Students’ perceptions of their first experiences of secondary-school science in New Zealand

Moos (1974) began to focus on people’s perceptions of their learning environment. Moos centred his projects on human environments on three dimensions: personal development, relationships and system maintenance and change. His studies included hospitals, army camps and schools. He posed a number of questions. How well do the people get on with one another? How orderly or innovative is the environment? How is the system maintained? The Classroom Environment Scale (CES) (Moos & Trickett, 1987), is made up of nine different scales that measure learning in the school classroom as a whole. It was designed for teacher-centred classrooms and contains scales such as Teacher Support, Innovation, Teacher Control and Task Orientation. Since this time, learning environment measures have gained popularity, with a variety of valid and widely-applicable instruments being developed for assessing students’ perceptions of classroom climates (Fisher & Fraser, 1991; Fraser, 2018; Skordi & Fraser, 2019; Taylor et al., 1995).

The following outlines some class climate instruments still prominent in the field today. The College and University Classroom Environment Inventory (CUCEI) developed by Fraser and Treagust (1986) was a response to smaller classes of up to 30 students in higher education environments. It has seven scales with seven items for each scale, and each item has four responses (Strongly agree, Agree, Disagree, Strongly Disagree). The Questionnaire on Teacher Interaction (QTI), specifically developed for the nature and quality of interpersonal relationships between teachers and students (Wubbles & Brekelmans, 1998), is well suited to measuring student perceptions of teacher co-operation and influence. Studies such as Khine and Lourdusamy (2005) have strongly supported the validity and potential use of the QTI in different cultural contexts and in teaching subjects such as English as a foreign language (Wei et al., 2009). Both the Learning Environment Inventory (Fraser, 1982) and the Science Laboratory Environment Inventory (Fraser et al., 1993) were developed as a response to the need to capture learner perceptions of group processes and co-operative interactions. The widely-used SLEI has been validated as a class climate questionnaire in many countries and so too has the What Is Happening In this Class? (WIHIC) instrument (Fraser, 1998, 2018), for investigating student cohesiveness and degree of teacher support in the learning environment. An analysis of five decades of class climate research (Alansari & Rubie-Davies, 2019) has established prevailing research gaps represented across both tertiary and secondary classroom environments where further exploration of the use of student voice could support shifts in teacher pedagogy. Overall, there has been significant progress in the development of quantitative instruments, but further future investigation of students’ perceptions of their current learning environments is anticipated.

Some research on learning environments has principally focused on the relationship between learners’ perceptions of their learning environments and outcomes (Fraser, 2012). One challenge that appeared in studies of the 1980s was that of groups of students who were found to be more directly involved in classroom discussions than other students who were not so active. This suggested that an individualized instrument would be beneficial in describing the learning environment. Hence the use of the traditional class form describing the class as a whole could pose a problem if the purpose is measuring the environment through the eyes of an individual student. Also, at this time, the traditional role of the teacher transmitting content knowledge to students was challenged with a developing curriculum design interest in different types of pedagogy which could be used in classrooms. These influences might have helped to pave the way for designing different forms of learning environment instruments. The personal and class forms considered the role of the individual within the class and the class form considered the class as a whole (Fraser et al., 1996). Additional scales such as Personal Relevance and the promotion of understanding rather than rote learning were shifts in the development of the instruments.

The Constructivist Learning Environment Survey (CLES) considers a socio-constructivist model of learning. Broadly speaking, the constructivist view of learning acknowledges that students make sense of the world in relation to their knowledge that they consider and construct. It encourages students to develop deeper understanding, to challenge what they learn and how they learn, to see relevance in what they learn, to negotiate their learning, and to reflect on what and how they learn (Bell, 2005; Brooks, 2002; Fraser, 2014). The CLES measures personal relevance, uncertainty, critical voice, shared control and student negotiation. It has been used not only to measure and quantify students’ perceptions, but also to assist teachers to reflect on their assumptions and support them in making shifts in their teaching practice (Ebrahimi, 2015). In this study, the “What happens in my science classroom?” Student Actual Form and “What I wish would happen in my science classroom?” Student Preferred Form of the CLES were used. These were pertinent instruments for New Zealand teachers for measuring learning climate with a constructivist view in a science classroom. The CLES was also preferred for this study because of its capability to characterise specific dimensions of co-constructive pedagogy and because it has demonstrated strong factorial validity and reliability in several different countries (Fraser, 2012). The first version of the CLES was introduced in 1991 (Taylor & Fraser, 1991) and was compatible with von Glasersfeld’s (1981, 1988) viewpoint of radical constructivism. The intention of the CLES was to measure students’ perception of their learning environment in a personal form. Three key themes were focused upon in the instrument: the degree to which learners used prior knowledge and reflected on their prior knowledge; the degree to which the student had freedom and autonomy in learning; and, thirdly, the degree to which students had the opportunity to negotiate their understanding with their peers.

