to 10 on how privileged they perceived themselves in different demographic categories. A picture of a ladder was used for each area, and this represented where people stand in our society. At the top of the ladder was the place in which the area was most valued in society, and the bottom rung represented the least valued. The participant had to mark the rung of the ladder where they thought they would fall. The “ladder” concept was borrowed from recent literature (Adler et al., 2000; Goodman et al., 2003) underscoring the importance of measuring the perceived impact of various status characteristics on the individual, as opposed to an objective measure (e.g., income, education level, age, gender, and so on)
The third domain is Protective Factors, which is indicative of an individual’s ability to survive and thrive in the face of obstacles. We initially developed 25 items including “I’m very focused on my academic goals,” “My parents told me success stories about others in my ethnic group,” and “Surmounting obstacles is something that is expected of me.” Some items were reverse scored, and all were randomly arranged to reduce bias. Responses ranged from 1 (strongly disagree) to 6 (strongly agree).
Participants
Sample 1
Data were collected from 292 undergraduate students at a large university in the Midwest and a historically black college in the South (Age, M = 22.47, SD = 5.75; 65.4% female, 34.6% male). Ethnicities represented included 63.4% White, 28.4% Black, 5.1% Hispanic, 2.4% American Indian, and 3% Asian.
Sample 2
Estimates of test-retest reliability across a two-week interval were obtained by administering the instruments to undergraduate students at a southwestern public university. The students were in four educational psychology classes. The 68 students who took part in this study ranged in age from 18 to 44 (M = 20.80, SD = 3.57), and 24 (35.3%) were female. Forty-seven (74.6%) of the 68 students were White.
Measures
Demographic questionnaire
This questionnaire consisted of questions designed to assess more objective measures of characteristics and descriptors that often coincide with privilege, such as gender, race, ethnicity, sexual orientation, citizenship status, parental occupations, and household income. This also served as a comparison tool to the DFP scales.
DFP was measured through the use of two scales—the Access to Resources Scale and the Ladder Scale. The Access to Resources Scale consists of 73 original items representing resources associated with privilege. Responses range from 1 (strongly disagree) to 6 (strongly agree). These items were randomly arranged and reverse-scored to reduce the possibility of response bias. The Ladder Scale is made up of 10 original items representing 10 distinct demographic areas developed to assess perceived DFP. Responses range from 1 (least privileged) to 10 (most privileged). This was patterned after recent research in the area of perceived status (Adler et al., 2000; Liu et al., 2004).
Procedures
Research assistants visited classes in education, natural sciences, and physical sciences at the two universities to recruit participants (Sample 1) to fill out the questionnaire packet. Students who agreed to participate were given one week to fill out the measures. An incentive was provided for participation. Data were collected and analyzed to produce a psychometrically sound and functional measure. Sample 2 (n = 68) received the 11-item DFP Resources and the 10-item DFP Status (Ladders) scales in an effort to obtain test–retest information. All participants signed consent forms.
Results
Access to Resources Scale
An exploratory factor analysis was conducted on the 73 original items of the DFP-Resources Scale. The extraction method utilized was a Principal Component Analysis with a Varimax with Kaiser Normalization rotation method. Results of the factor analyses indicated that there were two distinct and interpretable factors. The decision was made based on three criteria: parsimony, interpretability, and a priori criteria (e.g., adequate factor loadings, eigenvalues > 1.0, scree plot analysis). Item loadings ranged from .59 to .79 on their respective factors and the amount of variance explained by was 27.82% on the first factor and 22.21% on the second factor (Table 1).
Summary of Items and Factor Loadings for Varimax With Kaiser Normalization Two-Factor Solution for Distance From Privilege Resource Scale (N = 262)
This solution accounted for 50.03% of the total variance. A total of 11 items were retained to form the final version of the measure resulting in a possible total score ranging from 11 to 66, with higher scores indicating greater access to resources.
