Curriculum standardization, stratification, and students’ STEM-related occupational expectations: Evidence from PISA 2006

International Journal of Educational Research 72 (2015) 103–115

Contents lists available at ScienceDirect

International Journal of Educational Research journal homepage: www.elsevier.com/locate/ijedures

Curriculum standardization, stratification, and students’ STEM-related occupational expectations: Evidence from PISA 2006 Seong Won Han * [2_TD$IF]University at Buffalo, The State University of New York, USA

A R T I C L E I N F O

A B S T R A C T

Article history: Received 28 October 2014 Received in revised form 23 March 2015 Accepted 27 April 2015 Available online

This paper uses data from the Program for International Student Assessment (PISA) 2006 to examine the associations between characteristics of national education systems (the standardization of curriculum, the number of school types available to 15-year-old students, and early tracking) and students’ STEM occupational expectations. Results show that the associations between characteristics of national education systems and students’ STEM occupational expectations differ by gender as well as across STEM subfields and academic performance levels. Students’ computing and engineering occupational expectations are not associated with the characteristics of secondary education. The negative association between a standardized education system and students’ health service occupational expectations is stronger for students at the bottom of the performance distribution than students at the top. ß 2015 Elsevier Ltd. All rights reserved.

Keywords: Curriculum standardization Stratification STEM International PISA

1. Introduction There is a growing interest in cross-national differences in student STEM career aspirations and expectations as well as cross-national differences in math and science achievement. Prior research has shown that students’ occupational plans in high school are strong predictors of educational and occupational attainment in STEM fields (Morgan, Gelbgiser, & Weeden, 2013; Tai, Liu, Maltese, & Fan, 2006; Xie & Shauman, 2003). Given the importance of STEM occupational expectations for future educational and occupational attainments, cross-national research in education has examined which students want to pursue science-related careers in the future (Organisation for Economic Co-operation and Development, 2012, 2009c). These studies have found that in several countries students strongly desire science-related careers. A 2012 Organization for Economic Co-operation and Development (OECD) report showed that, on average across OECD countries, about 55 percent of students planned to work in high-status occupations such as legislators, senior officials, corporate managers, and professionals, and about 60 percent of those students planned to work in science-related professional careers (Organisation for Economic Co-operation and Development, 2012). However, in a number of countries such as Korea, Turkey, and the Slovak Republic, science-related careers were relatively less attractive to 15-year-old students. For example, in Korea, about 61 percent of students planned to work in high-status occupations, but only about 34 percent of those students planned to work in science-related careers (Organisation for Economic Co-operation and Development, 2012). Prior studies have shown

* Tel.: +1 7166451080; fax: +1 [10_TD$IF]7166452481. E-mail address: [email protected] http://dx.doi.org/10.1016/j.ijer.2015.04.012 0883-0355/ß 2015 Elsevier Ltd. All rights reserved.

104

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

that academically strong students are more likely than low-achieving students to report specific future career expectations (Sikora & Pokropek, 2011), and students who are academically strong in science have higher levels of interests in pursuing science-related careers than other students (Organisation for Economic Co-operation and Development, 2009c). Across OECD countries, an average of 61 percent of top performers in science reported that they would like to work in a career involving science; however, this proportion varied significantly across countries. For example, in France and Spain, 77 percent of top performers in science reported wanting a career involving science, while in the Czech Republic only about 39 percent of top performers in science were interested in a science-related career. High-performing students in Japan (43[1_TD$IF] percent), the Slovak Republic (45[12_TD$IF] percent), and Finland (47[13_TD$IF] percent) fell between these two extremes. This OECD report (2009c) concluded that while educational systems in the focal countries have had significant success teaching scientific knowledge and skills, they have had less success fostering science-related career aspirations among academically strong students; the report also suggested that STEM career expectations may be linked to the features of national education systems. However, there is a lack of empirical research on the link between features of education systems and sciencerelated career aspirations and expectations. A large amount of research in education has focused on the extent to which individual- and school-level factors, including math attitudes, self-assessments, and math and science achievement, predict STEM educational and occupational expectations (Correll[3_TD$IF], 2001; Riegle-Crumb, Moore, & Ramos-Wada, 2010; Xie & Shauman, 2003). Cross-national studies have also examined the extent to which individual- and school-level factors are associated with students’ STEM occupational expectations (Buccheri, Gurber, & Bruhwiler, 2011; Sikora & Pokropek, 2012b). Recent cross-national research in education has shifted from a focus on individual- and school-level factors to a focus on country-level factors, primarily the stratification level of secondary education systems, in examining students’ sciencerelated career expectations (Sikora & Pokropek, 2012a). Stratification indicates the degree to which students are selected into separate school types with clearly differentiated kinds of school curricula (e.g., academic versus vocational tracks). Researchers have also proposed that standardization, the degree to which school curricula are standardized nationwide, is another key feature of the organization of national education system (Allmendinger, 1989; Kerckhoff, 2001). However, there is a lack of empirical research on the link between standardized education systems and STEM occupational expectations. In the current study, I investigate the extent to which the features of national education systems are associated with nationallevel differences in students’ STEM occupational expectations by considering both standardization and stratification as features of national education systems. The study also examines whether the associations between features of national education systems and STEM occupational expectations remain consistent by gender and across different levels of science performance. To address these questions, the study used data from the Program for International Student Assessment (PISA) 2006, which was administered in OECD member and partner countries. 2. National education systems and students’ STEM occupational expectations: a comparative perspective Over the past decade, cross-national studies on students’ educational and occupational expectations have shifted their focus from individual- and school-level factors to country-level factors (e.g., national education systems) as well as the interactions between individual characteristics and macro-level educational contexts (Buchmann & Dalton, 2002; Buchmann & Park, 2009; McDaniel, 2010; Sikora & Pokropek, 2011, 2012a). These studies have employed the stratification– standardization framework proposed by Allmendinger (1989) to classify national education systems. In this context, standardization refers to ‘‘the degree to which the quality of education meets the same standards nationwide’’ (Allmendinger, 1989, p. 233). Standardization is generally higher when the central government controls curricular, learning, and assessment standards (Kerckhoff, 2001). Countries with highly standardized education systems (e.g., Japan and Korea) have national curriculum standards or courses of study that define the content to be taught by grade and subject. In Allmendinger’s framework, stratification (also called differentiation) refers to ‘‘the degree to which systems have clearly differentiated kinds of schools whose curricula are defined as higher and lower’’ (Kerckhoff, 2001, p. 4). Stratified educational systems most often include tracking, streaming, or grouping between secondary schools. Compared to comprehensive (unstratified) educational systems, stratified systems are more likely to provide diverse vocational education programs for secondary students. Stratified systems tend to sort students into different tracks (programs) at early age. Within the standardization-stratification framework, for example, the U.S. education system is characterized by low levels of both standardization and stratification, while the Japanese education system is characterized by high levels of standardization but low levels of stratification (Shavit, Mu¨ller, & Tame, 1998). Prior research has focused specifically on the stratification of educational systems as a potential explanation for crossnational variation in educational and occupational expectations among youths (Buchmann & Park, 2009). Researchers have found that students in highly stratified educational systems tend to have more realistic occupational expectations than those in undifferentiated systems (Buchmann & Park, 2009). This pattern may be due to the greater restriction of students’ options for educational and occupational trajectories at the secondary level in highly differentiated systems. Recently, cross-national research on students’ occupational expectations has expanded its focus to specific occupationbased expectations. In particular, a number of studies have reported significant cross-national differences in students’ science-related career expectations (Martin, Mullis, Gonzalez, & Chrostowski, 2004; Mullis, Martin, Gonzalez, & Chrostowski, 2004; Organisation for Economic Co-operation and Development, 2007, 2009c). Because PISA 2006 focused on assessments of scientific literacy, students’ attitudes toward science, and students’ expectations of the jobs they would hold at age 30,

