Multilevel Effects of Student and School Factors on Senior High School Students’ Ocean Literacy

Abstract

This study explored the variance in ocean literacy accounted for by student and school levels and examined the influence of these two predictors on senior high school students’ ocean literacy using a hierarchical linear model. Data were collected from 1944 students from 99 schools and used to construct the two-level hierarchical linear model. The results indicated that the variance in ocean literacy accounted for by students was larger than that accounted for by schools; approximately a quarter of the total variance in ocean literacy was accounted for by schools. At the student level, attitude toward the ocean and frequency of reading ocean-themed books or magazines were predictors of ocean literacy, whereas at the school level, school region and location were significant influential factors. This study’s results have significance for policy-making regarding ocean literacy improvement.

1. Introduction

Marine environmental protection is a very important sustainability issue for the entire world. Relatedly, enhancing ocean literacy (OL) is a very important task for ensuring the sustainability of marine environmental resources. The National Oceanic and Atmospheric Administration (NOAA) explicitly defines OL according to marine science as “an understanding of the ocean’s influence on you and your influence on the ocean” [1,2]. Kean, Posnanski, Wisniewski, and Lundberg [3] also indicate that people with OL can understand the basic origins and fundamental concepts of the ocean, disseminate knowledge regarding the ocean through meaningful approaches, and make precise conclusions regarding the ocean and its resources [2,3].

Enhancing the OL of students or the public can also promote the “Life below water” objective of the UN’s sustainable development goals (SDGs) for international development [4]. Before upgrading OL, we must firstly understand which factors affect students’ OL. The present study determined the influences of students and schools on the ocean literacy of Taiwanese senior high school students. Student and school data are nested, which means they are clustered or embedded together. When an individual student is selected as an analysis sample, this student is considered first-level data and the school or classroom group is considered second-level data. In this type of data structure, the OL performance of a student is profoundly influenced by the school environment, such as school location, school region (coastal or noncoastal), and school size, as well as personal factors, such as the student’s gender, reading habits, attitude toward the ocean (ATO), and their family’s socioeconomic status. Therefore, student characteristics and school factors could influence the learning process and OL performance of a student.

Taiwan is an island nation, and thus, the lives of its people and national development are closely linked to the ocean. However, investigations into whether students or the public possess sufficient marine knowledge or OL are relatively scarce, which suggests that research on marine knowledge or OL is of relatively low importance in Taiwan. In addition, marine education has not received enough attention compared with other subjects, which may be the result of it not being a mainstream subject, such as math and languages. Nevertheless, understanding the marine knowledge or OL of students, focusing on the importance of the ocean’s ecological environment, and equipping students and the public with basic OL are critical educational concerns.

In Taiwan, two studies [5,6] have investigated the influence of students’ personal characteristics on their OL performance. In addition, other studies [6,7] have demonstrated the influence of school factors on students’ OL. However, few studies have conducted research to explore the relationship between OL and students’ personal level and school level; even fewer studies have targeted Taiwanese senior high school students. Moreover, early literature on OL did not clearly reveal how the total variance was simultaneously accounted for by student and school levels. In this study, therefore, a two-level hierarchical linear model was utilized in order to concurrently evaluate the impacts of both the student and school levels on the OL of Taiwanese senior high school students.

The various factors that influence the OL performance of students exist in a multilevel relationship, and while previous studies have determined that both school-related and student-related factors have effects on OL, these two types of influential factor occurred simultaneously and could not be separated. If researchers separate the influence of school factors, whether the variance in senior high school students’ OL is accounted for more by school or student factors cannot be determined, and thus, this topic must be addressed.

1.1. Relationship between Student-Level Predictors and OL

Theoretical and empirical evidence [5,6,7] has indicated that several student-level factors affect students’ OL, such as students’ gender, parental education level, ATO, and frequency of reading ocean-themed books and magazines.

