Adaptation and Validation of the DUNDEE Ready Education Environment Measure (DREEM) for a Practice Teaching Program

The study examined the psychometric properties of the DUNDEE Ready Education Environment Measure (DREEM) in the context of a practice teaching program environment. The DREEM has been validated in many countries and is a widely recognized tool to measure students' perception of their learning environment in medical and allied health education programs. Data for this study come from 316 fourth year teacher education students who have already finished their practice teaching program. Principal components analysis (PCA) with Varimax rotation was used to examine the underlying component structure of the items of the instrument. PCA results indicated the presence of a five-component structure which supports other validation studies. However, component loadings of the items were not consistent with the original DREEM resulting to a shortened version. The shortened DREEM has a Cronbach alpha of 0.83, indicating a high internal consistency of its items. The adapted DREEM is a valid and reliable instrument for measuring practice teachers' perception of their learning environment. The instrument will be called Practice Teacher Education Environment Measure (PTEEM).

undergraduate students in other allied health programs (Patil et al, 2016;Bakhshialiabad, et al, 2015: Koopayehzadeh, 2014Khursheed, I., & Baig, L., 2014;Hamid et al, 2013, Brown et al. 2011Aghamolaei & Fazal, 2010;and Edgren et al, 2010). In the Philippines, two published studies were found utilizing the DREEM for determining the perceptions of students concerning their learning environment. Barcelo (2016) compared the perception of the academic learning environment between medical laboratory science students and nursing students at Saint Louis University, Baguio City, Philippines. Overall result show that the perceptions of medical laboratory science students on their academic learning environment were not significantly different from those of the nursing students. Bay Jr. B. E. & Subido, H. (2014) adopted the DREEM to evaluate the learning environment of the College of Dentistry in Lyceum of the Philippines University. The study found that dental students perceived their academic environment as conducive to learning. However, both studies did not present any validation process done to the instrument before it was used for data gathering.
Studies about adoption and adaptation of the DREEM in non-medical curricular programs are scarce. Atapattu et al, (2015) used a modified version of the DREEM questionnaire to assess the learning environment of students enrolled in the Bachelor Science (Agriculture) degree program in Sri Lanka while Riga (2015) adopted the Greek version of the scale to measure the practical training environment of students enrolled in Educational Studies and Early Childhood Education at the University of Patras in Greece. To the author's knowledge, no study has been conducted about validating the DREEM instrument in the Philippines, hence, this study. The aim of this study is to evaluate the psychometric properties the DREEM as a measure for assessing the learning environment of the practice teaching program of the teacher education curriculum in a Philippine university.

Instrument
The Dundee Ready Education Environment Measure (DREEM) consists of 50 statements, 41 of which are positive and nine are negative. Each item is rated using a five-point Likert-type scale ranging from 0-strongly disagree to 5-strongly agree. The instrument contains five subscales namely; perceptions of learning, perceptions of teachers, academic self-perceptions, perceptions of atmosphere, and social self-perceptions. Although the authors claim that the DREEM is culture-free, changing some words and phrases in the original items that are not in the context of practice teaching is necessary. For instance, the words teaching, teachers, patients, and department were replaced with practice teaching, cooperating teachers, practice teachers and cooperating school, respectively. A team composed of five cooperating teachers evaluated the items that were modified to ensure its relevance and appropriateness for the context of the study. Thirty practice teachers were asked to answer the modified item pool and also to give their comments and suggestions to ensure the meaning and clarity of each statement. The instrument was further modified by incorporating all the comments and suggestions from the cooperating teachers and practice teachers. The final form of the instrument was pilot tested to a group of practice teachers who were not part of the study. The pilot test revealed an internal consistency reliability of 0.83 which is considered to be adequate (Nunnally, 1978). Moreover, the researcher was granted permission by the authors to adapt and validate their instrument in this study.

Sample
Three hundred ninety-eight practice teachers chosen randomly from two batches of the Bachelor of Elementary Education and Bachelor of Secondary Education of Leyte Normal University in the Philippines anonymously completed the modified DREEM questionnaire. Permission from the practice teaching supervisors was sought before the administration of the questionnaire. The researcher conducted an orientation with the practice teachers on the purpose of the research and how the questionnaire will be answered before they completed the scale. Participation in the study was voluntary. The scale was administered after the students have already completed both in-campus and off-campus shifts of their practice teaching program.

