The participants in the current research were 216 Indonesian preservice biology teachers from one public educational studies-based university located on the western Java Island. The university itself is known as the central university for preparing future Indonesian teachers, and also contributes to the establishment of the curriculum for Indonesian primary and secondary school levels. The participants joining the current research were in their second (30%), third (38%) and fourth (32%) year, and were affiliated to the biology education program. One of the issues in most biology education programs in Indonesian universities is the limited number of male students; and so in the current research, only 15% of the total participants were males. We are aware of limited number of male participants, which might impact the findings of the current research, especially regarding the findings of gender differences. Thus we acknowledge this issue as one of the limitations of the current research. Regardless of this limitation issue, we still believe that the current research can be considered as the starting point, or maybe as the first research regarding the exploration of environmental beliefs, attitudes, intention, and self-reported behaviour of preservice teachers in Indonesia, given that exploration of the general population is still found to be limited.

Research instruments

Sample of research was chosen based on the objectives of the study: validity, reliability and perception. Four content experts were assigned to evaluate the content validity of the e-content module (three experts) and the e-assessment module (three experts). Since the Chemistry MOOC was developed for the undergraduate chemistry topics, a group of 23 undergraduate students was involved in the pilot study to answer the reliability questionnaire. A total of 129 undergraduate students from the Faculty of Science and Mathematics enrolled in Chemistry courses were randomly chosen as the respondents for perception study. The demographic information of the respondents is shown in Table I.

Survey

As mentioned above in the introduction, to examine the environmental beliefs, attitudes, and self-reported behaviour of Indonesian preservice biology teachers, the Environmental Attitudes Inventory (EAI) developed by Milfont and Duckitt (2010) was used as the main research instrument. The main reason that the EAI was used in the current research instrument is that it was developed by considering the multidimensionality of environmental attitudes and instrument may appropriate to exploring behaviour change theories. Many other instruments do not consider this dimensionality issue (see Milfont & Duckitt, 2010 for more detail), which can lead to misinterpretation of the results. The instrument embraces environmental beliefs, attitudes (affects), intention, and self-reported environmental behaviour. It, therefore, consists of a total of 12 environmental scales, with ten items in every scale. However, only 10 out of the 12 environmental scales were used to answer the research questions. The other two scales were excluded from the current research, because of the low psychometrical properties that the two scales had. After the evaluation to the ten scales through the item response theory (IRT) Rasch model and classical test theory (CTT), and only 97 out of the 100 items from the ten scales met the psychometrical requirements performed through Rasch analysis based on Boone et al. (2014). All the items were on Likert-scales with five choices, of 1 = strongly disagree to 5 = strongly agree. Table 1 shows the psychometrical properties of the ten scales. The Table shows that the ten scales were also categorized into four environmental domains, based on every scale’s construct definition, as provided in the original Milfont and Duckitt (2010) paper. Those four domains are environmental beliefs (EB), environmental attitudes (EA), environmental intention (EI), and self-reported environmental behaviour (SEB).

The EB domain consisted of four EAI scales,namely environmental fragility (EVF), which measures participants’ belief in the fragile and easily-damaged characteristic of the environment through human activities; confidence in science and technology (CST), which measures participants’ belief in human intellectuality in solving environmental problems, such as through science and technology; and anti-human utilization of and dominance over nature (aHUD) which measures beliefs on the equity between human and nature, and so protecting the environment should be first, rather than prioritizing economic development. aHUD originally was consisted of two different scales, human dominance over nature and human utilization of nature, but in the current research, both have been combined into one-joint construct, given that both scales have similarity in the construct definition that connects the relationship between human rights over nature. Additionally, we took the negative statements side, and thus we put ‘anti’ on the scale’s name. The EA domain was also composed of four EAI scales, without any joint-construct. The four scales are enjoyment of nature (EJN), which measures participants’ perception of spending time in nature and open-green areas; support for interventionist conservation policies (SIC), which measures participants’ perceptions of support for policies related to conservation and regulation of the use of raw materials; ecocentric concern (ECO), which measures participants’ emotional states towards environmental degradation and loss; and support for population growth policies (SPG), which measures participants’ perceptions and concerns of support for policies related to overpopulation. The EI and SEB domains consisted of one EAI scale for each. The EI consisted of environmental movement activism (MOV), which measures participants’ readiness and intention to be involved in the action of protecting nature and the environment. Moreover, the SEB consisted of personal conservation behaviour, which measures participants’ self-reported conservation behaviour.