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

105

a number of researchers have used the data to examine science-related career expectations in a comparative perspective (e.g., Kjærnsli & Lie, 2011; Sikora & Pokropek, 2012a). PISA is a triennial survey that measures the knowledge and skills of representative samples of 15-year-old students nearing the end of compulsory education in either public or private schools in each participating country. PISA assessed performance in reading, mathematics, and science literacy. In each PISA survey wave, three subject domains were tested and one of the three was assessed as the major domain. According to a recent OECD report (2007), across OECD nations, an average of 25 percent of students expected to attain a science-related career by age 30. However, only 8 percent of Japanese 15-year-olds expected to have a science-related career at age 30 while about 40 percent of 15-year-olds in Portugal expected to have a science-related career around age 30. Levels in the United States (38 percent) trail closely behind those in Portugal, and are followed by Canada (37 percent), Mexico (35 percent), Iceland (32 percent), Italy (32 percent), and Poland (31 percent). Relative to students in other OECD countries, students in high-achieving countries such as Finland, Japan, and Korea reported low-levels of interest in pursuing STEM occupations. Prior research (Han, 2013) has also examined net country trends in STEM occupational expectations after controlling for heterogeneity at the student and school levels (e.g., gender, parental educational attainment, family SES, immigration background of both students and parents, parental jobs in STEM fields, science literacy score, school-level mean SES, and school community location). Using data from the 2000, 2003, and 2006 waves of PISA, Han (2013) found that national trends in STEM career expectations vary across STEM subfields as well as students’ science performance levels; there was a downward trend in students’ career expectations for computing and engineering related fields between 2000 and 2006 in several countries (including Belgium, Germany, and Korea), while there was no downward trend in students’ career expectations for health services. In particular, many developed countries including Belgium, Germany, and Korea experienced downward national trends in computing and engineering-related occupational expectations among top performers in science. These findings should be interpreted with caution, however, because it is difficult to fully capture changes in national trends with only six years of PISA data. Using the framework of stratification to measure cross-national differences in education systems, a few studies have investigated the association between features of national education systems and science-related career expectations (Sikora & Pokropek, 2011, 2012a[4_TD$IF]). These studies analyzed data from PISA 2006 and measured the level of stratification in education systems by the number of school types available to 15-year-old students. Sikora and Pokropek (2011) found that the level of stratification in a country’s secondary education system affected boys and girls differently. Specifically, for girls but not boys, higher levels of stratification lowered the likelihood of planning a career in computer science or engineering. In the current study, I contribute to this line of comparative research by incorporating the level of standardization – a crucial feature of educational systems – in the examination of cross-national variation in students’ STEM occupational expectations. Since the early 1980s, national education reform in many countries, including the United States and Germany, has focused on improving the quality and equity of student outcomes by increasing the standardization of the education system, specifically by creating and enforcing centrally prescribed curricular, learning, and assessment standards for all students, teachers, and schools (Ertl, 2006; National Commission on Excellence in Education, 1983; Organisation for Economic Co-operation and Development, 2010; Sahlberg, 2006). It may be that students in highly standardized education systems are exposed to less differentiated math and science knowledge and courses than students in unstandardized education systems, which allow individual freedom of choice in curriculum. In standardized education systems, schools have little autonomy in determining course offerings and course content because curriculum and textbooks are established at the national level. Teachers in standardized systems are expected to teach a centrally prescribed curriculum and use the same textbooks, and all students within a given grade level are expected to meet the same standards. In standardized systems that require universal academic standards, classroom instruction is less likely to be adjusted to match the characteristics of students (Stevenson & Baker, 1991) and teachers are more likely to be unresponsive to individual students (Sheldon & Biddle, 1998). Because exposure to math and science courses in high school affects students’ intent to major in STEM and the likelihood of enrolling in a STEM field of study (Wang, 2013), students in standardized education system may be more likely to expect STEM occupations than students in less standardized education systems. However, a high level of standardization may have a negative association with students’ intent to pursue STEM education and occupations if a math and science curriculum that is less tailored to individual students’ needs, interests, and abilities decreases students’ intrinsic interest in math and science (Sheldon & Biddle, 1998). In the current study, I conduct the first empirical analysis of whether standardized education systems are positively or negatively associated with students’ STEM occupational expectations. 3. Research questions Using insights from comparative studies of educational and occupational expectations among youths, I examine the degree to which the characteristics of national educational systems are associated with cross-national variation in students’ STEM occupational expectations. I pay particular attention to the features of national secondary education systems, especially standardization and stratification. The analysis proceeds in three main steps. First, I examine the association between the degree of standardization of educational systems and students’ STEM occupational expectations. Because occupational preferences of students differ across STEM subfields, I also investigate whether these associations are consistent across subfields. This study focuses on two STEM subfields: (a) computing and