Data Analysis
Data for this study were subjected to Principal Component Analysis (PCA). This is a statistical technique used to reduce the dimensionality of uncorrelated variables (in this case, items in the DREEM instrument) into a smaller set retaining most of the information of the original data set (Dunteman, 1989). Aside from reducing dimensionality, PCA is also used to modify or shorten an instrument and to explore whether the structure of the original version is supported by a sample data set (Kellows, 2007). The analysis followed the general guidelines for principal component analysis. Before performing PCA, suitability of data for analysis and sampling adequacy was determined through the Kaiser-Meyer-Olkin (KMO) measure of Sampling Adequacy. Field (2005) states that KMO values between 0.5 and 0.7 are mediocre, values between 0.7 and 0.8 are good, values between 0.8 and 0.9 are great and values above 0.9 are superb. Bartlett's test of sphericity was performed to assess whether the data set is suitable for reduction. The Catell's Scree test and parallel analysis as recommended by (Costello & Osborne, 2005;Hayton, J. et al, 2004) and the interpretability criteria were used in deciding on the number of meaningful components to be retained. A given component is retained if it has a factor loading of at least 0.40 (Stevens, 1986) and items that load to more than one component are candidates for deletion. An item that does not load to any component is also a candidate for deletion. Cronbach alpha was used to determine the internal consistency of the revised DREEM instrument. An item has an acceptable level of internal consistency if its Cronbach alpha is at least 0.70 (Nunally, 1978;Streimer & Norman, 2008). An item is considered to contribute highly to the construct being measured if its corrected item-total correlation has a value of more than 0.3 (Yusoff et al., 2010).

Principal Component Analysis
Out of the 398 practice teachers who completed the revised DREEM only 316 data sets were used in the analysis because questionnaires with missing cases were excluded. There were 82 missing cases spread through 42 out of the 50 items in the scale. Results of the Kaiser-Meyer-Olkin (KMO) sampling adequacy measure was 0.846 while Bartlett's Test for Sphericity was 4848.007, significant ( ≤ 0.001). These results confirmed that the data set was suitable for principal component analysis. The scree plot showed a sharp point of inflection after the first component ( Figure 1) and started to level off after the fifth, suggesting 5 subscales.  (Table 1) shows that there are five components from the actual data whose eigenvalues exceeded the 95 th percentile eigenvalues. The scree plot and parallel analysis results were the bases for retaining number of meaningful components. After the rotation, components 1 to 5 accounted for a total variance of 39.59%.  29,37,40,2,18,6,3,32 and 1 refers to students' perception of their cooperating teachers. The second component contained items 10, 22, 16, 45, 13, 31, 30, 12, 20 and 15. These items are about students' perception of teaching. Items 23, 11, 34, and 24 made up the third component which refers to students' perception of atmosphere. The fourth component consisted of items 28,21,41,27,42,47,14 and 43 describes students' perception of learning. Finally, the fifth component refers to negative perception of cooperating teachers and teaching consists of items 35,39,9,8,4,50 and 48. Further analysis show that items 5,7,17,19,25,26,33,36,38,44,46 and 49 have loadings which are less than .40, hence these were deleted. Component loadings ranged from .40 to .722. Table 2  The teachers are good at providing feedback to students. .702 37 The teachers give clear examples.
.691 40 The teachers are well prepared for their classes.
.640 2 The teachers are knowledgeable.
.631 18 The teachers have good communications skills. .602 6 The teachers are patient with students. .592 3 There is a good support system for students who get stressed. .499 32 The teachers provide constructive criticism.
. My social life is good. 44 The teaching encourages me to be an active learner. 23 The atmosphere is relaxed during lectures. .722 11 The atmosphere is relaxed during the practice teaching. .699 34 The atmosphere is relaxed during seminars/tutorials. .665 24 The teaching time is put to good use. .432 26 Last year's work has been a good preparation for this year's work. 25 The teaching over-emphasizes factual learning. The teaching is too teacher-centered.
.409 17 Cheating is a problem in this school 38 I am clear about the learning objectives of the course. 7 The teaching is often simulating.

