The study of attitudes toward science has received considerable attention in recent decades. Researchers, educators and policy makers are expressing increasing concern at the decreasing interest of young people in pursuing scientific careers, and these concerns are being reinforced by research indicating the scientific ignorance of the general public (Osborne & Dillon, 2008; Osborne, Simon, & Collins, 2003). While the lack of interest in science often manifests itself during secondary school when young people must choose the subjects they will study, most students have already excluded the choice of scientific or technological subjects long before, i.e., during their years in primary school (Osborne & Dillon, 2008, Tai, Qi Liu, Maltese, & Fan, 2006; Young & Kellogg, 1993). Science education should therefore pay explicit attention to improving students' interest in and attitudes toward science, and this should take place beginning at the primary school level (Haney, Czerniak, & Lumpe, 1996; Tobin, Tippins, & Gallard, 1994; Van Driel, Beijaard, & Verloop, 2001).
Many initiatives across the world have focused on increasing the scientific knowledge, inquiry skills, and attitudes of primary students by allocating more time to science education. Some of these initiatives seem to have been reasonably effective, particularly with regard to children's enjoyment of science projects (Duschl, Schweingruber, & Shouse, 2007). However, a major problem remains that cannot be solved by simply allocating more time to science in primary education. The problem is that primary school teachers are not adequately trained to teach science. The present article is based on the contention that, to achieve sustainable improvements in primary science education, it is crucial for primary teachers to develop their own positive attitudes toward science.
Various studies (e.g., Harlen & Holroyd, 1997; Palmer, 2004; Shrigley, 1983; Trumper, 1998) have shown a generally low level of scientific and technological literacy among preservice and inservice primary school teachers, and these teachers generally tend to have negative attitudes toward science. The negative attitudes often originate from negative experiences that teachers had during their own primary and secondary education, and these attitudes persist during their preservice teacher training (Jarret, 1999; Mulholland & Wallace, 1996; Palmer, 2001; Tosun, 2000; Young & Kellogg, 1993). Studies investigating the attitudes of primary teachers toward science have indicated that teachers with less positive attitudes share a number of characteristics. They have lower confidence and self-efficacy beliefs about teaching science (Skamp, 1991; Tosun, 2000; Yates & Goodrum, 1990); they spend less time discussing and teaching these topics in their classrooms (Goodrum, Hackling, & Rennie, 2001; Harlen & Holroyd, 1997); they rely more on standardized methods and top–down instruction (Appleton & Kindt, 1999; Harlen & Holroyd, 1997; Jarvis & Pell, 2004; Plonczak, 2008); and they are less able to stimulate the attitudes of their students (Harlen & Holroyd, 1997; Jarvis & Pell, 2004; Osborne et al., 2003; Van Driel et al., 2001; Weinburgh, 2007). On the positive side, research has shown that when teachers gain greater confidence and self-efficacy and a more positive attitude through continuing education efforts, they subsequently teach science in a better manner and are able to improve the attitudes of their students in this area (Osborne & Dillon, 2008; Osborne et al., 2003).
The above discussion implies that explicit attention to improving the attitudes of preservice and inservice primary teachers toward science is of fundamental importance to the professionalization of teachers within the domain of science (Bleicher, 2007; Haney et al., 1996; Johnston & Ahtee, 2006). Educational change projects, research projects, and preservice or inservice training programs should therefore investigate the attitudes of teachers. Although such attitudes have been investigated in a range of scientific studies in many countries, scientific progress in this field has been slow due to several major theoretical and methodological issues (Bennett, Rollnick, Green, & White, 2001; Gardner, 1995; Kind, Jones, & Barmby, 2007; Osborne et al., 2003).
Most importantly, the concept of an attitude toward science is often poorly articulated, both in research and in educational change projects (Barmby, Kind, & Jones, 2008; Bennett et al., 2001; Coulson, 1992; Osborne et al., 2003; Pajares, 1992). Many studies provide incomplete definitions (or no definition at all) for the construct of attitude, fail to explicate the components of attitude that they measured, or do not distinguish between attitudes toward science and other related concepts (e.g., opinions or motivation). It is therefore difficult to determine exactly what is being measured or investigated (Blalock et al., 2008). Furthermore, most researchers fail to justify their choices to measure certain dimensions or objects of attitude. These choices often seem to be based on pragmatic or convenience arguments. Because of the poor articulation of the construct of attitude toward science, at least two profound ambiguities are present in scientific research on this topic to date.
The first ambiguity concerns the difference between “attitudes toward science” and “scientific attitudes,” a distinction that has also been described elsewhere (Gardner, 1975; Osborne et al., 2003). Many projects and studies that claim to improve or measure attitude actually focus primarily on improving or measuring “scientific attitudes.” These attitudes consist of features that characterize scientific thinking, such as curiosity, being critical about all statements, a demand for verification, or a respect for logic. Although such scientific attitudes are very important to the advancement of science in primary school, they do not constitute what should be considered as attitudes toward science. The latter type of attitudes indicate a variety of thoughts, values, feelings, and behaviors concerning such matters as an individual's thoughts about the level of difficulty characteristic of the sciences, the value attached to science for society, feelings of pleasure or interest with regard to science, and the desire or intention to learn more about science. In the present article, we focus on research that explicitly focuses on the attitudes of teachers toward science.
The second important ambiguity that we would like to address concerns the difference between the personal attitudes of teachers toward science and their professional attitudes toward the teaching of science. Personal attitude toward science refers to the attitude of a person as a citizen, independent of profession (in this case, teaching at primary school level). Examples of personal attitude include beliefs about the historical or economic relevance of science for society or daily life and the general interest in or affect toward staying informed about science through various media. In contrast, teachers' professional attitudes toward teaching science in primary school involve the beliefs and feelings that they may have with respect to teaching these topics within the school context. Examples include beliefs about the appropriateness and importance of science for children at the primary school level or feelings of joy or anxiety with regard to teaching these topics. Many studies combine aspects of the personal and professional attitudes of teachers. We specifically address this conceptual difference in the present article.
In addition to (and probably also because of) the poor theoretical definition of what constitutes primary teachers' attitudes toward science, many researchers have used poorly designed measurement instruments or inadequate methods of analysis (for a review of these issues, see Blalock et al., 2008; Coulson, 1992; Gardner, 1995; Reid, 2006). In addition, most researchers failed to conduct pilot testing, validation, and evaluation of their measurement instrument according to current psychometric standards (Blalock et al., 2008; Coulson, 1992; Gardner, 1995; Reid, 2006). These methodological problems have at least two important consequences. First, the results of these studies are difficult to interpret, as it is not clear what was actually measured. Second, it is difficult or impossible to compare or replicate the results of different studies on attitudes toward science when the various dimensions of attitude that were the objective of the intervention were not explicated (Pardo & Calvo, 2002).
To remedy the issues described above, the present article focuses on improving the fundamental problem of these methodological issues, namely the underlying theoretical construct. We do so by describing the development of a new comprehensive theoretical framework for the concept of primary teachers' attitudes toward science. The components for the framework were derived from an extensive review of the concepts of attitude used by existing studies that claimed to investigate the construct of interest. In doing so, this review of concepts aimed to elucidate and structure the often incomplete construct of primary teachers' attitudes toward science. It should be noted that, because of the theoretical and methodological issues described above, this review was not intended to compare or tally the results of these studies. Instead, it was meant to derive, explicate, and structure the range of underlying (implicit) definitions of primary teachers' attitudes toward science as used in these studies. In addition, in the second half of this article we evaluated the derived attitude components in light of general psychological attitude theories and other theoretical considerations that we deemed relevant to the construction of the framework. On the basis of this theoretical elaboration, we developed our new comprehensive conceptual framework for the underlying dimensions of primary teachers' attitude toward science. This framework is presented in the last part of the article.