106

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

engineering (CE) and (b) health services (HS) (Organisation for Economic Co-operation and Development, 2012; Sikora & Pokropek, 2012a). I empirically examine whether the association between the degree of standardization and students’ STEM occupational expectations remains consistent across these two STEM subfields. Second, I examine the association between the degree of stratification of educational systems and students’ STEM occupational expectations. As discussed in the literature review, previous studies have indicated that students in more stratified secondary education systems tend to have lower educational and occupational expectations due to greater restriction of students’ options for educational and occupational trajectories. Because the focal STEM occupations in this study require at least a bachelor’s degree at job entry, I expect to find a negative association between stratified systems and students’ STEM occupational expectations. Finally, I assess the degree to which the associations between features of secondary education systems and students’ STEM occupational expectations differ by student ability. High-performing students may want to pursue STEM occupations, which are among the highest-paying and fastest-growing of any occupational areas, no matter how national education systems are organized. Given that low-achieving students are less likely to enroll in math and science courses in systems that maximize individual freedom of choice, low-performing students in standardized systems may have a greater interest in STEM occupations than low-performing students in unstandardized systems. In contrast, however, standardized education systems may make STEM occupations less attractive to low-performing students because classroom instruction is ruled by a nationally prescribed curriculum and does not allow schools or teachers to tailor the science curriculum to meet the needs and interests of individual students. To tease out these potentially contradictory influences, I empirically examine the interaction effects of the features of secondary education systems and student performance levels on STEM occupational expectations. 4. Data and methods This study used data from PISA 2006, which is the latest available large-scale international survey containing information on students’ career expectations, and which assessed scientific literacy as a primary focus in 57 countries. Because PISA asked 15-year-old students what particular type of jobs they expected to have at age 30, the survey data offer a unique opportunity to explore STEM occupational expectations in a comparative perspective. Two countries are excluded from the analyses due to missing data on the dependent variable or key independent variables. First, no data is available for Qatar on the dependent variable in PISA 2006. Second, no data is available on the characteristics of the Albanian education system. As a result, the analytic sample includes 55 countries from PISA 2006. 4.1. Dependent variables The outcome measures for this study are binary variables; each variable indicates whether or not a student expects to have a certain type of STEM-related occupation around age 30. The PISA student questionnaire included a single open-ended question measuring students’ occupational expectations: ‘‘What kind of job do you expect to have when you are about 30 years old?’’ PISA coded student responses to this open-ended question manually and classified them using the International Standard Classification of Occupations 88 (ISCO-88). In this study, STEM-related fields include mathematics, natural science, engineering/computing, and health services. Because prior research has indicated that boys and girls expect to have science careers in different STEM subfields, this study examines expectations for two STEM subfields: (a) computing and engineering (CE) and (b) health services, excluding nursing (HS) (Organisation for Economic Co-operation and Development, 2012; Sikora & Pokropek, 2011). This study focuses on professional STEM occupations that require at least a bachelor’s degree at entry (Elias, 1997). 4.2. Country-level independent variables The main independent variables in this study are country-level indicators of characteristics of national education systems. These variables include: Standardization of educational systems. To measure curriculum standardization, prior research has used a PISA school questionnaire in which principals responded to a series of questions about whether regional or national education authorities are responsible for determining course content and deciding which courses are offered (e.g., Horn, 2009). These studies aggregated the responses to produce a national-level measure of curriculum standardization. However, this aggregate measure does not capture cross-national differences in curriculum standardization. According to the measure, for example, 83 percent of students in the United States attend schools in which the principal reported that regional or national education authorities exert a direct influence on decision making about instructional content, while about 37 percent of students in Japan attend schools in which the principal reported that regional or national education authorities exert a direct influence on decision making about instructional content. The aggregate measure suggests that the level of curriculum standardization in Japan is lower than the level of standardization in any other Western countries, including Australia, Austria, Belgium, Canada, and Germany. This result, however, is not consistent with the features of the Japanese education system: ‘‘Even with recent liberalization, there is still less choice for students in the Japanese curriculum than is typically the case in any Western country’’ (Organisation for Economic Co-operation and Development, 2010, p. 142).

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

107

The current study used the classification of curricular policies developed by Montt (2011). Montt (2011) classified countries into three groups based on their curricular policies: (a) countries in which there is no central government control over the curriculum (coded 0), (b) countries in which regional or local agencies have some ability to adapt a centrally prescribed curriculum (coded 1), and (c) countries in which the central government determines the curriculum (coded 2). Stratification of educational systems. The level of stratification of each national education system was measured by the number of school types available to 15-year-old students in each country, and the age of first selection into different school types or tracks. Among the PISA participating countries, the number of school types ranged from one to five. The age of first selection is a dummy variable for early tracking based on whether students are sorted into different tracks before or after the age of 14 (before age 14 is coded 1 and after age 14 is coded 0). The source of data for both indicators is the OECD report (2007). National economic development indicators. I used two indicators to capture national economic development levels: (a) a measure of the gross domestic product (GDP) per capita (in current U.S. dollars), and (b) an indicator of the level of educational investment, as measured by public education expenditures per student in secondary education as a percent of the GDP per capita. These two indicators were collected from the UNESCO Institute for Statistics and the World Bank. [14_TD$IF]4.3. Other control variables [5_TD$IF]At the student level, I control for gender, grade level, parental educational attainment, parental occupational status, immigration background of both student and parents, whether parents have STEM-field occupations, and science literacy score. At the school level, I control for school’s mean socio-economic status. [15_TD$IF]4.4. Analytic strategy [5_TD$IF]To investigate cross-national variation in students’ STEM occupational expectations and the association between this variation and macro-level features of education systems, this study uses three-level hierarchical generalized linear models (HGLMs) in which students (level 1) are nested within schools (level 2) and within countries (level 3). Because the dependent variable (whether or not a student expects to have a STEM-related occupation around age 30) is binary, this study employs HGLMs in which the level 1 sampling model is a Bernoulli distribution (Raudenbush & Bryk, 2002). The final student weights are normalized at the country level to ensure that each country contributes equally to the analysis (Organisation for Economic Co-operation and Development, 2009b). Model specification for three-level HGLM. Level 1 (Student)

hi jk ¼ log½’i jk =ð1  ’i jk Þ ¼ p0 jk þ p1 jk ðFemaleÞ þ p2 jk ðParent in STEM occpÞ þ p3 jk ðImmigrant backgroundÞ þ p4 jk ðGradeÞ þ    þ pP jk ðStudent predictor PÞ; [17_TD$IF]where w[18_TD$IF]ijk is the probability that a student i in school j in country k expects to have a STEM-related occupation around age 30; h[18_TD$IF]ijk is the log odds that a student i in school j in country k expects to have a STEM-related occupation around age 30. Level 2 (School)