Extraction Method: Principal Component Analysis; Rotation Method: Varimax with Kaiser Normalization
Further examination of the individual items whose component loadings were .40 and above, revealed that some items loaded to a component with different conceptual meaning. For instance item 3 (There is a good support system for students who get stressed) and item 1 (I am encouraged to participate in class) loaded to component 1(perception of cooperating teachers). Similarly, item 15 (I have good friends in this school) loaded to component 2 (students' perception of teaching). Item 30 (There are opportunities for me to develop interpersonal skills) loaded to component 1 (perception of cooperating teachers) and component 2 (perception of teaching). These items also did not pass the criteria for inclusion in the final item grouping of the instrument. A total of 17 items were eliminated reducing the modified DREEM instrument to 33 items. The final item grouping is presented in table 3. The subscales of the shortened instrument are labeled as: perception of teachers, perception of teaching, perception of atmosphere, perception of learning, and negative perception of teachers and teaching. The teachers get angry in class.
.826 9 The teachers are authoritarian. .824 8 The teachers ridicule the students. .825 4 I am too tired to enjoy the course. .828 50 The students irritate the teachers. .839 48 The teaching is too teacher-centered. .832

Reliability Analysis
Three internal consistency analysis were done for this study. The first was the overall Cronbach alpha of the modified 50 -item DREEM scale, second was the analysis done for the shortened DREEM and the reliability analyses for each of the subscales of the shortened instrument. Results show that the Cronbach alpha of the modified DREEM and the shortened DREEM were 0. 86 and 0.83 respectively and the subscale reliability analysis resulted to Cronbach alphas ranging from 0.71 to 0.80.

Discussion
The aim of the current study was to adapt and establish the validity and reliability of the DREEM scale in a sample of practice teachers from a state university in the Philippines. To explore the underlying structure of the adapted DREEM, a principal component analysis with Varimax rotation was performed using the data set gathered from 316 practice teachers. Result of the parallel analysis was used in deciding the number of components retained in the final form of the scale since it is considered as the most accurate technique as recommended by Ledesma & Valero -Mora (2007). The result of this study did support the five-component structure of original DREEM instrument, however, loading of the items to specific components were inconsistent with the original loading. For instance, the components academic self-perceptions and social self-perceptions were not identified in this study, instead For instance, in the original DREEM, the components are perception of learning, perception of teachers, academic self-perceptions, perceptions of atmosphere and social self-perceptions, in this study, component are named as perception of teachers, perception of teaching, perception of atmosphere, perception of learning and negative perception of teaching and teachers while in this study, the five comp perceptions of cooperating teachers, perception of teaching, perception of atmosphere, perception of learning and negative perception of teaching and teachers. These findings are consistent with results reported by (Schonrock-Adema 2009;Khan et al. 2011;Jakobson et al. 2011;Yusoff 2012;Hammond et al. 2012;Vaughan et al. 2014;Mogre & Amalba, 2016;Leman 2017;). As has been suggested by the previous studies, removal of specific items might improve the goodness of fit of the five-factor structure (Dimoliatis et al., 2010;Hammond et al., 2012;). The mismatch in the item loadings and subscales between the original and the shortened DREEM may be attributed to disparity of the environments and samples where the study took place. The original DREEM was intended for medical environments and medical undergraduate students while in the current study the environment are the laboratory schools where practice teaching is done and students are teacher education students. After eliminating items that did not satisfy the interpretability criteria, although retaining the original five subscales the instrument was reduced to 33 with the following proposed subscales: Perception of Cooperating Teachers, Perception of Teaching, Perceptions of Atmosphere, Perception Learning and Negative Perception of Teachers and Teaching.
Reliability analyses for both adapted and shortened version of the DREEM resulted to items with acceptable internal consistency. Similarly, its subscales have acceptable Cronbach alpha values which are acceptable. These values show that the five dimension of the shortened instrument also contain items with acceptable internal consistency.
There are some limitations of this study. First, the student teaching program is typically divided into shifts consisting of the on campus and off-campus shifts. These students while undergoing the student teaching program are assigned to different learning environments, the cooperating schools for the off-campus shift and the laboratory school in the university for the in-campus shift. It is possible that they were confused as to which of the in-campus or off-campus learning environment they should assess hence the mismatch in item loading with the original instrument. The sample size in this study could also be an issue considering that there were only 316 respondents, short of the ten respondents per item as suggested by Costello & Osborne (2005).

Conclusion
This study did support the five-component structure of the DREEM, however, loadings of items to the five dimensions in the shortened version were inconsistent with the original DREEM. With minor modifications of some items which seem to have different conceptual meaning from the rest and considering that the items in each subscale have reasonably good internal consistency indices, the shortened version may be use to assess practice teachers' perception of their learning environment. The shortened instrument will be called as Practice Teachers'