To our knowledge, no such theoretical elaboration has been conducted in this domain before. The useful and important overviews that were presented by Blalock and colleagues (2008) and by Osborne and colleagues (2003) focused on the attitudes of students toward science (in school) and were intended to identify important flaws in the methodology of these studies. On the other hand, our focus was not on methodological issues, but primarily on evaluating the theoretical construct of primary teachers' attitude toward science. However, we do discuss several methodological problems when they affected the evaluation of the concepts that we derived. For extensive reviews of methodological issues related to the study of attitudes toward science, we refer to Blalock et al. (2008) and Osborne et al. (2003). In the remainder of this section, we introduce the general psychological concept of attitude and provide some important considerations that serve to clarify the measurable objects of primary teachers' attitudes toward science.
Definition of Attitude
Although definitions of the construct of attitude vary greatly within the literature, there is general agreement that attitude is a psychological tendency to evaluate an object in terms of favorable or unfavorable attribute dimensions such as good/bad or positive/negative (Ajzen, 2001; Eagly & Chaiken, 1993). This evaluative quality is central to the definition of attitude, and it distinguishes the concept from beliefs or opinions. Attitudes are quite stable (once formed they are hard to change), and they are highly dependent upon context (i.e., attitudes toward to related objects can be contradictory, e.g., attitude toward sustainable energy and attitudes toward nuclear energy plants) (Ramey-Gassert, Shroyer, & Staver, 1996; Schoon & Boone, 1998).
Furthermore, attitude is not a single unitary concept; it is a construct consisting of multiple dimensions and subcomponents. The separate evaluations of each of these subcomponents contribute in varying degrees toward the overall attitude toward the object (Ajzen, 2001). As Pajares (1992) describes, “When clusters of beliefs are organized around an object or situation and predisposed to action, this holistic organization becomes an attitude” (p. 314). This implies that measuring an individual's attitude toward an object should consist of measuring the various dimensions and subcomponents of attitude toward that specific attitude object (Ajzen, 2001; Schwarz, 2008). To date, however, there is no consensus on the number or identity of the dimensions and subcomponents that constitute the construct of attitude toward science.
The overall psychological construct of attitude is often divided into three components: cognition, affect, and behavior (Eagly & Chaiken, 1993). For our literature review, we used this tripartite division as a starting point from which to categorize the various concepts of attitude toward science that we derived from previous research. The cognitive component of attitude encompasses the evaluative thoughts and beliefs that a person has about the attitude object. These beliefs range from a positive to a negative evaluation of attributes along a continuum. The second component of attitude is affect. This component consists of feelings and moods that a person experiences in relation to the attitude object. A positive attitude is characterized by the experience of positive reactions (physical or otherwise) and emotions when confronted with the object, whereas a negative attitude is accompanied by negative affective reactions. The third component of this tripartite division of attitude is behavior, which constitutes the behavioral responses or actions of a person when confronted with the attitude object. This response can be either overt (with the person actually acting out the behavioral response or action) or covert (with the person intending to act out the behavior, although the action has yet to take place). In addition to referring to the attitudinal responses a person may have, these three components of attitude refer to the different processes underlying the formation of an attitude. In the literature, these components have also been labeled as dimensions, classes, or categories (Eagly & Chaiken, 1993; Klop, 2008).
Objects of Attitude Toward Science
In addition to the need for theoretical and methodological distinctions among various components or dimensions of attitude, it is important to distinguish among various objects of attitude. The attitude object is the entity about which an attitudinal evaluation is made; it is the object of evaluation. Although this seems a simple description at first glance, many research articles lack clarity on this subject. In many studies, multiple objects of attitude are measured (e.g., general attitude toward science, attitudes toward science in secondary school, and attitudes toward the effective teaching of science) as if they constitute a single attitude object. In many cases, scores on questionnaire items measuring these different attitude objects are added together to yield one overall score of attitude toward science. This score, however, represents a variety of underlying concepts and is therefore impossible to interpret. As observed by Osborne and colleagues (2003), “if there is no single construct underlying a given scale, then there is no reason served by applying a summated rating technique to produce a unitary score” (p. 1058). One common example of the blending of attitude objects occurs when items on teachers' personal attitude toward science are combined with items regarding their professional attitude toward teaching science.
Personal and Professional Attitude
There are multiple studies that address the distinction between personal and professional attitude. However, most of these studies make this distinction only in the introduction of the study but fail to make a clear conceptual and methodological distinction in their measurement of attitude. To our knowledge, only four studies have made such an explicit distinction between the personal and professional attitudes of primary teachers toward science (Asma, Walma van der Molen, & van Aalderen-Smeets, in press; Atwater, Gardner, & Kight, 1991; Harty, Samuel, & Anderson, 1991; Koballa, 1986). Atwater and colleagues (1991) measure both attitude objects with separate instruments but provide minimal information on the constructs of attitude that they measure. Although the article does include references to the used measurement instruments, we were unable to locate the literature and analyze the content of the questionnaires.
The study by Harty and colleagues (1991) measures the attitudes of primary teachers toward science and toward the teaching of science before and after inservice training sessions. The results show that both attitudes improved after training. They also show a positive correlation between the two types of attitude. However, the validity of the measurements can be questioned. Because no information is provided concerning the instrument that was used to measure professional attitude, it is not possible to analyze what was actually measured. Furthermore, the Science Attitude Scale (SAS) (Shrigley, 1974) that the researchers used to measure personal attitude, actually contains items addressing both personal and professional attitude.
Koballa (1986) compared the predictive power of personal attitude toward science with professional attitude toward teaching science on teachers' intention to teach science. As expected, attitude toward teaching science was more strongly correlated with intention to teach than personal attitude toward science. However, personal attitude and professional attitude were measured with inherently different types of measurement instruments (i.e., Likert scales as opposed to semantic differential scales), which undermines the validity of the comparison.
A qualitative focus-group study conducted by Asma and colleagues (in press) investigated the perceptions of primary teachers regarding their own personal and professional attitudes toward science. Results from this study show that preservice and inservice teachers do indeed perceive a distinction between their personal and professional attitude toward science and that it is possible for these attitudes to develop independently. Although relationships between the two attitude objects were put forward, teachers reported that this relationship could work in two directions. Some teachers observed that their professional attitude toward teaching science were influenced by their personal attitude, whereas others reported the reverse.
Although the studies reported above raise conceptual and methodological concerns, their results do indicate that primary teachers' personal and professional attitudes toward science may represent two underlying objects of attitude. We treated these two attitudes separately in our review so that potential differences in the underlying dimensions of the attitude constructs could surface. This is necessary to be able to explore the distinction and the relationship between the two attitude objects in future projects.
Concept of Science
The concept of science as used in this article refers to the broad domain of natural and life sciences, including the concepts of “science and technology,” and “science, technology, engineering and mathematics (STEM).” Science as used in this article encompasses a broad domain of topics relating to physical, life, earth, space, and technical systems of knowledge, all of which involve the modeling of objective reality (for a conceptual account, see the Rocard Report, 2007). In addition, it also refers to the ways of acquiring and applying knowledge according to the methods of natural science and technology, as well as the structured whole of knowledge and skills resulting from these methods (see Walma van der Molen, van Aalderen-Smeets, & Asma, 2010).