p0 jk ¼ b00k þ b01k ðSchool mean SESÞ þ ro jk p1 jk ¼ b10k ; p21 jk ¼ b20k ; p3 jk ¼ b30k ; p4 jk ¼ b40k ;    ; pP jk ¼ bP0k Level 3 (Country)

b00k ¼ g 000 þ g 001 ðCountry predictor 1Þk    þg 00P ðCountry predictor PÞk þ u00k b10k ¼ g 100 ; b20k ¼ g 200 ; b30k ¼ g 300 ; b40k ¼ g 400 ;    ; bP0k ¼ g P00 Note that, at the country level, N is small enough to be potentially problematic. Therefore, I run two types of models: (a) a model in which the level of curriculum standardization is included as three dummy variables (low, medium, and high), and (b) a model in which the level of curriculum standardization is included as a single ordinal variable. If the former model shows an approximately linear association between the level of standardization and student expectations, I report the results from the latter model. The same procedure is applied to the number of school types available to 15-year-old students as one indicator of stratification. I run the HGLMs using five dummy variables to indicate the number of school types available to 15-year-old students. If the results show an approximately linear association between the number of school

108

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

types and student expectations, I report results from the model in which the number of school types available to 15-year-old students is included as an ordinal variable. To examine possible interactions between student performance and the features of secondary education systems, threelevel HGLMs were run separately by science performance quartile.1 This approach reveals whether the associations between the features of secondary education systems and students’ STEM occupational expectations differ at various levels of academic performance.

5. Results The first step in the analysis was to examine the association between the features of national education systems and STEM occupational expectations using the full sample of students. I fit three-level HGLMs to examine this question, the results of which are displayed in Table 1. The models contain a set of student-level and school-level variables. The primary purpose of including these variables in the model is to control for differences in student and school characteristics. 5.1. Individual characteristics and students’ [19_TD$IF]STEM occupational expectations The analytic results show that several individual-level characteristics are associated with students’ STEM occupational expectations. First, there are female-favorable gender gaps in general STEM occupational expectations (see column 1 in Table 1). The female coefficient for general STEM occupational expectations is 0.047, indicating that a female student is 5 percent more likely than a male student to expect to have a STEM occupation. However, further analysis revealed that the gender gap differs across STEM subfields (see Table 2): compared to boys, girls are less likely to expect to have CE occupations and more likely to expect to have health-related occupations. As shown in Table 2, the female coefficient for CE occupational expectation is 1.358, indicating that a female student is 74 percent less likely than a male student to expect to have a CE occupation. The female coefficient for HS occupational expectation is 0.907, indicting a female student is 2.5 times more likely than a male student to expect to have a HS occupation. Accounting for academic ability does not change these crossnational gender patterns in STEM occupational expectations. In addition, the results show that family background is associated with students’ STEM occupational expectations. As shown in Table 1, the analyses revealed that first- and second-generation immigrant students are more likely to expect STEM occupations than native students. For example, the first-generation immigrant student coefficient for general STEM occupational expectations is 0.389, indicating that these students are 48 percent more likely than native students to expect STEM occupations. Likewise, the second-generation immigrant student coefficient is 0.308, suggesting that second-generation immigrant students are 36 percent more likely than native students to expect STEM occupations. The pattern holds across STEM subfields and by gender. Even after controlling for individual-level factors, both immigrant boys and girls are significantly more likely than their counterparts to expect to have an occupation in each of the STEM subfields. The model results also demonstrate that having a parent working in a STEM field is positively associated with students’ STEM occupational expectations. The parents in a STEM field coefficient for general STEM occupational expectations is 0.326, suggesting that having a parent in a STEM field increases the likelihood of expecting a STEM occupation at age 30 by 39 percent (see column 1 in Table 1). Likewise, having a parent in a CE occupation or a health service occupation increases students’ likelihood of expecting a CE or HS-related occupation, respectively. The positive association between parental occupation and students’ occupational expectations holds for both boys and girls. For example, the CE parent coefficients for boys and girls are 0.629 and 0.417 (not reported in Table 2), respectively. This suggests that having a parent in a CE occupation increases the likelihood of expecting to have a CE occupation by 88 percent for boys and 52 percent for girls. Likewise, the HS parent coefficients for boys and girls are 0.656 and 0.438 (not reported in Table 2), respectively. This indicates that having a parent in an HS occupation increases the likelihood of expecting an HS occupation by 93 percent for boys and 55 percent for girls. 5.2. National education systems and students’ STEM occupational expectations The analyses revealed that the level of standardization in educational systems is negatively associated with expectations of having STEM occupations for girls, but boys’ expectations are constant across levels of standardization. As shown in Table 1, the standardization coefficient for girls is 0.266. This suggests that, for girls, a one-level increase in the standardization of a country’s secondary schools is linked to a 23 percent decrease in the odds of expecting STEM occupations. Further analyses revealed that the negative association between standardization and occupational expectations among girls differs across STEM subfields, while the lack of association among boys remains constant for both subfields. For girls, standardization is not associated with CE occupational expectations, but has a negative association with HS occupational expectations. As seen in Panel B of Table 2, for girls, the standardization coefficient is 0.236, indicating

1 PISA adopted a balanced incomplete block design for assessment, which pairs every block with every other block, but does not include all possible orderings of block pairs. Because of this design, PISA student test scores were estimated as five plausible values. I used the first plausible value for science to create each performance quartile in each country. This means that cut-off scores for performance quartile varied across countries. I used this approach because I assume that students will compare their performance levels with peers in their home country when they consider their future career expectations.

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

109

Table 1 Results of hierarchical Bernoulli logit models to explain variation in expectations for STEM-related occupations. Full sample

Intercept Country-level: Standardization Number of school type Early tracking School-level: School mean SES Individual-level: Female Gender Grade (ref: 10th) 7th 8th 9th 11th or above Parents’ education Primary Lower secondary Upper secondary 1 Upper secondary 2 University Parent’s Occ status Immigration status Second-generation First-generation Parents have STEM occ. Science score N Students (unit of observations) Schools Countries Variance components School level County level

Boys

Girls

b

O.R.

S.E.

b

O.R.

S.E.

b

O.R.

