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

In this article, I report a two-year study of working closely with science teachers and examining perceptions of Year 9 (12–13 year olds) students in 13 New Zealand secondary schools. The Constructivist Learning Environment Survey (CLES) was used. The questionnaire was administered to 327 students in the first year and 362 students in the second year to find out their perceptions of their preferred learning environments in order to compare this with their perceptions of the actual situation. The data were used to plan improvements in learning environments through a teacher professional learning process by redesigning programmes and encouraging changes in students’ classroom behaviours. Co-constructive learning strategies and reshaping of lessons to include current topics were used as tools to encourage students’ expression of opinions, personal relevance and shared control in their learning. After each subsequent year-long intervention, the CLES was re-administered to reveal that there had indeed been an improvement. The research is distinctive because it is a learning environment study in New Zealand schools examining students’ psychosocial perceptions related to their first-year experiences of science at the secondary level.


Introduction
For some time, there has been prevailing unease about the declining numbers of students choosing careers in science and technology (Bolstad & Hipkins, 2009). Reasons for this decline include a shift in the balance between activity-based and more-transmissive teaching methods, student perceptions of a lack of relevance of the topics, and a decline in opportunities for students to express their opinions in lessons (Tytler et al., 2008). Shirazi (2017) claims that there is a strong relationship between positive student experiences in their science learning as adolescents and the likelihood that students will continue to pursue science after this age. New Zealand data show a decline in engagement and attitude which begins at Year 8 (11-12 year-olds), accelerates at Year 9 (12-13 year-olds) and 1 3 continues into Year 10 (13-14 year-olds) (Crooks et al., 2008). By Year 9, there is a noticeable cooling of general interest to science and few students appear to actively foresee a possible adult role in a science career (Bolstad & Hipkins, 2009). A New Zealand Council of Education Research national survey (Bonne & MacDonald, 2018) of New Zealand secondary schools reports that science teachers were the least likely group of teachers to say that their students sometimes or often have learning experiences in which they work together on a project or activity to make a difference to their class, school, local environment or community. Furthermore, they were the least likely group of teachers to say that their students investigate their own questions, make connections with things in their own culture or life outside of school, or discuss different ways of looking at things. Collectively, these differences tend to support a view of science, somewhat more than other subjects, as a subject in which developing co-constructive approaches in the learning environment is not a high priority.
Focusing on the development of engaging teaching practices and experiences in the science classroom could encourage students to stay in science (Maltese & Tai, 2010). Co-constructivist teaching methods can offer successful engagement strategies to science education to reignite motivation (Shirazi, 2017). Previous researchers examining coconstructivist practices (Driver & Oldham, 1985;Fleer et al., 2007;Fraser, 2018;Nix & Fraser, 2011;Tobin, 1990) characterise an emphasis on constructing knowledge and meaning through experience with others, with learners engaged in active participation and helping each other learn through a collaborative social process (Watkins et al., 2007). There are different forms of constructivism (Gilbert, 2017), whose roots typically can be traced back to Kelly's (1955) personal construct psychology in which learning is the building of an individual's intellectual constructs, although this construction can take place not in individuals alone but in the spaces between them. Past research has identified that, when teachers develop co-constructivist learning environments, importance is placed on: creating opportunities for students to air their developing ideas with others; students having an opportunity to decide and share control of learning tasks; activities being managed by the students themselves; and, importantly, teacher facilitating and monitoring the learning (Kaufman, 2004;Nuthall, 2007;Watkins, 2015).
New Zealand secondary schools have been developing teaching and learning programmes using a national curriculum (Ministry of Education, 2007) which encompasses significant changes to the way in which teachers are expected to teach using overarching Key Competencies which describe a set of capabilities for living and lifelong living (e.g. Managing Self, Relating to others, Participating and Contributing, Thinking, and Using Language, Symbols, and Texts). Because the introduction of these, rapid changes to science programmes have been anticipated in terms of the nature of design and co-constructive pedagogy. To catch up with these changes, teachers are expected to place emphasis on how they teach rather than what they teach. However, recent reports indicate that many science teachers remain hesitant to make changes, and some are challenged to know how to begin (Hipkins, 2015). Goldspink and Foster (2014) note that teachers want to do the best by their students but, for many, this means doing their best within existing systems and practices. They argue teacher reluctance to change from traditional practices can be due to anxiety from expectations of high stakes content coverage and summative assessment.
To help address the tensions highlighted, it is necessary to provide researchers and educators with instruments with which they can assess the degree to which a particular science learning environment is consistent with different types of pedagogy. Feedback from instruments can help science teachers to reflect on their assumptions and change their teaching practice and policies towards more constructivist ones in science classrooms (Bell, 2005). Some instruments aid researchers not only to investigate how constructivist education affects science students' final outcomes, but also to assess the effects of constructivist education ideas on students' actual and preferred perceptions of their learning environments (Khine & Lourdusamy, 2005). In this research, the rationale to measure perceptions of 'preferred' or ideal classroom environment as well as the 'actual' or experience classroom environment was to gain insight into the goals and values orientations of the participants' ideal learning environment. The Preferred items in the questionnaire are worded slightly different from the Actual form. For example, an item in the actual form such as "It's Ok for me to express my opinion" would be changed in the preferred form to "I wish it was Ok for me to express my opinion".
This article describes an existing instrument (i.e. the Constructivist Learning Environment Survey, CLES) for assessing science students' perceptions of the psychosocial environment that should exist in constructivist classrooms, and reports comprehensive validation information for this instrument for a sample of science students from New Zealand. The CLES was administered each year for two years with one cohort of Year 9 (12-13 year-olds) students and then repeated the following year with the same teachers with the new cohort of Year 9 students. Because the students were attending their first year at a secondary school, it most likely was the first-time when they experienced science taught by a specialist science teacher. The study supported teacher professional learning and acted as a stimulus to encourage critical professional discourse with the science teachers. It offered the opportunity to make measurements of the learning environment using the CLES and help to strengthen teacher pedagogical knowledge of co-constructive learning environments. Because there had been little research in New Zealand involving the use of the CLES among science teachers teaching students in their first year of secondary schooling, this was the purpose that was focussed on.
In this article, the purpose of the research is established first before past learning environments research is summarised, including the field's quantitative instruments and specific aspects of the CLES such as its underpinning constructivist paradigm, the rationale for having two forms (Actual and Preferred) and a description of each of the five scales in the instrument. A section describes the methods used in the research before a brief description of the reliability of the CLES, especially the internal consistency of the Actual and Preferred forms. The following sections outline key findings and specific results for the Actual and Preferred forms, comparisons of Actual and post Actual measures. In terms of the nature of professional learning workshops, this article does not provide a full explanation of the intervention process, but a section describes the key features that have been identified in the teacher professional development. Following this, further analysis focuses on comparisons between the two years. Finally, limitations of the study, suggested future directions and a conclusion are provided.