In 1997, revised versions of the CLES (Taylor et al., 1995, 1997) were developed from the initial version. The three themes were refined, based on the original version (Taylor & Fraser, 1991), and they also embraced the ideas of critical constructivism (Taylor & Campbell-Williams, 1993). These modified versions adopted five key dimensions of a critical constructivist learning environment from the learner’s perspective. The key elements measured were: how relevant the world outside of school is in the learning classroom; how much empowerment the student gains to express issues regarding teaching and learning; how much control is shared between the teacher and the student and the meta-cognitive awareness of the student; how much engagement and interaction between peers is there for improving understanding; and the extent to which science is viewed as ever changing (Taylor et al., 1995, 1997).

The CLES has been validated and applied in various science education studies in different countries including the USA, Taiwan, Western Australia and Korea. (e.g., Aldridge et al., 2000; Johnson & McClure, 2004; Kim et al., 1999; Nix et al., 2005; Peiro & Fraser, 2009; Taylor et al., 1997). It has also been used in a wide range of curriculum areas as well as science: mathematics, business studies, computer studies, and English (e.g., Aldridge et al., 2004; Ebrahimi, 2015; Koh & Fraser, 2014; Wanpen & Fisher, 2006). The CLES has been translated and adjusted to accommodate specific situations in both English and non-English speaking countries such as a Korean version (Kim et al., 1999), a Spanish translation for science students in Miami, USA (Peiro & Fraser, 2009), and an online version called the Constructivist On-line Learning Environment Survey (Taylor & Maor, 2000). Despite the wide use of the CLES, few published articles can be found that report its use in action research to promote constructivist learning environments in New Zealand.

The CLES is available in two forms: Actual and the Preferred (Taylor et al., 1995). The actual form considers the learning environment as perceived by the student, whereas the preferred form considers what environment is favoured by the student. The preferred form identifies the key elements as goals and value orientations (Fraser, 1998). The preferred form has the same five elements (scales) and the same number of items, but the wording is slightly changed to imply the ideal or the preferred environment. For example, the title of the preferred form is worded with “What I wish would happen in my science classroom” rather than “What happens in my science classroom” so that each item of the preferred form has “In this class I wish that….” Each form contains 30 items altogether with five scales and six items to each scale. The scales offer a five-point range with alternative responses such as almost always, often, sometimes, seldom and almost never. On the front page of each of the forms, there are details of the purpose of the questionnaire and directions on how to answer each question. The CLES in this research has 25 items with five items for each of the five scales. Table 1 provides a description of each scale (Personal Relevance, Uncertainty, Critical Voice, Shared Control, and Student Negotiation) (Taylor et al., 1997). A sample of one item from each of the scales is included.

This study applied an action research strategy under a quantitative research paradigm, firstly, to gain further understanding of students’ perceptions of their immediate learning environment in science lessons and, secondly, to invite reflection and intervention with the teachers using the data gathered. Action research is an approach that enables practitioners to develop themselves personally and professionally by investigating and evaluating their work (McNiff & Whitehead, 2005). Detailed information about the study, general guidelines and consent forms were distributed to the schools and participating teachers. Teacher professional workshops were held to inform the teachers about the potential benefits of using the CLES. Steps were taken to ensure that the CLES was carefully administered so that it would provide valid and reliable data for assessing the secondary classroom environments. To satisfy ethical considerations, questionnaires were numbered to ensure school, teacher, and student anonymity. Questionnaires were printed and administered to the participating students; each questionnaire was assigned numeric codes to maintain anonymity throughout the study. Coding was used to track participants and classes, so that the numbering protocol was used to link the actual, preferred and, later in the year, post-actual questionnaires for all participants.

Guidelines describing how the CLES would measure actual, preferred and post-actual student perceptions were provided and teachers were given an opportunity to ask questions, make comments about the overall design of the study and make any changes to the method of administration. In the second consecutive year of the study, the identical CLES instrument was administered to the new cohort of Year 9 (12–13 year-old) students, repeating the same administration protocols and working with the same teachers from each school. The study adopted a quasi-experimental design to establish a cause-and-effect relationship (Shadish & Cook, 2009) and allow examination of changes in data over time and the drawing of inferences from the intervention.