REVIEW OF CONCEPTS
The main goal of our review was to explicate the dimensions and attributes of the construct of primary teachers' attitudes toward science as used in previous studies. The review was conducted in line with the aforementioned definitions of attitude and attitude objects. We thus used the tripartite division of attitude as an initial frame of reference to categorize the components that have been measured under the banner of attitude in previous studies. However, it should be noted that we used this model only as a starting point. As described below, we added new elements or labels in cases where they provided a better representation of the measured dimensions. In doing so, we developed a new and more detailed structure for the construct of primary teachers' attitudes toward science. In our view, this structure adds to the relatively broad labeling of cognitive, affective, and behavioral components that are described in the original tripartite division. Unlike most previous studies, the current review also includes an explicit distinction between personal attitude and professional attitude and the observed attitude components are categorized accordingly.
A literature search was conducted to select all studies that have investigated the attitude of preservice and inservice primary teachers toward science (and the teaching of these subjects) from 1985 until 2010, including quantitative and qualitative studies. To our knowledge there is no theoretical reason to assume or expect that the components of the construct of attitude are different for preservice and inservice teachers and we therefore included studies investigating both groups. Furthermore, several large studies on the attitude of the general public toward science were included, because of the relevance of these studies for further explication of the construct of personal attitude toward science.
The following databases were searched: Education Resources Information Centre (ERIC), Web of Science, PsychInfo, and Scopus. In addition, the reference lists of selected studies were searched to ensure that all relevant studies had been included. Several keywords and terms were used, including attitude and attitude-related concepts (e.g., beliefs, confidence, and self-efficacy; science, technology, and STEM; teaching; primary and elementary; inservice and preservice; public). The search yielded 41 quantitative studies and 4 qualitative studies. For each study, we analyzed which dimensions or aspects of attitude were measured in the study. We categorized these aspects as either one of the dimensions of the tripartite model or included a new dimension when it was conceptually different from these dimensions of the tripartite model. In addition, we categorized the aspects of attitude as pertaining to personal or professional attitude toward science. When all aspects and dimensions of attitude were derived, we subcategorized the aspects of attitude within dimensions were possible, resulting in multiple subcategories. Table 1 provides an overview of the resulting dimensions and subcategories that we derived from the reviewed studies.
Table 1. Overview of Studies on Primary Teachers' Personal and Professional Attitudes Toward Science, Categorized per Investigated Attitude Component
Burgerpeiling wetenschap (2009), Cobern and Loving (2002), Coulson (1992), Eurobarometer 38.1 (1992), Evans and Durant (1995), Koballa (1986), McDevitt et al. (1993), Palmer (2004), and Young (1998). (Total 9)
Appleton and Kindt (1999), Carleton et al. (2008), Cobern and Loving (2002), Coulson (1992), Harty et al. (1991), Haury (1989), Johnston and Ahtee (2006), Koballa (1986), Liang and Gabel (2005), McDevitt et al. (1993), Pederson and McCurdy (1992), and Ramey-Gassert et al. (1996). (Total 12)
Hartshorne (2008), Harty et al. (1991), Haury (1989), Koballa (1986), McDevitt et al. (1993), Mulholland and Wallace (1996), Pederson and McCurdy (1992), and Ramey-Gassert et al. (1996). (Total 8)
Harlen (1997), Harlen and Holroyd (1997), Holroyd and Harlen (1996), Johnston and Ahtee (2006), and Liang and Gabel (2005). (Total 5)
Cobern and Loving (2002), McDevitt et al. (1993), Mulholland and Wallace (1996). (Total 3)
McDevitt et al. (1993) and Shepardson and Pizzini (1992). (Total 2)
Hartshorne (2008), Harty et al. (1991), Haury (1989), Koballa (1986), Liang and Gabel (2005), McDevitt et al. (1993), Mulholland and Wallace (1996), Palmer (2001, 2004), Pederson and McCurdy (1992), Ramey-Gassert et al. (1996), and Young (1998). (Total 12)
Hartshorne (2008), Harty et al. (1991), Haury (1989), Johnston and Ahtee (2006), Koballa (1986), Liang and Gabel (2005), Martin-Dunlop and Fraser (2007), McDevitt et al. (1993), Pederson and McCurdy (1992), and Ramey-Gassert et al. (1996). (Total 10)
Atwater et al. (1991), Bursal and Paznokas (2006), Evans and Durant (1995), Hartshorne (2008), Haury (1989), McDevitt et al. (1993), Palmer (2004), and Young (1998). (Total 8)
Carleton et al. (2008), Hartshorne (2008) Harty et al. (1991), Haury (1989), Koballa (1986), Liang and Gabel (2005), McDevitt et al. (1993), Palmer (2001), Pederson and McCurdy (1992), and Ramey-Gassert et al. (1996). (Total 10)
Behavior/ behavioral intention
Appleton and Kindt (1999), Goodrum et al. (2001), Haney et al. (1996), Palmer (2001), and Yates and Goodrum (1990). (Total 5)
Atwater et al. (1991), Coulson (1992), Mulholland and Wallace (1996), Palmer (2001, 2004, 2006), Skamp (1991), Yates and Goodrum (1990), and Young (1998). (Total 9)
Appleton (1995), Appleton and Kindt (1999), Bleicher (2007), Bursal and Paznokas (2006), Carleton et al. (2008), Coulson (1992) Harlen (1997) Harlen and Holroyd (1997), Hartshorne (2008), Harty et al. (1991), Holroyd and Harlen (1996), Jarret (1999), Jarvis and Pell (2004), Koballa (1986), McDevitt et al. (1993), Murphy et al. (2007), Palmer (2001, 2006), Pederson and McCurdy (1992), Ramey-Gassert et al. (1996), Scharmann and Orth Hampton (1995), Schoon and Boone (1998), Skamp (1991), Tosun (2000), Wenner (1993, 1995), Yates and Goodrum (1990), and Yilmaz-Tuzun (2007). (Total 28)
Asma et al. (in press), Carleton et al. (2008), Haney et al. (1996), Harty et al. (1991), Koballa (1986), Liang and Gabel (2005), Lumpe et al. (2000), McDevitt et al. (1993), Pederson and McCurdy (1992), and Ramey-Gassert et al. (1996). (Total 10)
Inspection of Table 1 shows that, in some cases, the same study appears in different cells. This is because various aspects of attitude may be measured within a single study. For this reason, although many studies do not make explicit distinctions between different attitude components or attributes, we identified separate dimensions and subcategories according to the actual measurement instrument items or reference literature presented in the articles. For those studies that neither reported measurement items nor included references to item lists (or those for which the references could not be obtained), we derived the measured dimensions and aspects either from the global descriptions of the instruments or from textual references in the results and discussion sections of the articles.
Personal Attitude Toward Science
From the reviewed literature, we derived multiple distinct beliefs and thoughts that could be categorized under the cognitive component of personal attitude. As listed in Table 1, we organized this broad array of beliefs into three subcategories that we will discuss briefly: the perceived relevance or importance of science for society and daily life, the perceived difficulty of science, and beliefs about gender differences in science.
The first attribute is perceived relevance or importance of science, referring to the extent to which people consider science relevant or important for their personal lives, for society, for prosperity, or for health. The following are examples of questionnaire items that are commonly used to measure perceived relevance: “Science is important for solving problems” and “Science is useful for daily life.” This attribute has been measured in several studies on the attitudes of preservice teachers (Cobern & Loving, 2002; Coulson, 1992; Koballa, 1986; McDevitt, Heikkinen, Alcorn, Ambrosio, & Gardner, 1993; Palmer, 2004; Young, 1998) as well as in studies of public attitudes toward science (Burgerpeiling Wetenschap, 2009; Eurobarometer 38.1, 1992; Evans & Durant, 1995). Although the latter studies on public attitude are important sources of information for governmental policy, they are often methodologically and conceptually weak (Pardo & Calvo, 2002). The objects of attitude measured in such studies include a wide array of attitudinal objects (predominantly controversial, science-related issues), and these attitude objects are treated as similar concepts for the purposes of analysis. In addition, these studies measure mostly one subcomponent of the complete construct of attitude toward science, namely the relevance of science.