Table I.Psychometric properties of the used Environmental Attitudes Inventory (EAI) scales

Domain	EAI Scale		Total Item	Measure	Infit MNSQ	Outfit MNSQ	Item Reliability	Person Reliability	α if item deleted	Cronbach's α
Environmental Beliefs (EB)	Environmental fragility	EVF	10	-1.02 ~ 0.66	0.73 ~ 1.22	0.81 ~ 1.27	0.97	0.77	.685 ~ .746	.734
	Anti-Human utilization of and domination over nature	aHUD	20	-0.97 ~ 1.32	0.69 ~ 1.69	0.68 ~ 1.69	0.99	0.77	.748 ~ .798	.777
	Confidence in science and technology	CST	9	-1.19 ~ 2.41	0.50 ~ 1.40	0.50 ~ 1.40	0.99	0.69	.632 ~ .697	.690
Environmental Attitudes (EA)	Enjoyment of nature	EJN	10	-1.33 ~ 1.22	0.74 ~ 1.35	0.77 ~ 1.31	0.98	0.80	.798 ~ .818	.825
	Support for interventionist conservation policies	SIC	10	-0.86 ~ 0.88	0.71 ~ 1.39	0.73 ~ 1.33	0.98	0.74	.712 ~ .745	.750
	Ecocentric concern	ECO	9	-0.73 ~ 0.64	0.80 ~ 1.48	0.81 ~ 1.25	0.97	0.74	.737 ~ .766	.773
	Support for population growth policies	SPG	10	-0.89 ~ 1.00	0.67 ~ 1.37	0.64 ~ 1.28	0.98	0.75	.715 ~ .767	.758
Environmental Intention (EI)	Environmental movement activism	MOV	10	-0.38 ~ 0.61	0.77 ~ 1.39	0.74 ~ 1.37	0.86	0.82	.797 ~ .816	.821
Self-reported Environmental Behavior (SEB)	Personal conservation behavior	PCB	9	-0.94 ~ 0.49	0.63 ~ 1.26	0.58 ~ 1.24	0.97	0.64	.598 ~ 679	.675

As mentioned above, only three out of the 100 items from the 10 EAI scales were excluded from the analysis, because these did not meet the IRT-Rasch cut-off. These excluded items were CST_3, ECO_3, and PCB_1. However, as is shown in Table 1, we still included some items that did not meet the cut-off (the Linacre (2012) and Boone et al. (2014) cut-off MNSQ is from (0.50 – 1.50). Those items were aHUD_1 (outfit MNSQ = 1.69) and aHUD_10 (outfit MNSQ = 1.53). We believe that because those items were included in the joint-construct, the items became misfitting items, and the MNSQ vales were still considered not too high from the cut-off. Therefore, we still kept the items in the further analysis.

Semi-structured interview

Following the survey, a semi-structured interview was conducted. A total of six participants, two (one for each gender) from every academic year (second, third, and fourth years), participated in the interviews. These participants were randomly selected based on their academic year and gender. The interview process was conducted to gather more in-depth data on what participants think about engaging in the pro-environmentally behaviour. Thus several questions, such as “how do you react to the phenomena of environmental damage?”, “how do you explain and connect your own self to the emergent phenomena of environmental damage?”, and “how do you explain the intention to behave and the behaviour itself in the context of the environment?” were used as prompting questions to uncover their ideas about pro-environmental behaviour. The interview took between about fifteen and twenty-five minutes for each respondent.

Data analyses

To respond to the research questions, independent sample t-test was used to determine gender differences, and one-way analysis of variance (ANOVA) test was used to uncover educational year differences. Following the analyses, the Pearson’s correlation and partial correlation were performed, to find out how one domain correlates with other domains without and with any control variables. Addressing the main research question, evaluating the hypothesized model was done through structural equation modeling and was performed with AMOS software. The indexes suggested by Schumacker and Lomax (1996) and Yu (2002) were used to evaluate the model. Based on these indexes, a model with a p-value of chi-square (χ2) > .01, AGFI > .90, NFI > .90, CFI > .95, TLI > .95, and RMSEA < .06 or .08 is considered a good model. The interview data were analysed thematically.

Participants

Research instruments

Survey

Semi-structured interview

Data analyses