After the actual and preferred CLES questionnaire was administered to the participants, the data from each class were collated and analysed for class means. Detailed profiles for each teacher were constructed from the class mean scores, including graphical information for each scale: Personal relevance, Uncertainty, Critical Voice, Shared Control and Student Negotiation. Individual participant items were also reported as a summary, so that the associated classroom teacher could build further understanding of individual perceptions in their current class. The differences presented in profiles between the actual and the preferred learning environments were used to consider which aspects of the co-constructive classroom environment needed to be changed in order to align the actual environment more closely with that of the environment preferred by the students. Each CLES scale was examined and discussed in conjunction with the numerical values of class means. After this initial stage of analysis and reflection, and working closely with the teachers, professional development workshops were held to develop constructivist learning strategies over the course of the year. Subsequently, later in the year, the Actual CLES instrument was re-administered (post-Actual) to all participants to see if students’ perceptions of their current classroom climate had changed. Workshops with teachers were held to evaluate the post-Actual data and to target specific constructivist strategies to improve the learning environments.

The research was undertaken in the central North Island of New Zealand, with the 13 secondary schools located in the Bay of Plenty and Waikato regions. These regions cover 37,000 square kilometres of mainly rural and coastal land, and all the schools were situated in small towns. Nine of the schools were classed as rural and four classed as urban. All schools had a similar structure in the makeup of subjects taught at the junior level, with science taught by specialist science teachers who were qualified with a science degree and a graduate or postgraduate teaching diploma. In all the schools, there were three or sometimes four lessons of science per week depending on the timetable structure. Students remained together in the class for the entire year and their teacher taught them over the course of the year. There were 10% Pacifica,¹ 31% of NZ Māori,² and 58% NZ European³ students who were identified in terms of ethnicity in the first year of the study, and 5% Pacifica, 24% NZ Māori, and 69% NZ European students who were identified in terms of ethnicity in the second year. There were 57% girls and 43% boys in the first year with a total of 327 students who participated, and 62% girls and 38% boys in the second year with a total of 362 participants. Some of the schools in this research have students who live below the poverty line, with some children struggling to bring lunch to school. However, there is a wide cross-section of social demographics across the schools. The research involved 15 teachers whose participation was through invitation but entirely voluntary. Each year, the CLES was administered consecutively to each science class of Year 9 (12–13 year-old) students. The actual/preferred forms of the CLES were administered in March and the (post)Actual form was repeated in November, in both years.

Fraser and Fisher (1986) have proposed a plan for changing the classroom environment. Teachers can use information obtained from quantitative instruments such as the CLES to guide attempts to improve their classroom environment. This tactical approach seemed suited to the type of situation and context of secondary-school science learning and teaching in this project. Other researchers have used this tactic with a variety of instruments (Fisher & Fraser, 1991; Fraser et al., 1988). The steps for changing the classroom environment in summary are as follows (Fraser & Fisher, 1986):

These steps were used to plan the professional learning sessions with the teachers in the project and maintain a consistent process to the methodology. The profiles were carefully considered in the reflection phase. Specific learning environment scales were identified for each class and an individual classroom improvement plan was developed after discussions with the teachers. Because statistically-significant differences were identified for scales of the CLES across the classes, all aspects of the co-constructive learning (Personal Relevance, Uncertainty, Critical Voice, Shared Control, and Student Negotiation) were selected for improvement. Classes had a range of differences and class means. The intervention was introduced to the classes in order that all participants would be aware of and agree about attempts to improve each aspect of the learning environment. The intervention was initiated in teacher professional learning workshops starting in May and lasted for six months. At the end of the intervention, the Actual Form of the CLES was re-administered to determine whether the students perceived their actual environment differently. These reassessment results were used to indicate whether changes in the learning environment had been achieved.

Students’ responses to the frequency scale consisting of Almost Never, Seldom, Sometimes, Often and Very Often were scored 1, 2, 3, 4 and 5, respectively. The data were analysed using SPSS and various analyses were conducted to measure validity and reliability of the CLES: factorial validity, internal consistency reliability, and the ability to differentiate between the perceptions of students in different classrooms.

Factor analysis is a mathematical technique designed to take a large set of variables and summarises them using a reduced set of factors or components (Pallant, 2016). There are two challenges to consider in determining whether a particular data set is suitable for factor analysis, namely, sample size and the strength of the relationship among the variables. Tabachnick and Fidell (2013) recommend at least 300 cases for factor analysis but the general suggestion is the larger, the better. Yamini and Rahimi (2007) advocate a ratio of participants to items of 10 to 1 (i.e. 10 respondents for each item). In this study, the ratio was approximately 20 to 1 (more than 20 participants for each item of 25-item CLES).