Furthermore, the used instruments contain questions that were phrased in such way that they could have encouraged socially desirable answers. Many studies investigating respondents' beliefs about the relevance or value of science used items that are almost impossible to disagree with and that therefore do not discriminate between respondents with positive attitudes toward science from those with negative attitudes toward these topics. Nevertheless, we consider the relevance or importance that individuals assign to science with regard to society and daily life, as an important aspect of their attitudes toward science.
The second cognitive attribute that we derived—perceived difficulty—refers to the beliefs of individuals concerning the general difficulty of science relative to other fields of study. Several studies included items that covered this category, including the following: “Science is too difficult” and “Only smart people can understand science” (Hartshorne, 2008; Harty et al., 1991; Haury, 1989; Koballa, 1986; McDevitt et al., 1993; Mulholland & Wallace, 1996; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996).
The third attribute that we derived concerns beliefs about different gender roles and the different abilities of men and women in science (e.g., “I believe that men are better in understanding science than women are”). Although gender differences are a much discussed and investigated topic in the literature on science education, to our surprise we found only three studies investigating such gender stereotypical beliefs (Cobern & Loving, 2002; McDevitt et al., 1993; Mulholland & Wallace, 1996). All other studies focusing on gender differences are directed at outcome differences between men and women (or between boys and girls) with regard to their knowledge, skills, or attitudes in the area of science. For example, these studies focus on the difference in science performance indicated by test scores or the difference in self-efficacy beliefs. For the current review, however, we focused on respondents' beliefs about gender differences (i.e., what are teachers thoughts about a difference between men and women or boys and girls in science skills, knowledge, or attitudes), since these constitute an essential part of the attitudes of primary teachers toward science and may influence their teaching. Although these beliefs may be difficult to measure directly (respondents are known to be prone to give socially desirable answers), educational research suggests that gender-differentiated expectations are among the most influential educational factors affecting student motivation, attitudes, and academic achievement (Rosenthal, 1994). We therefore believe that gender-stereotypical beliefs should be included in any theoretical framework of teacher attitudes toward science (and the teaching of this subject) and that improved explicit and implicit measurements should be developed.
A considerable number of studies investigated primary teachers' emotions or feelings related to science in ways that are not directly related to their teaching practice. We categorized such feelings under the affective component of personal attitude toward science and subdivided this dimension into two categories, as shown in Table 1. The first subcategory contains positive feelings of pleasure and joy related to science in daily life or as part of leisure time (as studied by Hartshorne, 2008; Harty et al., 1991; Haury, 1989; Koballa, 1986; Liang & Gabel, 2005; McDevitt et al., 1993; Mulholland & Wallace, 1996; Palmer, 2001, 2004; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996; Young, 1998). These investigations include items concerning the extent to which respondents enjoy informing themselves about science in their daily lives (e.g., reading a science magazine for fun) or their pleasure in taking part in various science-related activities (e.g., conducting experiments or visiting a science museum).
The second attribute that we derived comprises negative feelings of anxiety related to science in general or in daily life. Several studies (Atwater et al., 1991; Bursal & Paznokas, 2006; Evans & Durant, 1995; Hartshorne, 2008; Haury, 1989; McDevitt et al., 1993; Palmer, 2004; Young, 1998) investigated respondents' fear of or aversion toward science-related concepts and activities that are not necessarily related to the context of primary school context. For example, several studies measured how nervous or tense people feel when they encounter science-related topics or when they have to discuss science topics with someone. As shown in Table 1, studies investigating such general negative emotions do not necessarily include the simultaneous investigation of the positive affect of respondents toward science.
The third component of the tripartite division of attitude is behavior or behavioral intention. None of the articles that we reviewed reported having measured behavior related to science in respondents' daily lives or measured behavioral intention to engage in activities related to science or technology. As we discuss more elaborately in our evaluation of the proposed attitude dimensions, the behavioral (intentional) component of attitude is difficult to measure, as it is not easily separated from affect (e.g., the item “I would like to read about science more often” could relate either to someone's affect toward reading more about science or to someone's intention to do so). Examples of activities that might fit within the category of teachers' personal behavior include reading the science section in the newspaper, watching scientific television programs, or wondering about natural phenomena.
In addition to the above-described dimensions, which can be directly categorized as one of the dimensions of the tripartite division of attitude, we derived additional components that are highly relevant to the construct of primary teachers' attitudes toward science but that, in our view, do not necessarily fit into one of the overall cognitive, affective, or behavioral dimension. One of these additional components is self-efficacy, which refers to the perceived level of capability or confidence that an individual has for performing a particular behavior that may contain difficult and stressful elements (Bandura, 1997). In the context of personal attitude, one example could involve individuals' beliefs about their capacity to understand science-related articles in the newspaper. The definition of self-efficacy resembles that of self-confidence, and the two concepts have been used interchangeably in some cases (Palmer, 2006). In theoretical terms, we suggest that self-efficacy is qualitatively different from the purely cognitive and affective components. In our view, self-efficacy results from a combination of feelings and beliefs about internal factors (abilities, knowledge, and experience) that influence the feeling of being in control with regard to the execution of particular behaviors.
Several studies (Atwater et al., 1991; Coulson, 1992; Mulholland & Wallace, 1996; Palmer, 2001, 2004, 2006; Skamp, 1991; Yates & Goodrum, 1990; Young, 1998) report having measured aspects that can be categorized as self-efficacy or self-confidence related to personal attitude toward science. Mulholland and Wallace (1996) report confidence measures in science knowledge and skills related to the daily lives of individuals. Atwater and colleagues (1991) include at least one item on self-confidence and self-efficacy: “I feel a sense of insecurity when attempting physical science” (p. 8). However, some studies that investigated aspects of attitude that we categorize here as self-efficacy used items that were ambiguous and unclear. For example, Coulson (1992) asked preservice teachers about their beliefs in their ability to succeed in science. It is unclear what is meant by succeed in science. It could refer to succeeding in science class, succeeding in science experiments, or succeeding in science-related professions.
Professional Attitude Toward Science
As presented in Table 1, the distinct beliefs categorized under professional attitude that we derived from the literature were similar to the beliefs that we derived for personal attitude. We distinguished three cognitive attributes for primary teachers' professional attitude: teachers' perceptions of the relevance or importance of teaching science in primary school, perceptions about the relative difficulty of teaching science, and their beliefs about gender differences related to teaching science. As listed in Table 1, perceived relevance was investigated most frequently of these three attributes (Appleton & Kindt, 1999; Carleton, Fitch, & Krockover, 2008; Cobern & Loving, 2002; Coulson, 1992; Harty et al., 1991; Haury, 1989; Johnston & Ahtee, 2006; Koballa, 1986; Liang & Gabel, 2005; McDevitt et al., 1993; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996). We did not observe the methodological issues that were present in the items and scales that measured beliefs about the relevance of science regarding personal attitude. The items concerning teachers' beliefs about the relevance of teaching science in primary school (e.g., “It is important to teach science at the primary school level”) seemed to discriminate well and to be designed properly.
The second attribute that we derived for the cognitive component of professional attitude concerns the level of difficulty that teachers attribute to teaching science in primary school. As listed in Table 1, this attribute was investigated less frequently than were teachers' perceptions regarding the relevance of teaching science (Harlen, 1997; Harlen & Holroyd, 1997; Holroyd & Harlen, 1996; Johnston & Ahtee, 2006; Liang & Gabel, 2005). It should be noted that the cognitive beliefs about the difficulty of teaching science are, in our view, qualitatively different from teachers' perceived self-efficacy in teaching science. The following is an example of the type of items used to measure the cognitive attribute: “Most teachers believe that teaching science is more difficult than teaching other subjects.” This statement represents teachers' general beliefs about the difficulty of teaching these topics. In contrast, professional self-efficacy is defined as the perception that teachers have about their own capability in teaching science. This point is discussed in more detail below.