Internal consistency estimates were determined for each factor and the total score on DFP-Resources Scale. The estimates were as follows: Factor I (.81) with 6 items, Factor II (.72) with 5 items and the Total Score (.77) with all 11 items. Factor I items were clearly related to social capital—that is, access to the social networks, the social relationships that provide value to the individual. Factor II items all concerned economic resources—the things money can buy that support a middle class or higher lifestyle.
DFP Status (Ladder Scale)
All 10 original items were retained for the DFP-Status (Ladders) scale (Figure 1).
The ladders scale
These 10 items showed adequate variability with actual scores ranging from 34 to 100 (M = 75.8, SD = 12.99). Internal reliability estimates demonstrated an adequate reliability for this 10-item measure (α = .70).
DFP-Protective Factors
An exploratory factor analysis was conducted using the original 25 items. The extraction method utilized was a Principal Component Analysis with a Varimax with Kaiser Normalization rotation method. Again, three criteria were used in the decision-making process: parsimony, interpretability, and a priori criteria (e.g., adequate factor loadings, eigenvalues > 1.0, scree plot analysis). Although six factors were extracted with adequate factor loadings and eigenvalues > 1.0, the factors were not easily interpretable. It was decided this particular measure needed more work to be useful and, thus, is not discussed further.
The two remaining measures, Resources and Ladders, were administered together to the same group of 68 students, 2 weeks apart, as one instrument. The overall test–retest reliability for the 2-week period was .82, indicating good stability over this period.
Analysis of Group Differences in Perceived Privilege: Validity
A further examination of privilege at the group level was also conducted on select variables. In the case of gender privilege, women reported feeling significantly less gender privileged (M = 6.31, SD = 2.13) than their male counterparts (M = 8.0, SD = 1.47), t(281) = −6.97, p < .001. This corresponded with expected results based on previous research regarding the status of being male as compared with being female. Notably, perceived geographic privilege was significantly different among various living situations, F(5, 276) = 5.67, p < .001. The highest geographic privilege was noted among those living in a suburban area (M = 8.17, SD = 1.40), and significantly higher than those living in midsize cities (M = 6.75, SD = 1.95), rural areas (M = 6.47, SD = 2.37), and small towns (M = 6.20, SD = 2.01). When considering perceived privilege due to social class (M = 7.24, SD = 1.69) an interesting trend was found when comparing perceived privilege to objective data (household income, parental education levels). It was expected that perceived privilege of social status would increase as the objective variables increased in status, that is, higher parental education levels and household income level the higher the perceived privilege. The results did follow this trend for income and father’s education; however, there was not a significant relationship between perceived social status and mother’s education level. When examining race, it was found that “White” individuals (M = 8.74, SD = 1.38) reported significantly higher race privilege than “non-White” individuals (M = 4.98, SD = 2.35), F(1, 287) = 294.58, p < .001.
Other areas of perceived privilege explored in this sample were sexual orientation, citizenship status, and religion. Heterosexual orientation; U.S. citizenship; and Christian religion all had higher means, but tests of significance were not possible because of small numbers of nonheterosexuals, noncitizens, and nonChristians.
Discussion
This study developed and validated two scales for measuring DFP, a theorized barrier for talented women in STEM. The first measure, Access to Resources, was found to have two clear factors accounting for 50% of the variance, corresponding to social capital and economic resources. This finding reinforces what sociologists have contended for over two decades: that financial capital and social capital are distinctly different kinds of resources. Having money is not the same as having beneficial social connections, and for many less privileged women, scholarships and other support for their education may not, by themselves, provide the necessary social networks that can support and encourage persistence in college and in STEM fields.
The second measure, DFP Status, was found to have acceptable internal consistency and to provide a measure of the individual’s own perceptions of privileged status. The Perceived Privilege scale effectively differentiated between the following: males and females; minorities and nonminorities; heterosexuals and nonheterosexuals; and levels of socioeconomic status, as well as types of geographic locations, as predicted by theory and research on privilege. It, therefore, has strong construct validity. The Perceived Privilege scale also reinforces the idea that college students internalize their society’s valuing of their gender, race, geographic origin, sexual orientation, religion, social class, and citizenship status. This has implications for students vulnerability to stereotype threat (Steele, 1997), self-efficacy (Betz & Voyton, 1997), and a host of other internal variables that affect achievement and persistence in college.