One of the important considerations in the field of learning environment research is the reliability and validity of the scales used in an instrument (see Table 2). The reliability of a scale indicates how free it is from random error (Pallant, 2016). One frequently-used indicator of a scale’s reliability is internal consistency which is the degree to which the items that make up the scale are all measuring the same underlying attribute. The most-common indicator is Cronbach’s coefficient alpha, which provides an indication of the average correlation among all the items that make up a scale. With values ranging from 0 to 1, with higher values indicating greater reliability, Nunnally (1978) recommends a minimum level of 0.7. Validity refers to the degree to which a scale measures what it is supposed to measure. The discriminant validity of a scale is measured by its relationship with other constructs, with one of the methods being to use the mean correlation with other scales. Another desirable statistical analysis for differentiating between the actual and preferred perceptions of students is the t-test. T-tests are used when two sets of data are analysed (e.g. before and after) and compared using the mean score on a continuous variable. In this research, paired sample t-tests (repeated measures) were used to compare participants’ actual and preferred scores, as well as actual and post-actual perceptions, for both years.

Because they were the same participants tested each time in each year (Pallant, 2016), students in the same class would perceive their class relatively similarly, while mean within-class perceptions would vary from actual and preferred, and from actual and post-actual measurements. The results of these analyses, reported in Tables 5, 6, 7 and 8, indicate that scales differed significantly (p < 0.01 and p < 0.001). If this p-value is less than 0.05, there is a significant difference between the two scores. Therefore, the measurements in this research show significant difference in the range of scales. Although the magnitudes of statistically-significant differences in the following tables are modest, nevertheless, they are all positive and identify important differences.

Table 2 provides a summary of previously-reported statistical information for the CLES (Ebrahimi, 2015; Ogbuehi & Fraser, 2007; Taylor et al., 1997). This information includes each scale’s internal consistency reliability (Cronbach alpha coefficient) and discriminant validity (using the mean correlation of a scale with other scales in the same instrument).

Table 3 provides reliability estimates for 327 students in the first year and 362 students from the second year of my New Zealand study. Cronbach alpha coefficients in Table 3 demonstrate that the reliability measure for each CLES scale was robust. With the student as the unit of analysis, scale alpha reliabilities ranged from 0.74 to 0.88 for the Actual Form and from 0.80 to 0.89 for the Preferred Form for the first year. In the second year, alpha reliabilities ranged from 0.80 to 0.85 for the Actual Form and from 0.81 to 0.89 for the Preferred Form. This suggests that all scales of the CLES possess satisfactory internal consistency in both Actual and Preferred Forms for both years of this study (see Table 2). The results in Table 3 compare favourably with past research in terms of internal consistency reliabilities in secondary and tertiary classes (Ebrahimi, 2015; Ogbuehi & Fraser, 2007; Taylor et al., 1997).

The discriminant validity of the CLES was measured using each scale’s mean correlation with the other scales. Table 4 reports three statistics from the second year, namely the Actual, Preferred and Post Actual Forms of the mean correlations. The mean correlations ranged from 0.23 to 0.49 as the unit of analysis. This range indicates that the instrument has acceptable discriminant validity and each scale measures generally distinct although to some extent overlapping aspects of the constructivist learning environment.

The results in Table 4 for the mean correlation of a scale with other scales compare favourably with past research in secondary and tertiary classes (Ebrahimi, 2015; Ogbuehi & Fraser, 2007; Taylor et al., 1997) (see Table 2).

Overall, the data presented support the contention that the CLES is a reliable and valid learning environment instrument for the assessment of secondary students’ perceptions of science learning environments in New Zealand classrooms.

The CLES provided specific information about the students’ actual perceptions and preferred perceptions of their learning environment. The initial administration of the CLES provided results for the Actual and Preferred scores, whereas the final administration later in the year provided results for the Post-Actual means of the classes.

Table 5 reports the mean scores for the Actual and Preferred Forms of the CLES in the first year, the corresponding standard deviations, the mean Actual–Preferred difference for each scale and results of paired sample t-tests for differences between the Actual and Preferred values. Actual–Preferred differences were significant for all the scales except the Uncertainty scale. These differences indicate that students preferred a greater opportunity to experience relevant everyday contexts in their science lessons, express their opinions in class, discuss ideas with their peers and plan their learning with the teacher.