S.E.

0.448

0.639

0.213*

0.702

0.495

0.242**

0.241

0.786

0.221

0.207 0.192 0.101

0.813 0.825 1.106

0.109[7_TD$IF]y 0.071** 0.186

0.164 0.132 0.030

0.849 0.877 1.030

0.121 0.079 0.207

0.266 0.233 0.164

0.767 0.792 1.179

0.112* 0.072** 0.192

0.046

1.048

0.016**

0.064

1.066

0.021**

0.047

1.048

0.020*

0.047

1.048

0.013***

0.221 0.095 0.015 0.073

0.802 0.909 1.015 0.929

0.071** 0.037* 0.018 0.034*

0.338 0.137 0.011 0.033

0.713 0.872 0.989 0.968

0.100** 0.052** 0.026 0.051

0.006 0.041 0.044 0.094

1.006 0.960 1.045 0.911

0.102 0.052 0.025[7_TD$IF]y 0.045*

0.021 0.056 0.030 0.136 0.213 0.001

1.021 1.058 1.030 1.145 1.238 1.001

0.049 0.048 0.052 0.046** 0.047*** 0.000*

0.065 0.117 0.097 0.168 0.250 0.003

1.068 1.124 1.102 1.183 1.285 1.003

0.076 0.073 0.077 0.070* 0.071*** 0.001***

0.005 0.034 0.007 0.144 0.214 0.000

0.995 1.034 1.007 1.155 1.238 1.000

0.064 0.063 0.069 0.060* 0.061*** 0.001

0.308 0.389 0.326 0.005

1.361 1.476 1.386 1.005

0.035*** 0.040*** 0.017*** 0.000***

0.364 0.464 0.395 0.006

1.439 1.590 1.485 1.006

0.049*** 0.058*** 0.024*** 0.000***

0.281 0.331 0.268 0.004

1.324 1.392 1.307 1.004

0.047*** 0.054*** 0.024*** 0.000***

387,752 13,903 55 0.175*** 0.246***

191,931 13,213 55 0.132*** 0.296***

195,817 13,065 55 0.173*** 0.252***

Each column in each panel reports results from one regression. All regressions control for GDP per capita ($1000) and educational expenditure (percent of GDP) at the national level. [8_TD$IF]y p  .10. * p  .05. ** p  .01. *** p  .001. [1_TD$IF]b = Coefficient, O.R. = Odds ratio, S.E. = Standard error.

that an additional level of standardization in a country’s secondary schools is associated with a 21 percent drop in the odds of expecting to have a health service occupation. As described above, the stratification of education systems was measured via two indicators: (a) the number of school types available to 15-year-old students and (b) the presence of early tracking into different school types (implemented before age 14). The analytical results show that a higher number of school types are linked to lower student STEM occupational expectations, but early tracking is not associated with STEM expectations. As shown in Table 1, the coefficient for number of school types is -0.192. The odds ratio for number of school types is 0.825, indicating that each additional school type available to 15-year-old students is associated with a 17 percent drop in the odds of expecting STEM occupations. This association between the stratification of education systems and students’ STEM occupational expectations differs across STEM subfields: while there is no association between number of school tracks and CE occupational expectations, number of school tracks is negatively associated with occupational expectations for health services. This negative association between the degree of stratification in education systems and health services occupational expectations does not vary by gender. As shown in Panel B of Table 2, for boys, the number of school types coefficient is 0.228, indicating that each additional school type is associated with a 20 percent decrease in the odds of expecting an HS occupation. For girls, the school types coefficient is 0.265, indicating that each additional school type is linked to a 23 percent decrease in the odds of expecting an HS occupation. Additional analyses examined whether the association between features of secondary education systems and students’ STEM occupational expectations differed across science performance quartiles. The first column in Table 3 displays the coefficients for the features of secondary education systems for the top quartile of students in each country; the second, third, and fourth columns show results for students in the upper-middle, lower-middle, and lowest quartiles, respectively. Table 3 reveals that higher levels of standardization in secondary education are associated with lower STEM occupational expectations, particularly in the health services field. This negative association is stronger for students at the bottom of the

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

110

Table 2 Results of hierarchical Bernoulli logit models to explain variation in expectations for Computing and Engineering (CE) and health service-related occupations (excluding nursing). Full sample

Panel A: CE Intercept Individual-level: Female Gender Country-level: Standardization Number of school type Early tracking Panel B: HS Intercept Individual-level: Female Gender Country-level: Standardization Number of school type Early tracking

Boys

Girls

b

O.R.

S.E.

b

O.R.

S.E.

b

O.R.

S.E.

1.593

0.203

0.201***

1.638

0.194

0.215***

2.858

0.057

0.263***

1.358

0.257

0.021***

0.123 0.071 0.095

0.885 0.931 1.100

0.098 0.063 0.167

0.148 0.051 0.095

0.862 0.951 1.099

0.103 0.067 0.177

0.074 0.105 0.096

0.929 0.900 1.101

0.122 0.079 0.210

2.113

0.121

0.235***

2.385

0.092

0.298***

1.119

0.326

0.228***

0.907

2.477

0.020***

0.170 0.273 0.097

0.844 0.761 1.102

0.118 0.077*** 0.202

0.092 0.228 0.063

0.912 0.796 0.939

0.145 0.093* 0.247

0.236 0.265 0.145

0.790 0.767 1.157

0.113* 0.074*** 0.195

Each column in each panel reports results from one regression. All regressions control for all the student-level, school-level, and national-level variables reported in Appendix A (descriptive statistics table). [9_TD$IF]y p  .10. * p  .05. ** p  .01. *** p  .001. [1_TD$IF]b = Coefficient, O.R. = Odds Ratio, S.E. = Standard Error

academic performance distribution than for students at the top. For top-quartile students, the level of standardization is not associated with students’ HS occupational expectations. However, for bottom-quartile students (Panel C in Table 3), the standardization coefficient is 0.239, indicating that a one-level increase in the standardization of a country’s secondary school system is linked to a 21 percent decrease in the odds of expecting an HS occupation. The results shown in Panel C of Table 3 suggest that higher levels of stratification in education systems – as measured by the number of school types available to 15-year-old students – are tied to lower expectations of having health service occupations across all performance quartiles. For top-quartile students, the school types coefficient is 0.116, indicating that each additional school type is associated with an 11 percent decrease in the odds of expecting an HS occupation. For uppermiddle quartile students, the school types coefficient is 0.178, indicating that each additional school type is associated with a 14 percent decrease in the odds of expecting an HS occupation. For lower-middle quartile students, the school types

Table 3 Results of hierarchical Bernoulli logit models to explain variation in aspirations for stem-related occupations by performance quartile. Top-quartile

b

O.R.