With regard to persistence in STEM, both perceived and objective DFP probably interact with other major variables that predict women’s probability of continuously progressing toward a career in STEM. Most studies in this area have had a strong focus on cognitive variables, such as self-efficacy rather than on environmental variables. Although enhancing girls’ self-esteem and young women’s self-efficacy in STEM may contribute significantly to their choices to enter science, it is apparent from studies of women further along in the pipeline that social cognitive variables account for only a little of the variance in persistence. For college students in general, Pascarella and Terenzini (2005) found that engagement with faculty and involvement with the campus are the strongest predictors of persistence. Gagliardia (2005) found that for female honors students, self-perception variables accounted for little of the variance in the prediction of bright women’s persistence in STEM through graduation. Instead, perceptions of the college environment and perceptions of mentoring accounted for 52% of the variance in persistence.
It is also evident that the majority of STEM equity theory and research, while tangentially addressing career/family issues, seldom target the variable of gender relations. At the end of the pipeline, however, gender relations, particularly characteristics of marriage and family, have been found to distinguish between those women academic scientists who succeed in attaining tenure track placement, tenure, and rank. In summarizing the results of an extensive study of women in academe, the NSF report, “Gender Differences in the Careers of Academic Scientists and Engineers” (Division of Science Resources Statistics of NSF’s Directorate for Social, Behavioral, and Economic Sciences, [NSF, 2004]) stated:
We find evidence that female scientists and engineers are less successful than their male counterparts in traveling along the academic career path. Some of this disparity appears to be related to differences between the sexes in the influence of family characteristics. Typically, married women and women with children are less successful than men who are married and have children. (p.2)
It is therefore critical that these major predictors of persistence in college (engagement and mentoring) and persistence in career (gender relations) be entered into the equation with DFP. An earlier version of the DFP scale was combined with other variables in this way in major NSF project conducted by the authors. They developed a valid and reliable measure of positive, egalitarian gender relations called the Gender Relations Inventory (Multon, Kerr, Fry & Syme, 2007) and conducted a multisample SEM study in which DFP (perceived high social capital, actual middle and lower status), positive gender relations (equal valuing of career and relationship), mentoring and high achievement predicted college engagement, confidence in major career engagement, and hope for the future in a large, diverse sample of college women (Multon, Kerr, Kurpius, & Hammond, 2009). In this way, the Noble, Subotnik, and Arnold model of women’s talent development received its first empirical support.
There is evidence, therefore, that the DFP may resolve a number of problems in research and practice. First, it may help in assessing the psychological impact of belonging to a particular status (Fouad & Brown, 2000) rather than simply providing objective information to researchers and practitioners. This avoids the problem of assuming that a particular status means the same thing to everybody. In addition, the DFP measures more variables and approaches the breadth of the ADDRESSING model, as well as other models, of ways in which individual identities differ, making it more useful than unidimensional measures of race, gender, SES, and so on.
In gifted education, the DFP may be useful in providing a more comprehensive measure of status, rather than the simple demographic facts of race, class, or gender, or the common, but less descriptive variable of reduced or free lunch status for students. College personnel might also take interest in the possibility of going beyond these conventional measures, to select talented students who could benefit from mentoring and career development interventions.
More research will be needed to provide support for the DFP as an effective research and practice measure. It is not clear if the assessment is appropriate only to college students, as in this study, or if it can be used with adolescents or working adults. Studies using the DFP in assessment of counseling clients would be particularly useful; it would be important to evaluate the usefulness of the DFP both to the counselor in understanding the worldview of the client, as well as the usefulness to the outcome of counseling.
Besides establishing and validating these measures that can be used to test the theoretical model of women’s talent development of Noble et al. (1999), this study also created ways to assess clients’ and students’ understanding of their DFP as a way of improving counseling the talented women in STEM and other fields.
Footnotes
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This manuscript was funded by a grant from the National Science Foundation.