For the second year, Table 6 reports that the Actual–Preferred differences year were statistically significant for all the scales except Personal Relevance. These differences indicate students’ preference for a greater opportunity to express their opinions in class, plan their learning with the teacher and learn about science. Overall, results for the first year and the second year clearly suggest that the New Zealand students were not satisfied with their current classroom environments and preferred a more-constructivist classrooms than the ones they perceived as being actually present in terms of the five dimensions of the CLES. These differences between students’ actual and preferred environments in this study are consistent with past research which has explored the congruence between actual and preferred environments in a number of countries (Ebrahimi, 2015).

Tables 7 and 8 present results for the statistical significance of difference between Actual and Post-Actual forms of the CLES. For the first year, differences between Actual and Post-Actual scores in Table 7 were statistically-significant for Personal Relevance, Critical Voice and Shared Control. This pattern highlights the shifts in student perceptions over the course of the year.

Table 8 reports mean scores for the Actual and Post-Actual Forms of the CLES in the second year. The largest difference occurred for Personal Relevance whose Actual scale mean of 3.14 increased in the Post-Actual to 3.32. For the scales of Uncertainty and Shared Control, means shifted slightly with a mean difference of 0.12 or more.

Actual Personal Relevance difference remained consistently significant for both years, perhaps due to the sustained teacher professional development focus of context and topic design. Critical Voice and Student Negotiation differences were not significant in the second year and there was little change in mean values. This could have been due to the teachers being able to develop contexts of Personal Relevance more easily in their lessons than strategies to develop Critical Voice and Student Negotiation. The teachers examined the mean values for all scales closely, each teacher had her/his particular class mean results and comparisons were discussed with respect to the overall mean values from all the classes.

In terms of the nature of professional learning workshops, this paper does not provide a full account of the intervention process, but the following briefly describes key aspects of the content and delivery. Firstly, the teachers identified that there was a need to make changes to future science topics in the programme to provide greater emphasis on current social and political issues. Early on, teachers recognised from the low scores for the Personal Relevance scale that more could be developed in terms of ensuring that lessons had contexts linked with real-life scenarios. This exercise was challenging, but juggling science content knowledge with social relevance in the topics fuelled teacher dialogue in the workshops. However, the discussions proved valuable in that teachers could develop their own topics through their interpretations of the CLES items for Personal Relevance. Prominence was placed on programme topics using current events and how these could lift Personal Relevance Actual scores. Contextualising the programmes was identified by the teachers as a lever to build greater engagement to the learning, and this was recognised as helping to connect the student’s world outside school to what happens in science lessons. Teaching and learning programme designs were customised so that local and New Zealand contexts were used, with specific details being shared by teachers at workshops. Examples included a sporting event with emphasis placed on human biology and diet, a local environmental challenge involving restoring a forest, a music event which had used thousands of plastic bottles that ended up in a land fill, and an earthquake that had killed people. The curriculum changes were carefully constructed to include specific scientific content knowledge linked with the context. Towards the end of the series of workshops, teachers commented that they were pleased to be teaching programmes designed with far more relevance, and that they felt directly involved in the process of the teacher professional development.

Critical Voice and Shared Control also are CLES scales which revealed illuminating differences. The professional development sessions provided teachers with opportunities to share experiences to identify classroom strategies which would enhance scores on these scales. Teachers shared their beliefs and values around the practice of increased student voice in the classroom and subsequent workshops included techniques to enhance student dialogue. Teacher professional development also considered classroom interactions between both teacher to student and between students. Development of specific classroom strategies to enhance co-construction between teacher and student included: shared learning goals; greater choice of learning tasks; and activities for which students helped to plan a series of lessons in an inquiry. Teachers spoke positively about how their students presented their scientific findings to other class members through the use of regular focussed episodes of evaluation, which is something that they had often overlooked in the past.

Tables 5, 6, 7 and 8 summarise results for CLES Actual, Preferred and Post-Actual means in the first and second years, especially the variation in mean values for each scale and the shifts in scores from Actual to Preferred to Post-Actual. The results demonstrate a similarity in the differences for all scales, with particular reference to the low means for Shared Control for both years. Student Preferred perceptions of Shared Control are much higher than their Actual perceptions, with Actual mean scores of 2.22 and 2.17 and Preferred mean scores of 3.08 and 3.10, respectively for both years. Subsequently, the need for teaching strategies to incorporate co-operation and goal sharing for students was specifically identified. The Student Negotiation scale examines students’ perception of talking to others and learning to communicate. This had the highest mean values compared with other scales in both years; however, Preferred perceptions remained higher than the Actual perceptions, signalling that further development was required in this specific scale. This involved opportunities in the lessons for students to interact with their peers, experts, and scientists, and hence contribute to a shift in power relationships between teachers and students. A focus in re-designing lessons was to support further student autonomy and agency, using talk and composition as a window into their developing knowledge about science.