Panel A: STEM Intercept 0.018 0.982 Standardization 0.092 0.912 Number of school type 0.082 0.922 Early tracking 0.034 1.035 Panel B: CE Intercept 1.163 0.313 Standardization 0.058 0.944 Number of school type 0.060 0.941 Early tracking 0.191 1.211 Panel C: Health services (excluding nursing) Intercept 1.807 0.164 Standardization 0.035 0.965 Number of school type 0.116 0.890 Early tracking 0.088 0.916

Upper-middle-quartile

Lower-middle-quartile

Bottom-quartile

S.E.

b

O.R.

S.E.

b

O.R.

S.E.

b

O.R.

S.E.

0.237 0.104 0.070 0.182

0.418 0.140 0.098 0.011

0.659 0.869 0.907 0.989

0.195* 0.088 0.060 0.160

0.728 0.202 0.160 0.167

0.483 0.817 0.852 1.182

0.200*** 0.089* 0.060** 0.157

1.128 0.194 0.255 0.268

0.324 0.824 0.775 1.307

0.241*** 0.113[7_TD$IF]y 0.078** 0.203

0.272*** 0.111 0.074 0.194

1.624 0.060 0.006 0.120

0.197 0.941 0.994 1.127

0.258*** 0.111 0.075 0.199

1.788 0.133 0.036 0.096

0.167 0.876 0.964 1.100

0.253*** 0.108 0.071 0.188

1.975 0.055 0.210 0.273

0.139 0.946 0.810 1.314

0.329*** 0.141 0.098* 0.253

0.244*** 0.102 0.068[7_TD$IF]y 0.178

2.010 0.186 0.178 0.112

0.134 0.831 0.837 0.894

0.241*** 0.107[7_TD$IF]y 0.071* 0.193

2.305 0.198 0.243 0.143

0.100 0.820 0.784 1.154

0.263*** 0.115[7_TD$IF]y 0.076** 0.201

2.788 0.239 0.368 0.331

0.062 0.787 0.692 1.393

0.293*** 0.131[7_TD$IF]y 0.091* 0.237

Each column in each panel reports results from one regression All regressions control for all the student-level, school-level, and national-level variables reported in Appendix A (descriptive statistics table). [9_TD$IF]y p  .10. * p  .05. ** p  .01. *** p  .001. [1_TD$IF]b = Coefficient, O.R. = Odds Ratio, S.E. = Standard Error

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

111

coefficient is 0.243, indicating that each additional school type is linked to a 22 percent decrease in the odds of expecting an HS occupation. For bottom-quartile students, the school types coefficient is 0.368, indicating that each additional school type is linked to a 31 percent decrease in the odds of expecting an HS occupation. In sum, the results show consistent negative associations between the number of school types available to 15-year-old students and HS occupational expectations across different science ability levels. Further, as shown in Panel B of Table 3, higher levels of stratification are linked to lower CE occupational expectations among students in the bottom performance quartile. 6. Conclusions Using data from a large-scale international survey of student achievement, PISA 2006, this study examined the extent to which features of secondary education systems are associated with STEM occupational expectations across countries. The study employed the standardization-stratification framework to classify national educational systems. Standardization refers to the degree to which school curricula are standardized nationwide and stratification indicates the degree to which students are sorted into school types that are valued differently by higher education institutions and labor markets. First, the analyses revealed that associations between features of secondary education systems and STEM occupational expectations differed across STEM subfields. Overall, higher levels of stratification in secondary education were linked to lower student expectations for health service occupations. In contrast, students’ computing and engineering (CE) occupational expectations were not associated with any of the characteristics of secondary education systems measured in the current study—the standardization of curriculum, the number of school types available to 15-year-old students, or early tracking. Second, gender differences emerged with respect to the associations between the features of national education systems and STEM-related occupational expectations. Higher levels of curricular standardization were associated with lower health service occupational expectations among girls, but boys’ HS expectations remained constant across different levels of standardization. In contrast, the associations between stratified education systems and STEM occupational expectations were similar for boys and girls. Levels of stratification in a country’s secondary education system were not associated with either boys’ or girls’ CE occupational expectations, but were linked to lower health service occupational expectations for both boys and girls. The analytical results also showed that the association between the features of national education systems and certain types of STEM occupational expectations differed across student performance levels. Specifically, the negative association between standardized systems and health service occupational expectations was stronger for low performers in science than for top performers in science. However, the negative association between stratified systems and heath service occupational expectations remained consistent across academic performance levels. The findings of the current research suggest that policymakers and researchers must pay attention to macro-level features of education systems as well as individual- and school-level factors as they seek to explain students’ STEM occupational expectations. In Korea, for example, educational researchers and professors in STEM fields have argued that the standardization of curriculum and instruction (which precludes the influence of a student’s interest) in secondary science education is leading students to avoid studying in the fields of science and engineering when they attend college (Korea Research Institute for Vocational Education & Training, 2002). Like Korea, Japan has also experienced a ‘‘rikei banare’’ (flight from science) in its education system, a phenomenon in which students avoid the study of science, engineering, and mathematics (Gordon, 2009). The Japanese government has expressed concerns about flagging student interest in studying natural science or engineering (National Science Board, 2008). Japanese students are ranked at the top in cross-national comparisons of mathematics and science achievement. However, Japanese students have reported taking less pleasure in and having a lower level of interest in learning mathematics and science as well as having low levels of interest in pursuing science-related careers (Organisation for Economic Co-operation and Development, 2007, 2013). Further research is needed to determine whether there is a casual relationship between highly standardized education systems and STEM educational and occupational expectations such that highly standardized education systems decrease STEM occupational expectations and encourage a flight from science. Addressing these questions with cross-sectional data (such as the PISA data used in this study) requires at least two time points of data, one before and one after changes in the standardization of an education system. For example, Japan reformed its education system in 2006, decreasing standardization by reducing uniformity and specificity, and moving away from top-down management (Organisation for Economic Co-operation and Development, 2010). Thus one possible way to test whether these changes in curriculum standardization affected Japanese students’ STEM occupational expectations is to compare students’ STEM expectations before and after 2006. If the next PISA wave (i.e., PISA 2015) includes students’ expected occupations at age 30, researchers should use the data to examine whether decentralization reform in Japan has changed students’ STEM occupational expectations. Given the concerns about top-performing students’ flight from mathematics, the natural sciences, and engineering, policymakers and educational researchers in many countries are interested in fostering students’ interest in STEM education and occupations (Organisation for Economic Co-operation and Development, 2009b). The findings of the current research suggest that the factors that make STEM occupations attractive might differ across STEM subfields and by gender as well as across student performance levels. For example, the pattern found in this research project – that the associations between the features of education systems and students’ STEM occupational expectations differ across