Numerous studies [8,9,10] have revealed that students’ attitudes positively and substantially influence their learning outcomes. In addition, positive learning attitudes and learning motivations are exceptionally helpful for students’ learning and can effectively improve learning outcomes. Greely used students’ ATO as a mediation factor to determine their influence on students’ OL [8]. The results revealed that content knowledge and environmental attitudes were remarkably influential on OL performance. Fortner and Mayer conducted research into the relationship between students’ marine knowledge and ATO and Great Lakes, revealing an exceptional correlation [9]. Moreover, students who had a more positive ATO scored higher on marine knowledge. In Taiwan, Tsai et al. applied structural equation modeling to analyze the correlation of the OL and attitudes of students aged between 16 and 18 years [10]. The study revealed that students with more positive ATO scored higher on OL, which indicated that ATO has a powerful and positive influence on students’ OL.

The relationship between socioeconomic factors and OL has not been extensively discussed in the literature. Furthermore, parental education level is usually considered one of the most stable socioeconomic factors because it establishes its influence on children at a young age, which then remains consistent over time [11,12]. Kalender and Berberoglu regarded parental education level as a critical indicator of socioeconomic status and determined that it was a substantial variance in student learning outcomes [13]. Numerous studies have emphasized the relationship of parental education level with the scientific achievements of students [13,14,15].

In addition, the present study identified students’ gender and frequency of reading ocean-themed books or magazines as crucial factors. Relevant studies have discussed the difference between male and female students in terms of OL [5,7,16]. Chang proved that the OL scores of male and female Taiwanese junior high school students were not notably different [5]. However, Lwo et al. revealed that female students scored remarkably higher in oceanography than male students [7]. Greely conducted research on an oceanography camp exclusively for 30 teenage girls and found that marine exploration and experiences notably improved the girls’ marine knowledge and ATO [8]. Zelezny, Chua, and Aldrich surveyed 1293 teenagers and revealed gender differences in various aspects of environmental literacy as well as diverse points of view on environmental topics [17]. Chang suggested that students who frequently read ocean-themed books or magazines outperformed their peers who did so less frequently [5]. Furthermore, Steel et al. found that newspapers and the Internet are likely to improve citizens’ knowledge about the ocean, whereas television and radio broadcasts negatively affect marine knowledge improvement [16].

1.2. Relationship between School-Level Predictors and Achievement

Several studies have revealed that school-level predictors, such as location, had a critical effect on students’ learning outcomes [5,16]. In the United States (US), Steel et al. conducted research on 1233 US citizens regarding OL knowledge and revealed that people who lived in coastal states generally believed they had superior marine and coastal knowledge compared with those who lived in noncoastal states [16]. Moreover, their research results revealed that people who lived in coastal states had higher accuracy rates on OL tests than did those who lived in noncoastal states. In Taiwan, Chang defined coastal and noncoastal schools based on whether the administrative area of the school is located near the ocean [5]. The results of Chang’s study showed that students who attended coastal schools scored significantly higher than those who did not.

In addition to whether the school is near the ocean, its scale and distance to a city are some other factors that affect students’ learning outcomes. Kyei and Nemaorani revealed a negative correlation between school location and students’ math scores; specifically, students’ performance at city schools was lower than that of suburban schools [18]. Lamb and Fullarton revealed that students in rural areas scored higher in math than did those in the city [19]. Beck and Shoffstall revealed superior performance of students in rural areas compared with that of students from city and suburban areas [20]. The Nippon Foundation Ocean Policy Research Foundation conducted a census on marine education in 6707 Japanese high schools and elementary schools, and revealed a remarkably high correlation between school location and the cognition level of marine education [21]. In other words, the closer the school was to the sea, the higher the students’ cognition level of marine education was. In addition, schools’ location and implementation of marine education were exceptionally correlated; the closer the school was to the sea, the more marine education-related courses it implemented (both through textbooks and extracurricular activities). However, no relevant studies have determined the relationship between school location and students’ OL.

1.3. Research Purposes

The objective of the present study was to use hierarchical linear model to determine the influence of the personal characteristics of Taiwanese senior high school students and school-level predictors on their OL performance. Hierarchical linear modeling is the most appropriate statistical method for analyzing data that are organized in more than one level [11,22]. Thus, this study employed hierarchical linear model to determine factors influencing OL from the perspectives of student and school levels. The three research questions were as follows:

(1)

What degree of variance in OL are the selected school-level and student-level factors responsible for?

