Issues of stability and change in interest development.

Interest development is not an easily studied process. There are at least 4 methods for examining the process of stability and change over time: relative stability, absolute stability, profile stability, and structural stability. A program of research that focuses on examining these 4 types of stability is summarized relative to the issues pertinent to the development of vocational interests in children and adolescents.

Vocational interests are a cornerstone of career interventions, wherein efforts are aimed at matching interests with majors and occupations. There are two general models differentiating interests: those viewing interests as traitlike commodities and those focusing on the development and maturation of interests. Person-environment models (Holland, 1985, 1997; Rounds & Tracey, 1990) view interests are fairly traitlike entities to be assessed and matched to the environment. Developmental theorists focus on how interests change over time (e.g., Savickas, 1999; Vondracek, Lerner, & Schulenberg, 1986) as a function of maturation and interaction with the environment. Thus, one set of models stresses stability and the other change. We argue that a focus on development necessitates an examination of both stability and change. Although these two concepts may seem redundant in that one is sometimes viewed as the absence of the other, this is not true at least with respect to the development of interests. There are at least four nonoverlapping conceptions of development: relative stability, absolute stability, profile stability, and structural stability. Changes in one or more of these dimensions may be evident, whereas the others can remain constant. We argue that to have a thorough representation of interest development, all four types of interest continuity must be examined. The present article is an exposition of these different types of stability and change as well as a review of one program of research (i.e., the first author's) on these areas as it applies to the development of interests in children and adolescents ages 8 to 18 years.

Two recent reviews of the literature on children's vocational development (Hartung, Porfeli, & Vondracek, 2005; Watson & McMahon, 2005) focus on the broad literature within the area. Our focus is much more specific; we cover issues in longitudinal research on interests. Although cross-sectional research on interests has and continues to contribute to our understanding of interests, it cannot reveal key aspects of development and change because there are too many confounds. Furthermore, we focus on the more global traitlike quality of interests, which covers school and academic content as well as nonschool content. Although research that focuses on a narrower band of school-only content (usually science and math) is important (e.g., Eccles et al., 1993; Eccles & Wigfield, 2002; Fouad & Smith, 1996; Hilton & Berglund, 1974; Lapan, Shaughnessy, & Boggs, 1996; Lopez, Lent, Brown, & Gore, 1997; Meece, Parsons, Kaczala, Goff, & Futterman, 1982; O'Brien, Martinez-Pons, & Kopala, 1999), we view interests and their impact on decision making more broadly. As such, our focus is on the broader, more general representation of interests as typically summarized by Holland's (1997) six interest types: Realistic (R), Investigative (I), Artistic (A), Social (S), Enterprising (E), and Conventional (C). Although some have argued that these six types may not be an all-inclusive representation of interests (e.g., Tracey & Rounds, 1993), the domain representativeness of Holland's (1985, 1997) model is generally accepted (Rounds, 1995).

Relative Stability

The most common method of thinking about stability is the test-retest correlation of scales. This index provides information on the invariance of individuals' ranking relative to others. If the test-retest correlation is high, it indicates that an individual's position does not change appreciably relative to the positions of others over time. Individuals who scored higher than others at Time 1 will tend to score higher than others at Time 2. Hence the focus is on the invariance of the relative ordering of individuals on a specific scale. A common question of relative stability is the following: How does the ranking of an individual's score change over time relative to the scores of others within his or her cohort?

Using examinations of test-retest correlations, several researchers have concluded that interests are highly stable and change little after the ages of 25 to 30 years (Campbell, 1966, 1971; Hansen, 1984; Swanson, 1999). Common lore in the field is that vocational interests demonstrate too little relative stability over time for individuals younger than age 16 years (Low, Yoon, Roberts, & Rounds, 2005), as evidenced by many inventories not being recommended for use with individuals below this cutoff of 16 years (e.g., Harmon, Hansen, Borgen, & Hammer, 1994; Zytowski, n.d.). In a meta-analysis of test-retest correlations across different age groups, Low et al. demonstrated that down to ages of 12 to 14 years, there was moderate stability and that this stability increased over time, reaching a plateau at the ages of 18 to 20 years. The authors demonstrated that the relative stability of interests occurred earlier and at a higher level than previously thought. Furthermore, they found that this pattern did not vary across gender. In addition, results from studies of children even younger than the youngest age group (i.e., 12-14 years) examined by Low et al. have demonstrated similar results. Children in elementary school (Grades 4 to 6) and those transitioning into middle school have moderate levels of relative stability, and these indices increase as children age (Tracey, 2002; Tracey & Ward, 1998). Therefore, the research indicates that moderate relative stability of interests exists not only for ages 12 to 16 years but also for ages as young as 8 years old.

Although test-retest correlations are appropriate indicators of relative stability, they are relatively gross indicators. A finer tuned approach is to examine order relations over time using structural equation modeling, in which the amount of covariation across shorter and longer time spans can be partialed out. Tracey and Robbins (2005) used such an approach: They found that interests ...