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

112

student performance levels – suggests that the factors that make STEM occupations attractive might differ across student performance levels. This study found that features of national education systems were not associated with the STEM occupational expectations of top academic performers. In addition to national education systems, researchers have asserted that macro-social and macro-economic conditions such as the status of scientists at the societal level, their earnings relative to other professional occupations, and their employment prospects can influence student interest in STEM careers (for example, Stephan, 2012; Xie & Killewald, 2012). Thus, more attention should be focused a wide range of economic and social factors that may make STEM occupations attractive in the eyes of students who perform well in math and science. This study has several limitations. Because of the limited data on career expectations in large-scale international surveys, it is difficult to test the validity of the outcome measure in the current study. PISA does include one alternative measure of career expectations: an index of future-oriented science motivations (Organisation for Economic Co-operation and Development, 2009a). This index was derived from student responses about their interest in studying ‘‘broad science’’ and working on ‘‘broad science’’ projects. However, because it is not clear whether students across countries interpret the phrase ‘‘broad science’’ in the same way, I did not use the measure. This study utilized somewhat limited measures to capture the features of national education systems, and further research is needed to examine the degree to which the features of math and science education shape students’ STEM career expectations. Due to the limitations of the cross-sectional data used in the current study, it is difficult to determine exactly why students in countries with higher levels of both curricular standardization and educational stratification are less likely to expect to have STEM occupations. In particular, it is difficult to explain why the associations between characteristics of national education systems and STEM career expectations differ across STEM subfields. This question remains open for future research. Further, many aspects of the gender differences in the association between the features of national education systems and STEM career expectations are important topics for future research. Acknowledgments This research was supported by a grant from the American Educational Research Association which receives funds for its ‘‘AERA Grants Program’’ from the National Science Foundation under grant no. DRL-0941014. Opinions reflect those of the author and do not necessarily reflect those of the granting agencies. Appendix A Descriptive statistics PISA 2006 Mean Occupational expectations STEM, general Computing and Engineering (CE) Health services excluding nursing Student characteristics Grade in school 7th or lower 8th 9th 10th 11th or higher Female gender Student ability Science Family background Parents’ education None Primary Lower secondary Upper secondary 1 Upper secondary 2 University Parents have STEM occupation Parent’s occupational status Immigration status Native Second-generation immigrant First-generation immigrant School characteristics School mean SES

Std. Dev.

0.28 0.10 0.11

0.45 0.31 0.32

0.01 0.05 0.34 0.51 0.08 0.51

0.01 0.05 0.34 0.51 0.08 0.50

477.46

104.50

0.02 0.06 0.11 0.08 0.28 0.44 0.17 47.54

0.16 0.23 0.32 0.27 0.45 0.50 0.38 17.03

0.92 0.05 0.04

0.28 0.21 0.19

0.27

0.77

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

113

Appendix A (Continued ) PISA 2006

National economic development GDP per capita ($1,000) Educational expenditure (percent of GDP) Characteristics of national educational system Standardization Low Medium High Number of school types Early tracking

Mean

Std. Dev.

24596 19.99

23307 6.95

0.11 0.38 0.52 2.33 0.33

0.31 0.48 0.50 1.22 0.47

Appendix B STEM-related careers 2100 physical, mathematical and engineering science professionals 2110 physicists, chemists and related professionals 2111 physicists and astronomers 2112 meteorologists 2113 chemists 2114 geologists and geophysicists [incl. geodesist] 2120 mathematicians, statisticians, etc professionals 2121 mathematicians, etc professionals 2122 statisticians [incl. actuary] 2130 computing professionals 2131 computer systems designers and analysts [incl. software engineer] 2132 computer programmers 2139 computing professionals not elsewhere classified 2140 architects, engineers etc professionals 2141 architects town and traffic planners [incl. landscape architect] 2142 civil engineers [incl. construction engineer] 2143 electrical engineers 2144 electronics and telecommunications engineers 2145 mechanical engineers 2146 chemical engineers 2147 mining engineers, metallurgists, etc, professionals 2148 cartographers and surveyors 2149 architects, engineers, etc professionals not elsewhere classified [incl. consultant] 2200 life science and health professionals 2210 life science professionals 2211 biologists, botanists, zoologists, etc professionals 2212 pharmacologists, pathologists etc profess. [incl. biochemist] 2213 agronomists, etc professionals 2220 health professionals (except nursing) 2221 medical doctors 2222 dentists 2223 veterinarians 2224 pharmacists 2229 health professionals except nursing not elsewhere classified 2230 nursing and midwifery profess. [incl. registered nurses, midwives] Careers in computing and engineering 2130 computing professionals 2131 computer systems designers and analysts [incl. software engineer] 2132 computer programmers 2139 computing professionals not elsewhere classified 2140 architects, engineers etc professionals 2141 architects town and traffic planners [incl. landscape architect] 2142 civil engineers [incl. construction engineer] 2143 electrical engineers