Purpose of the study
The aim of this paper is to report results from using the CLES, including its internal consistency reliability and tests of statistical significance for difference between Actual and Preferred, Actual and Post Actual results. The rationale behind the research was to explore the learning environments of Year 9 (13-14 year-olds) students as they experienced science education for their first year at a secondary school. The research question was "What are students' attitudes and perceptions of their experiences in Year 9 Science?" The specific purposes of this study were to: • Provide validation data for the use of the Constructivist Learning Environment Survey (CLES) in New Zealand secondary schools. • Explore how science learning environments for the Year 9 students could be developed by teachers using co-constructivist approaches. 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).

Development of instruments
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 cooperative 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(Fraser, , 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.

Learner perceptions of their learning environment
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 . Additional scales such as Personal Relevance and the promotion of understanding rather than rote learning were shifts in the development of the instruments.

The CLES
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 (1981Glasersfeld's ( , 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(Taylor et al., , 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(Taylor et al., , 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.

Overall design and approach
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.

Procedures
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.

Participants
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 European 3 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):

Plan
1. Assess student actual and student preferred perceptions of the classroom environment. 2. Draw profiles of student actual and student preferred perceptions. 3. Reflect upon the profiles and consider intervention strategies. 4. Establish an intervention to make change to the classroom environment. 5. Reassess student actual perceptions.
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.

Data analysis
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 Table 2 Alpha reliability coefficients and discriminant validity using (mean correlation with other scales) for CLES a (Taylor et al., 1997), CLES b (Ogbuehi & Fraser, 2007) and CLES c (Ebrahimi, 2015) CLES a N = 1081 high-school science students (Taylor et al., 1997) CLES b N = 661 middle-school mathematics students (Ogbuehi & Fraser, 2007) CLES c N = 622 English-language student-teachers (Ebrahimi, 2015)  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 postactual 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).

3
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.

Differences between actual and preferred forms of CLES
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.

Differences between actual and post-actual forms of CLES
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.

Teacher professional learning
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.