The third attribute that we derived involves gender-stereotypical beliefs about teaching science (McDevitt et al., 1993; Shepardson & Pizzini, 1992). This may refer to two conceptually distinct beliefs. The first type of belief involves possible differences between male and female teachers with regard to their ability to teach science and their level of interest and enjoyment in doing so (e.g., “I believe male teachers enjoy teaching science in primary school more than female teachers do”). The second type of belief involves perceived differences between boys and girls in science (e.g., “Boys have more fun than girls in doing technical assignments” or “it is easier to teach science to boys than to girls”). Although the studies that we reviewed do not specify the gender belief to which they referred, the items and discussion sections suggest that the primary focus was on teachers' perceptions of gender differences among pupils.
The affective component of the tripartite division for the professional attitude of teachers encompasses feelings and emotions related to teaching science in primary school. Both positive and negative feelings are addressed in the studies we reviewed (see Table 1). As with the construct of personal attitude, these feelings can be categorized as either enjoyment in teaching science (Hartshorne, 2008; Harty et al., 1991; Haury, 1989; Johnston & Ahtee, 2006; Koballa, 1986; Liang & Gabel, 2005; Martin-Dunlop & Fraser, 2007; McDevitt et al., 1993; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996) or anxiety about teaching science (Carleton et al., 2008; Hartshorne, 2008; Harty et al., 1991; Haury, 1989; Koballa, 1986; Liang & Gabel, 2005; McDevitt et al., 1993; Palmer, 2001; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996).
Anxiety about teaching science has been measured reliably. The following are typical examples of items used in these studies: “It scares me to have to teach science” and “I become tense when I have to teach science.” In contrast, the validity of the enjoyment measures can be questioned. In our view, a study investigating the attitudes toward teaching science should measure the enjoyment that teachers or preservice teachers experience from teaching science. Many of the studies in our review of concepts, however, investigated the attitudes of preservice teachers who had no teaching experience but who had taken science courses as part of their vocational training. Even though these studies claim to investigate attitudes toward teaching science, the items used often question the enjoyment of science classes, rather than enjoyment derived from teaching science (Martin-Dunlop & Fraser, 2007; Palmer, 2001).
A second issue concerns the ambiguous phrasing of some of the items that do address enjoyment in teaching. The following are examples: “I am looking forward to teaching science in the elementary school” (Haury, 1989, p. 508) and “I eagerly anticipate the teaching of science… ” (Hartshorne, 2008, p. 345). One can argue whether “looking forward” and “anticipate” denotes enjoyment, curiosity, or interest in teaching science. One can even question whether preservice teachers providing positive responses to these items necessarily have a positive attitude toward teaching science or whether they are simply eager to start working as teachers after graduating. Despite these methodological issues, we believe that enjoyment is an important factor to consider when establishing primary teachers' attitudes toward science when measured more validly.
The behavioral component of attitude toward teaching science was investigated in only a few studies (Appleton & Kindt, 1999; Goodrum et al., 2001; Haney et al., 1996; Palmer, 2001; Yates & Goodrum, 1990). Haney and colleagues (1996) measured the influence of attitude, subjective norms, and perceived behavioral control on intention to teach science. They showed that attitude is the best predictor of behavioral intention. In a qualitative study, Appleton and Kindt (1999) interviewed starting teachers concerning their attitudes toward teaching science. They report several statements from teachers that can be categorized as the intention to teach science. Examples include statements that science lessons are more likely to be postponed or canceled than are language and mathematics lessons or that teachers are not likely to teach these topics, as they require so much preparation time and effort. In addition to measuring the intention to teach, Palmer (2001) measured actual behavior by asking preservice teachers whether they had volunteered to teach science during their summer holidays.
As shown in Table 1, the most frequently investigated component of the construct of primary teachers' professional attitude is self-efficacy. Studies investigating self-efficacy focused on a variety of aspects and effects of self-efficacy, including the following: effect studies of interventions aimed at preservice and inservice teachers (Appleton, 1995; Bleicher, 2007; Carleton et al., 2008; Hartshorne, 2008; Harty et al., 1991; Jarrett, 1999; Jarvis & Pell, 2004; McDevitt et al., 1993; Palmer, 2001, 2006; Pedersen & McCurdy, 1992; Scharmann & Orth Hampton, 1995; Wenner, 1993, 1995), studies on the development of instruments (Coulson, 1992; Yilmaz-Tuzun, 2008), studies investigating the relationship between self-efficacy and science-related knowledge of preservice teachers (Appleton, 1995; Harlen, 1997; Harlen & Holroyd, 1997; Holroyd & Harlen, 1996; Murphy, Neil, & Beggs, 2007; Schoon & Boone, 1998; Tosun, 2000), one study investigating the relationship between anxiety and self-efficacy (Bursal & Paznokas, 2006), studies investigating the difference between preservice teachers and experienced inservice teachers with regard to self-efficacy (Skamp, 1991), studies focusing on the sources of self-efficacy in preservice teachers (Appleton & Kindt, 1999; Palmer, 2001; Ramey-Gassert et al., 1996), one study on the relation between self-efficacy and intention to teach science (Koballa, 1986), and one study focusing on the amount of self-efficacy of primary teachers (Yates & Goodrum, 1990).
The majority of these studies used the Science Teaching Efficacy Belief Instrument (STEBI) developed by Enochs and Riggs (1990) or one of the modified versions (i.e., the STEBI-B for preservice teachers or the MTEBI for mathematics beliefs) (Carleton et al., 2008; Palmer, 2006; Ramey-Gassert et al., 1996; Scharmann & Orth Hampton, 1995; Schoon & Boone, 1998; Wenner, 1993, 1995). Over the years, this instrument has received considerable support as a measure for self-efficacy (Bleicher, 2004; Palmer, 2006; Scharmann & Orth Hampton, 1995; Wenner, 1993, 1995). However, the instrument contains a design flaw, which has not yet been described elsewhere. The STEBI uses a 5-point Likert scale ranging from strongly agree to strongly disagree, with the assumption that the five response categories form a continuous scale indicating the extent of agreement with the item. However, the corresponding response labels (strongly agree, agree, uncertain, disagree,and strongly disagree) do not indicate a continuous scale. The middle response label—uncertain—is conceptually different from a middle response option between agree and disagree. This response could also indicate that the respondent is not sure about what to answer, does not understand the statement, or has no opinion on the topic. The results of the studies using the STEBI should therefore be interpreted with caution. It is well known that the midpoint in a 5-point (or 7-point) Likert scale is a conceptually difficult response category as it indicates that the respondent neither agrees nor disagrees with the statement. The meaning of such a response is difficult to determine. Nevertheless, the midpoint is often included as a forced-choice response scale (four or six response options), in which no middle point is present, could easily lead to a skewed image of the respondents' attitude.
The remaining studies, which did not use the STEBI, had devised their own surveys and questionnaires on self-efficacy (Appleton, 1995; Jarvis & Pell, 2004; Skamp, 1991) or questionnaires on self-confidence (Harlen & Holroyd, 1997; Holroyd & Harlen, 1996; Murphy et al., 2007). Many of the instruments used in these studies were neither validated nor reported reliability measures (Appleton, 1995; Hoslroyd & Harlen, 1996; Skamp, 1991).
In addition to self-efficacy, we derived a second component of professional attitude from the literature that was not present in the original tripartite model. This component can be described as the perceived influence of context factors for teaching science. The perceived influence of context factors refers to teachers' beliefs that certain external factors make it easier or harder to teach science. These beliefs range from the idea that certain factors bear a slight influence on teaching to the conviction that certain external factors are essential for teaching. In the latter case, teachers feel completely dependent on these factors.