114

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

2144 electronics and telecommunications engineers 2145 mechanical engineers 2146 chemical engineers 2147 mining engineers, metallurgists etc professionals 2149 architects engineers etc professionals not elsewhere classified [incl. consultant] Careers in health services (excluding nursing) 2212 pharmacologists, pathologists etc profess. incl. biochemist 2220 health professionals (except nursing) 2221 medical doctors 2222 dentists 2223 veterinarians 2224 pharmacists 2229 health professionals except nursing nec References Allmendinger, J. (1989). Career mobility dynamics: A comparative analysis of the United States, Norway, and West Germany. Berlin: Max-Planck-Institut fu¨r Bildungsforschung. Buccheri, G., Gurber, N. A., & Bruhwiler, G. (2011). The impact of gender on interest in science topics and the choice of scientific and technical vocations. International Journal of Science Education, 33(1), 159–178. Buchmann, C., & Dalton, B. (2002). Interpersonal influences and educational aspirations in 12 countries: The importance of institutional context. Sociology of Education, 75(2), 99–122. Buchmann, C., & Park, H. (2009). Stratification and the formation of expectations in highly differentiated educational systems. Research in Social Stratification and Mobility, 27(4), 245–267. Correll, S. J. (2001). Gender and the career choice process: The role of biased self-assessments. The American Journal of Sociology, 106(6), 1691–1730. Elias, P. (1997). Occupational classification (ISCO-88): Concepts, methods, reliability, validity and cross-national comparability. OECD labour market and social policy occasional papers, (No. 20). hhttp://dx.doi.org/10.1787/304441717388i. Ertl, H. (2006). Educational standards and the changing discourse on education: The reception and consequences of the PISA study in Germany. Oxford Review of Education, 32(5), 619–634. Gordon, E. (2009). Winning the global talent showdown: How businesses and communities can partner to rebuild the jobs pipeline. Oakland, CA: Berrett-Koehler Publishers. Han, S. W. (2013). Schooling, labor markets and STEM occupational expectations: A comparative perspective. Madison, WI: University of Wisconsin–Madison (Ph.D. dissertation). Horn, D. (2009). Age of selection counts: A cross-country analysis of educational institutions. Educational Research & Evaluation, 15(4), 343–366. Kerckhoff, A. C. (2001). Education and social stratification processes in comparative perspective. Sociology of Education, 74, 3–18. Kjærnsli, M., & Lie, S. (2011). Students’ preference for science carers: International comparisons based on PISA 2006. International Journal of Science Education, 33(1), 121–144. Korea Research Institute for Vocational Education & Training. (2002). The avoidance of science and engineering fields of high school students in attending colleges and universities in Korea. Seoul, Korea: Korea Research Institute for Vocational Education & Training. Martin, M. O., Mullis, I. V. S., Gonzalez, E. J., & Chrostowski, J. J. (2004). TIMSS 2003 international science report: Findings from IEA’s trends in international mathematics and science study at the fourth and eighth grades. Chestnut Hill, MA: TIMSS & PIRLS International Study Center, Boston College. McDaniel, A. (2010). Cross-national gender gaps in educational expectations: The influence of national-level gender ideology and educational systems. Comparative Education Review, 54(1), 27–50. Montt, G. (2011). Cross-national differences in educational achievement inequality. Sociology of Education, 84(1), 49–68. Morgan, S. L., Gelbgiser, D., & Weeden, K. A. (2013). Feeding the pipeline: Gender, occupational plans, and college major selection. Social Science Research, 42(4), 989–1005. Mullis, I. V. S., Martin, M. O., Gonzalez, E. J., & Chrostowski, J. J. (2004). TIMSS 2003 international mathematics report: Findings from IEA’s trends in international mathematics and science study at the fourth and eighth grades. Chestnut Hill, MA: TIMSS & PIRLS International Study Center, Boston College. National Commission on Excellence in Education. (1983). A nation at risk: The imperative for educational reform. Washington, DC: US Department of Education. National Science Board. (2008). Science and engineering indicators 2008. Washington, DC: National Science Foundation. Organisation for Economic Co-operation and Development. (2007). PISA 2006 science competencies for tomorrow’s world. Paris: Organisation for Economic Cooperation and Development. Organisation for Economic Co-operation and Development. (2012). Education at a glance 2012: OECD indicators. Paris: Organisation for Economic Co-operation and Development. Organisation for Economic Co-operation and Development. (2013). PISA 2012 results: Ready to learn: Students’ engagement, drive and self-beliefs. Paris: Organisation for Economic Co-operation and Development. Organisation for Economic Co-operation and Development. (2009a). PISA 2006 technical report. Paris: Organisation for Economic Co-operation and Development. Organisation for Economic Co-operation and Development. (2009b). PISA data analysis manual: SPSS second edition. Paris: Organisation for Economic Co-operation and Development. Organisation for Economic Co-operation and Development. (2009c). Top of the class: High performers in science in PISA 2006. Paris: Organisation for Economic Cooperation and Development. Organisation for Economic Co-operation and Development. (2010). Strong performers and successful reformers in education: Lessons from PISA for the United States. Paris: Organisation for Economic Co-operation and Development. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods ((2nd ed.)). Thousand Oaks: Sage Publications. Riegle-Crumb, C., Moore, C., & Ramos-Wada, A. (2010). Who wants to have a career in science or math? Exploring adolescents’ future aspirations by gender and race/ethnicity. Science Education, 95(3), 458–476. Sahlberg, P. (2006). Education reform for raising economic competitiveness. Journal of Educational Change, 7(4), 259–287. Shavit, Y., Mu¨ller, W., & Tame, C. (Eds.). (1998). From school to work: A comparative study of educational qualifications and occupational destinations. Oxford: Clarendon Press. Sheldon, K. M., & Biddle, B. J. (1998). Standards, accountability, and school reform: Perils and pitfalls. Teachers College Record, 100(1), 164–180. Sikora, J., & Pokropek, A. (2011). Gendered career expectations of students: Perspectives from PISA 2006. Paris: Organisation for Economic Co-operation and Development (OECD Education working papers no. 57). Sikora, J., & Pokropek, A. (2012a). Gender segregation of adolescent science career plans in 50 countries. Science Education, 96(2), 234–264. Sikora, J., & Pokropek, A. (2012b). Intergenerational transfers of preferences for science careers in comparative perspective. International Journal of Science Education, 36(16), 2501–2527.

S.W. Han / International Journal of Educational Research 72 (2015) 103–115

115

Stephan, P. (2012). How economics shapes science. Cambridge, MA: Harvard University Press. Stevenson, D. L., & Baker, D. P. (1991). State control of the curriculum and classroom instruction. Sociology of Education, 64(1), 1–10. Tai, R. H., Liu, C. Q., Maltese, A. V., & Fan, X. (2006). Planning early for careers in science. Science, 312, 1143–1144. Wang, X. (2013). Why students choose STEM majors: Motivation, high school learning, and postsecondary context of support. American Educational Research Journal, 50(5), 1081–1121. Xie, Y., & Killewald, A. (2012). Is American science in decline? Cambridge, MA: Harvard University Press. Xie, Y., & Shauman, K. A. (2003). Women in science: Career processes and outcomes. Cambridge, MA: Harvard University Press.