Further analysis
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.
During the year, students were encouraged to work in small groups, to have greater autonomy in planning their own investigations with the support of their teacher, and to make suggestions to improve the learning in their groups. An intervention strategy involving the use of small groups of students working collaboratively was applied to all classes, with students assigned roles of specific responsibility such as technician, leader and researcher in the science lessons. In the early part of the year, students appeared to be lukewarm about their actual experiences of sharing control, but their preferences showed much higher values and, by the end of the year, higher Post-Actual perceptions were found. Tests of significance of differences between Actual and Post-Actual, (e.g. t-value of 4.42 for Personal Relevance and 3.62 for Shared Control in the first year) demonstrate the the shift on these scales. Associated with this shift was an increase of student engagement levels, with teachers commenting that they were pleased to see their students sharing greater control with them and their peers. The teachers spoke about how they had incorporated strategies such as seminars into the programme with students in their groups presenting their current progress on topics; this had led to a lift in student interest and motivation. Both of these positive impacts on student engagement are a result of the active process of developing a more constructivist learning environment in the science lessons.

Limitations and strengths
The study was conducted with 30 science classes in 13 schools in the central North island of New Zealand. Therefore, findings are limited to the context of those schools and their student profiles; the sample was neither large enough nor representative enough to generalise findings to all Year 9 secondary students in New Zealand. A further constraint to the findings is whether there were student perception differences between schools or student ethnicities. Class means across the range of the five CLES scales did vary from school to school, but what was interesting is the Shared Control and Personal Relevance were the prevalent scales that stood out with the lowest values across all 13 schools. However, the scope of this study did not include comparing student perceptions for different schools, student ethnicities, and teachers. The CLES measures in this study identified specific actual and preferred differences from each class, and this was a valuable tool for the teachers in examining student perceptions and making informed changes to the learning and teaching. While the teachers had access to their own class means and individual student data, further analysis of comparisons between the classes and schools would have been worthwhile because the results reported in this article assume that the differences were uniformly distributed across all the schools. Also omitted from the study was the effect of constructivist learning on conceptual understanding because this would have required standardised topics across all the classes. However, there were strong signs of improved student dialogue and engagement through the development of co-constructivist experiences across the classes. A future replication of this research incorporating effects on conceptual understanding is strongly encouraged.
The exploratory nature of this research emphasised teaching and learning processes through a co-constructive view in relation with learning environment in Year 9 science classes. The opportunity for teachers to inquire into their own learning by gaining knowledge of students' perceptions appeared to be a valuable process in making cognisant steps to bring about change in the classes. It signals how the perceptions of the 12-13 year-old students in this study are dynamic and responsive particularly to personal relevance in topics and sharing control with the teacher. There were varying views of how students perceived their learning in science, but the opportunity to analyse these using the CLES was a robust mechanism for the teachers for reflecting on their practice and making informed steps into their own professional learning.

Conclusion
This research is unique because it has examined students' perceptions related to their firstyear experiences of science at secondary schools in New Zealand. An analysis using the CLES showed that students preferred a greater co-constructive and participatory learning environment than what was present in the actual environment. Because the first purpose of this study was to report the validation of the CLES in New Zealand secondary schools, this instrument was field tested with a combined sample of 689 students in 30 classes. The questionnaire exhibited strong validity and reliability in its actual, preferred and postactual forms. The second purpose was to develop the learning environment for students and teachers by making it more co-constructive. Through the use of the CLES, specific teaching practices and pedagogy were identified through the action-research initiative which supported teacher professional learning. This targeted the changes towards the preferred students' perceptions of their learning environments, which required intentional and explicit implementation. The CLES identified statistically-significant differences between actual and preferred perceptions for Shared Control, Critical Voice, Student Negotiation and Personal Relevance for both years.
This study was also a response to the lack of learning environment research in New Zealand, particularly the absence of quantitative measurements using questionnaires. By reporting data specifically for a New Zealand sample, it paves the way for future research on learning environments in New Zealand and, importantly, the opportunity for teachers to make informed changes to their junior science learning and teaching programmes. Using the CLES, this study has shown that New Zealand students were not satisfied with their current learning environment and preferred a more co-constructivist environment on all scales. Classroom environments in secondary science can be improved by connecting lesson activities and knowledge to students' everyday out-of-school experiences (i.e. Personal Relevance) and providing opportunities for students to experience the control of design and management of tasks (i.e. Shared Control). Future professional learning opportunities could be provided so that science teachers have the ability to develop their learning environments with the use of quantitative measures which support constructivist ideals. This study can assist those science educators who want to develop constructivist, collaborative and student-centred classroom environments.