Perceived influence of context factors is measured by several studies (Asma et al., in press; Carleton et al., 2008; Haney et al., 1996; Harty et al., 1991; Koballa, 1986; Liang & Gabel, 2001; Lumpe, Haney, & Czerniak, 2000; McDevitt et al., 1993; Pederson & McCurdy, 1992; Ramey-Gassert et al., 1996).
Appleton and Kindt (1999) asked starting teachers about context factors that would make it harder or easier for them to teach science. The study identifies four main factors: collegial support, lack of resources (materials and examples), time allocated for science in the curriculum, and the time and effort needed to prepare science lessons. Lumpe and colleagues (2000) developed a specialized instrument (CBATS: Context Beliefs About Teaching Science) for measuring perceived context factors. This instrument questions respondents about the perceived influence of 26 potential context factors on teaching science. These 26 items can be roughly categorized into the following five factors: standardized teaching methods and curriculum, social support, resources (materials and money), time available within the curriculum, and preparation time. These factors overlap partially with the factors found by Appleton and Kindt (1999). Carleton and colleagues (2008) also emphasize the importance of context factors as measured according to the CBATS instrument.
In summary, our review of concepts investigating the personal and professional attitudes of primary teachers toward science showed that a large portion of research in this field focuses on professional attitude and on teachers' perceptions of self-efficacy in teaching science. Overall, the components and aspects of attitude can be organized into a structure consisting of dimensions that are related to the tripartite model of attitude and to dimensions that were not originally embedded in that model, but which are conceptually related to teachers' attitudes toward science (i.e., self-efficacy and context beliefs).
Although the above review provides important and interesting information on the dimensions and aspects of attitude that have been investigated in the field of science education, several remarks must be made about methodological issues we encountered during the review. It is important to mention these issues for two reasons. First, they indicate the limits for the reliability of the current review of concepts. Second, they underscore the statement made in the Introduction that, because the construct of attitude is poorly defined, studies of this concept are difficult to interpret, compare, and replicate. We encountered the following issues:
First, as stated by various previous articles (see the Introduction), the construct of teachers' attitude toward science has thus far been defined very poorly. The studies addressed in this review underscore this observation. The majority of the studies provided no appropriate definition for the concept of attitude (Appleton, 1995, Appleton & Kindt, 1999; Atwater et al., 1991; Carleton et al., 2008; Cobern & Loving, 2002; Coulson, 1992; Eurobarometer 38.1, 1992; Evans & Durant, 1995; Haney et al. 1996; Harty et al., 1991; Hartshorne, 2008; Haury, 1989; Jarret, 1999; Jarvis & Pell, 2004; Johnston & Ahtee, 2006; Koballa, 1986; Liang & Gabel, 2005; Martin-Dunlop & Fraser, 2007; McDevitt et al., 1993; Mulholland & Wallace, 1996; Palmer, 2001, 2004; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996; Shepardson & Pizzini, 1992; Yates & Goodrum, 1990; Yilmaz-Tuzun, 2007; Young & Kellogg, 1993; Young, 1998). Some of these studies gave a coarse definition of attitude without specifying the various dimensions of the construct. None of the studies in this review defined attitude according to its constituent parts. The studies that did define the concept they were investigating were mostly focusing on self-efficacy or self-confidence and only defined these latter concepts (Bleicher, 2007; Bursal & Paznokas, 2006; Harlen & Holroyd, 1997; 1995 Lumpe et al., 2000; Palmer, 2006; Scharmann & Orth Hampton, 1995; 1998 Schoon & Boone, 1998; Skamp, 1991; Wenner, 1993, 1995).
Owing to the lack of consensus regarding the definition of the construct of attitude, we observed considerable variation in the concepts and attitude objects measured in different studies. In this review, we report only components of personal and professional attitude toward science, although a wide range of additional attitude objects was measured in many studies, such as opinions about science teaching methods, opinions about the required levels of knowledge for teachers, or views about teamwork in class (Atwater et al., 1991; Burgerpeiling Wetenschap, 2009; Carleton et al., 2008; Coulson, 1992; Eurobarometer 38.1, 1992; Evans & Durant, 1995; Haney et al. 1996; Haury, 1989; Jarvis & Pell, 2004; Johnston & Ahtee, 2006; Liang& Gabel, 2005; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996; Young, 1998).
The measurement of different aspects of attitude and attitude objects need not be problematic, as long as the scores on the instruments are analyzed according to psychometric standards. However, many studies in this review reported only one overall score that reflected the combination of many different underlying concepts. This renders the results impossible to interpret (e.g., Hartshorne, 2008; Harty et al., 1991; Haury, 1989). For example, many studies used a version of SAS (Harty et al., 1991; Koballa, 1986; Liang & Gabel, 2005; McDevitt et al., 1993; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996). This instrument (and the different versions thereof) measures aspects of personal attitude (feelings toward handling science equipment and toward anything that is associated with science), aspects of professional attitude (feelings and beliefs toward teaching science), feelings and beliefs toward college science courses, and the use of scientific ideas within daily life intermixed (Hassan & Shrigley, 1984; Shrigley, 1974; Shrigley & Johnson, 1974; Thompson & Shrigley, 1986). These different attitude objects are measured within one scale and result in one overall score.
We also examined the object of attitude that was measured (i.e., whether studies measured personal attitude, professional attitude, or both). Many of the studies reviewed here used instruments that focus solely on professional attitude (Appleton, 1995, Appleton & Kindt, 1999; Bleicher, 2007; Carleton et al., 2008; Haney et al. 1996; Harlen & Holroyd, 1997; Jarvis & Pell, 2004; Johnston & Ahtee, 2006; Koballa, 1986; Palmer, 2006; Scharmann & Orth Hampton, 1995; Shepardson & Pizzini, 1992; Schoon & Boone, 1998; Skamp; 1991; Wenner, 1993, 1995; Yilmaz-Tuzun, 2008), in contrast to the small number of studies focusing exclusively on personal attitude (Burgerpeiling Wetenschap, 2009; Eurobarometer 38.1, 1992; Evans & Durant, 1995; Palmer, 2004). A number of studies measured aspects of both attitudes. Only a few of these studies explicitly distinguished between these two attitudes and provided separate scores for personal and professional attitude components (Asma et al., in press; Atwater et al., 1991; Harty et al., 1991; Koballa, 1986). The other studies that we categorized as measuring both personal and professional attitude did not explicitly refer to this distinction in the text and do not include any comment or reference to a conceptual distinction within the items used. They used instruments that include items for both personal and professional attitude (as the aforementioned SAS) but did not make a theoretical distinction between the two attitudes, nor did they analyze these separately (Cobern e& Loving, 2002; Coulson, 1992; Hartshorne, 2008; Haury, 1989; Jarret, 1999; Liang & Gabel, 2005; McDevitt et al., 1993; Murphy et al., 2007; Pedersen & McCurdy, 1992; Ramey-Gassert et al., 1996; Yates & Goodrum, 1990).
In addition to the issues pertaining to the poor definition of attitude and attitude objects and those pertaining to poor instrument design and analysis, the studies we reviewed were characterized by considerable ambiguity regarding the phrasing of items. Some of the instruments contained items for which it is unclear which component of attitude is being measured or what a positive or negative response implies. This was, for example, the case for a number of items on the various versions of SAS (Hassan & Shrigley, 1984; Shrigley, 1974; Shrigley & Johnson, 1974; Thompson & Shrigley, 1986). This instrument includes the following items: “I daydream during science classes,” “I would not enjoy working in a science laboratory for a summer,” and “I have a preference for science teaching over other subjects.” A positive response to the item “I daydream… ” does not necessarily indicate that the respondent has a negative attitude toward science. It is possible that a respondent daydreams during all classes while actually liking science. A negative response to the same item is even harder to interpret. Never daydreaming during science class does not imply that an individual likes science class or is positively disposed toward science. The same logic applies to the other two items stated above. Positive responses to these items (or negative in the case of a negatively phrased item) do not necessarily indicate a positive attitude toward science. Responses indicating that one does not want to work in a science lab during the summer could very well mean that this respondent does not want to work at all during the summer, whether this is in a science lab or art fair. The ambiguity of positive and negative responses to these items could be due to the fact that they are phrased as statements and behaviors rather than as beliefs. Responses to these items are therefore affected by many underlying factors, only one of which is attitude. It is therefore impossible to deduce underlying attitudes from these items.
We were able to observe the aforementioned methodological issues only in articles presenting the items or the instruments they used. For those articles that did not present the items, we were in most cases able to obtain information on the instrument by following the literature references. For some studies, however, we were unable to locate the referenced literature, due to either the nature of the publication (Ph.D. thesis) or the publication date of the reported instruments (Appleton, 1995; Atwater et al., 1991; Carleton et al., 2008). In cases for which no information on the instrument could be obtained, we were forced to rely on titles or descriptions of the scales, sample items, or remarks included in the results or discussion section of the article. This suggests that the actual items in the questionnaires may have measured concepts or aspects of attitude other than those described in this review article. For one study, we could not obtain any information about the scale (Harty et al., 1991) as the only description of the employed instrument was that it measured “attitude toward science.”
THEORETICAL EVALUATION OF THE DIMENSIONS OF ATTITUDE: TOWARD A NEW FRAMEWORK
The objective of this review of concepts was to derive all relevant aspects and dimensions of the construct of primary teachers' attitude toward science as used in existing literature and to provide a structure for these dimensions and components. Despite the difficulties that we encountered in specifying the attributes of attitude that were measured in certain studies (as described above), the results of our review enhance our understanding of the underlying dimensional structure of primary teachers' attitudes toward science. This understanding is reflected in our formulation of a new theoretical framework. Before presenting this framework for attitude toward science, we evaluate the potential components and attributes in light of general theories on the relation between attitude and behavior.
Reconsidering the Tripartite Model
We used the tripartite model of attitude as the initial frame of reference for our review of concepts. The results of the review, however, indicate that very few studies incorporate components that could be categorized as behavior or behavioral intentions. In contrast, cognition and affect were found in various relevant attributes. These differences in results could be due to difficulties involved in distinguishing between intention and attitudinal beliefs or affect (e.g., research shows that such items as “I like or I look forward to teaching science” load predominantly on the affective component) and to the subsequent difficulty of establishing a behavioral component of attitude. Alternatively, this finding could reflect a more profound theoretical difference between attitudinal beliefs, emotions, and behavior.
The components of cognition and affect are well established within the tripartite division of attitude (Eagly & Chaiken, 1993), and they are supported by one of the most influential theories on the relation between general attitude and behavior: the theory of planned behavior (TPB) (Ajzen & Fishbein, 1980). According to this theory, attitude consists of a cognitive component and an affective component. The cognitive and affective components of attitudes partly determine behavioral intention, which is the immediate motivational factor for behavior itself. Behavioral intention is thus viewed as a direct outcome of these two dimensions of attitude, and not as a component of attitude itself. The TPB further argues that behavioral intention indicates the extent to which a person is willing to try to perform a certain behavior. Whether the action actually takes place is determined by the combination of behavioral intention and specific circumstances at a given time and place. On the basis of these general theories of attitude and behavior and on the results of our review of the concept of primary teachers' attitude toward science, we believe that the behavior and behavioral intentions of primary teachers should be seen as conceptually different from their attitudes toward science (and the teaching of these subjects) and should not be part of the construct of attitude itself.
Reconsidering Self-Efficacy and Context Beliefs
As shown in the present review, a large number of studies investigating the attitudes of primary teachers toward teaching science focused on the concept of self-efficacy. The concept of self-efficacy was first postulated by Bandura's social cognitive theory (Bandura, 1997). It refers to the perceived capability of an individual to perform a particular behavior that may contain difficult and stressful elements. In most of these studies, self-efficacy was investigated in isolation and was not explicitly related to other dimensions of attitude. In some cases, the concept was even equated with attitude. Although many primary teachers do indeed seem to feel insecure about teaching science, we strongly believe that self-efficacy is but one of several underlying attributes of teachers' attitudes toward science. We further consider it important to investigate the relationship of self-efficacy to the beliefs, feelings, and behavior of primary teachers.
In addition to the concept of self-efficacy, Bandura's (1997) social cognitive theory includes the concept of response–outcome expectancy. Response–outcome expectancy refers to a person's beliefs or opinions about whether a given behavior will produce a desired outcome. In practice, however, response–outcome expectancy is predominantly concerned with opinions about the effectiveness of particular behavior (in our case, the question of what constitutes effective primary science teaching) rather than with personal evaluative beliefs about the construct. In contrast to outcome expectancy, self-efficacy is concerned with evaluative beliefs that individuals hold regarding their own ability to perform certain actions. We therefore assume that self-efficacy is a part of teachers' attitude toward science but that the concept of outcome expectancy does not fit into our framework.
Psychological theories on the relationship between attitudes and behavior (e.g., the TPB or Ford's motivation systems theory; see Ford, 1992) also recognize self-efficacy as an important attitudinal factor. In the TPB, self-efficacy is part of the concept of perceived behavioral control, one of several factors that are hypothesized to influence behavioral intention. Ford's motivation systems theory states that an individual's performance is the result of several factors, including capability beliefs (Ford, 1992). Capability beliefs are beliefs that individuals hold about their own ability to perform particular behaviors and these beliefs can be regarded as synonymous to self-efficacy (Ford, 1992; Lumpe et al., 2000).
Self-efficacy refers to what individuals believe or feel about their own abilities. In this article, we will use the term self-efficacy as relating to perceptions about abilities determined by internal factors. In our view, such perceptions are distinct from perceptions that teachers may have about external, contextual factors that could hinder or foster their science teaching (e.g., their beliefs or feelings about the necessity of elaborate teaching methods or financial school support for science). As our review has shown, such perceived influence of context factors has been recognized as a salient factor related to teachers' attitudes toward science in a number of studies (Appleton & Kindt, 1999; Carleton et al., 2008; Lumpe et al., 2000). The importance of context factors is also endorsed by the motivation systems theory (Ford, 1992), which states that, in addition to capability beliefs, context beliefs are important factors that may determine behavior.
However, we argue that it is not so much the perceived influence as well as the perceived dependency on context factors that is of interest when investigating primary teachers' attitudes toward teaching science. Attitude is a strong predictor of behavioral intention and intention is the antecedent of the actual behavior (Ajzen, 2001). In our view, the amount of dependency on context factors that teachers perceive is a stronger predictor of the intention to teach than their general beliefs about the influence of these factors on teaching. The perceived influence of context factors is more descriptive, but says nothing about whether teachers believe and feel that these context factors play a role in whether or not they will teach science. Recent research (e.g., Asma et al., in press) has shown that a positive attitude change in primary teachers, who received additional professional training in science, was coupled with a diminishing dependency on external factors and a concomitant greater sense of being in control, regardless of the availability of teaching methods, materials, or school support.
Theoretical Framework for Primary Teachers' Attitudes Toward Science
As mentioned in the Introduction to this article, our main goal was to develop a new theoretical framework for describing and researching primary teachers' attitudes toward science and the teaching of science. This framework is presented in Figure 1. It is the result of (a) our review of attitude concepts used in previous studies and (b) our theoretical evaluation of the observed concepts, in light of general psychological theories. Our proposed framework integrates the insights that we developed during our evaluation process and consists of three components and seven underlying attributes (see Figure 1). In our view, the framework is suitable for describing both the professional and personal attitudes of primary teachers toward science. The framework retains the cognitive and affective components of the original tripartite model. These components consist of subattributes that we propose, based on the studies that we reviewed and that seem to be highly relevant to the concept of attitude in this domain. We nonetheless revised the original tripartite model in two ways. First, based on the results of our review and in line with the TPB, we excluded the behavioral component as part of the underlying construct of attitude and added a new third component, which we labeled perceived control. As described before, we agree with Ajzen and Fishbein's (1980) assertion that behavior and behavioral intention are conceptually different from attitudes and that this component should not be part of the construct of attitude itself. Instead, attitudes should be viewed as antecedents of behavioral intention, which determines actual behavior as stated by the TPB. Therefore, in our new framework, we include behavior and behavioral intention as outcomes of attitude and not as part of the theoretical construct of attitude itself.
Instead of behavior and behavioral intention as a third component of attitude, we propose a new third component called perceived control, which consists of the attributes of self-efficacy and perceived dependency on context factors. Our review of the literature revealed that many researchers have studied self-efficacy as a primary or, in many cases, the sole attitudinal predictor of the teaching of science at the primary school level. To form a complete picture of the possible roles played by the various aspects of self-efficacy, we propose that self-efficacy should not be investigated in isolation, but in conjunction with other components and attributes. It should therefore be included in a framework for attitude toward science. We suggest that self-efficacy is neither a part of the affective component nor a part of the cognitive component of attitude, because it contains aspects of both and at the same time is qualitatively different from both. In our view, self-efficacy consists of cognitive and affective aspects that are focused on people's internal beliefs about and feelings of being in control to execute particular behaviors and it is therefore included as part of the component of perceived control.
In addition to self-efficacy, we believe that teachers' perceptions of external factors should be included in a component of perceived control. The outcomes of the present review of concepts suggest that, in addition to internal beliefs and feelings associated with self-efficacy, the beliefs and feelings that teachers have about external (i.e., contextual) factors are apparently closely related to teachers' sense of being in control. In our view, the perception of teachers regarding their dependency on context factors (e.g., their belief that they can teach science only if their school ensures the availability of the proper materials and sufficient preparation time) is an indispensable component of a complete theoretical framework of primary teachers' attitudes toward science. It should be noted that the component of perceived control deliberately refers to teachers' perceptions of being in control. Similar to the cognitive and affective dimensions of attitude, which are necessarily subjective, this new component reflects the subjective beliefs and feelings of individuals about internal and external obstacles, and not the factual presence of such obstacles.
Ajzen (2002) proposed that perceived external, contextual obstacles should be viewed as a part of self-efficacy. However, the results of an in-depth focus group study (Asma et al., in press) suggest that teachers' dependency on context factors may differ from their self-efficacy beliefs. In general, the teachers in Asma's study felt that they had sufficient self-efficacy. Nonetheless, a large group still perceived low control in teaching science; they felt that they could not teach science unless they were provided with the proper materials or methods. Measuring both aspects as a single attitudinal attribute would make it impossible to deduce these clearly different aspects, and it would hinder a distinction between different combinations of the attributes. In our view, teachers could theoretically be (a) low on self-efficacy and highly dependent on context factors, (b) high on self-efficacy, but still highly dependent on context factors, or (c) high on self-efficacy and not dependent on context factors. These three categories reflect increasing levels of confidence or sense of being in control. A fourth possibility, in which teachers are low on self-efficacy and simultaneously not dependent on context factors, seems less realistic. The two constructs are clearly related, although they need not point in the same direction, in contrast to Ajzen's (2002) argumentation. In our framework, we therefore included the two constructs as separate subattributes of the dimension that we labeled perceived control.
DISCUSSION AND IMPLICATIONS
The primary goal of the present article was to develop a new theoretical framework for describing and investigating primary teachers' attitude toward science and the teaching thereof. This framework is based on a review of the concepts of attitude as they are used in existing literature. The proposed components for the framework were evaluated on the basis of general theories on attitude and behavior. It should be noted that this framework does not constitute a definition or model of attitude in general. Instead, it provides an overview of all relevant concepts that we believe should be considered when investigating primary teachers' attitude toward science.
The components of the framework correspond to parts of various general theories on the antecedents of behavior. The TPB incorporates components equivalent to cognition and affect (albeit not subdivided into different attributes. However, although the TPB provides a valuable tool for predicting behavior and has similarities to the framework presented here, it is neither specifically designed nor intended as a structure for the construct of attitude and it therefore lacks specificity. The motivation system theory emphasizes the role of perceived control (i.e., self-efficacy and beliefs about context factors), but it does not refer to the cognitive and affective components of attitude. The particular division of the construct of attitude into cognitive, affective, and perceived-control components that we propose is new and has, to our knowledge, not been proposed elsewhere. However, it is important to realize that the subdivision in the three components and their seven underlying attributes is framed within the context of primary teachers' attitudes toward science and the teaching of these subjects. When this framework is to be used or applied within different contexts, care should be taken to investigate the suitability of the framework for this context.
Our proposed framework contributes to the theoretical definition and conceptualization of the construct of primary teachers' attitude toward science and the teaching of science, which has been poorly developed to date. It also provides a valuable schema for the development of a new and integrated instrument for measuring teachers' attitudes in this domain. Although previous survey instruments have measured certain aspects of teachers' attitudes, self-efficacy or dependency on context factors, none of the previously used attitude instruments are capable of validly measuring all the components of the construct of primary teachers' attitudes toward science, including our newly proposed component of perceived control. However, we developed and validated a new attitude measuring instrument, called the Dimensions of Attitude towards Science instrument (van Aalderen-Smeets, Walma van der Molen, & Asma, submitted). This instrument is solely based on the theoretical framework proposed in this article. The instrument is validated and proved reliable for measuring all components and subcomponents of preservice and inservice primary teachers' attitudes toward teaching science. It allows teachers to display their complete collection of beliefs, thought, and emotions toward science instead of asking teachers to focus exclusively on a subpart of their attitude toward science as has been done before (such as self-efficacy or their beliefs about the importance of science). In addition to contributing to the development of measuring instruments, the new framework can assist in designing interventions and training aiming to improve the attitudes toward science. Furthermore, it allows such interventions to be tailored toward teachers' individual patterns of scores on the various attitude dimensions and components.
As a final remark, we would like to stipulate that the professionalization of primary teachers in the domain of science cannot be accomplished without improving teachers' knowledge of science and scientific processes. However, in the context of primary teachers and science it should be noted that increases in knowledge do not always lead to improvements in the amount and quality of teaching these topics. Only when teachers believe that science are relevant and important, when they have positive feelings toward these subjects, and when they perceive themselves to be capable of teaching them without being dependent on too many context factors will teachers change and improve in their teaching of these topics.
The framework presented in this article is an essential new step toward a convergence of the research in this field. Only when researchers are aware of the complexity of the construct of teachers' attitudes toward science, when explicit and substantiated decisions have been made regarding which components and objects should be measured, and when methodologically sound instruments and interventions are used can scientific progress be achieved in research on teachers' attitudes. Future research is needed to investigate the various aspects of the proposed framework, including the relationships between the components and the weights of the various components and subattributes in predicting behavioral intention. The investigation of these aspects is a prerequisite for gaining further insight into the dynamics of primary teachers' attitudes toward science and for the development of interventions that are better suited to improve specific aspects of teachers' attitudes.