(2)

Which student-level factors (e.g., gender, frequency of reading ocean-themed books or magazines, and parental education level) are significantly related to OL?

(3)

Which school-level factors (e.g., region, size, location, and frequency of marine education activities) are significantly related to OL?

2. Materials and Methods

2.1. Measurement Instruments

The scale established in this study contained 40 general OL knowledge question items. Its structure was based on the substantive connotations of marine education topics in the guidelines for 12-year compulsory education and comprised five learning topics, namely marine leisure, marine society, marine culture, marine science and technology, and marine resources and sustainability. A total of 19 different fundamental concepts are used to illustrate and explicate these topics, with each topic being delineated through the presentation of a series of those basic concepts (

Table 1. Learning topics and content of marine education.

Learning Topics	Content
	Senior High School
Marine leisure	1. Being familiar with water sports and possessing the knowledge of safety. 2. Planning and participating in various water recreation and tourist activities. 3. Understanding fishing villages and coastal landscapes and the relationship between cultural customs and ecotourism.
Marine society	1. Analyzing marine industries and technological development and evaluating the relationship between marine industries and economic activities. 2. Understanding and paying attention to ocean-related laws and policies. 3. Analyzing the evolution of the marine history of Taiwan and other countries and evaluating their similarity and difference. 4. Understanding Taiwan’s maritime rights and interests and strategic position.
Marine culture	1. Making good use of various writing styles and skills to create literary works with the ocean as the background. 2. Recognizing the value, style, and cultural context of various marine arts. 3. Comparing the evolution of ocean-related folk beliefs in Taiwan and that in other countries.
Marine Science and Technology	1. Understanding the physical and chemical properties of the ocean. 2. Understanding the influence of ocean structure, seabed topography, and ocean currents on the marine environment. 3. Investigating the relationship between marine environmental changes and climate change. 4. Understanding global hydrosphere relationships between ecosystems and biodiversity. 5. Being familiar with ocean-related applied science, such as seawater desalination, shipping, tidal power generation, and mineral extraction.
Marine resources and sustainability	1. Examining the management strategies and sustainable development of marine biological resources. 2. Understanding marine resources such as minerals and energy and their economic values. 3. Recognizing waste accumulation in marine life and the environment caused by marine environment pollution as well as proposing coping strategies. 4. Understanding problems of the hometown marine environment and actively participating in ocean preservation activity.

). A committee comprised of 15 members developed the questionnaire, with the members of the committee consisting of marine scientists, senior high school teachers, and both formal and informal marine educators. Moreover, the aforementioned five issues and the associated 19 basic concepts were the primary basis for the question items developed for the questionnaire. On this basis, the committee discussed and determined which concepts could be assessed by written tests before designing the test accordingly. After the first stage of development, three test evaluation experts were asked to review and revise the questionnaire to fit the principles of multiple-choice questions. Subsequently, five scholars and experts conducted further discussions and revisions on each revised item with respect to its content to reach a consensus and ensure that the items fit the basic concepts of OL. All items are multiple-choice questions that have one correct answer out of four choices. Each correct answer counts for one point and each wrong answer counts for zero points; the total score is 40.

Table 2. Number of items in the ocean literacy (OL) questionnaire that feature the five factors.

Essential Principle	No. of Items	Instrument
Marine leisure	8	1–8
Marine society	10	9–16, 37–38
Marine culture	8	17–24
Marine science and technology	8	25–32
Marine resources and sustainability	6	33–36, 39–40
Total	40

presents the number of items in the OL questionnaire that feature the five factors and their corresponding numbers. The aforementioned procedure indicated that the development of this questionnaire had content validity.

Before the official test, a pretest was conducted and its results were employed to revise the items once more to establish the reliability and validity of the questionnaire. A total of 466 senior high school students participated in the pretest. Both the Cronbach’s $\mathsf{\alpha }$ coefficient and construct reliability of the pretest data analysis were 0.91, meaning that both values were above the respective threshold values of 0.8 and 0.60 advocated by Fornell and Larcker [23]. In other words, the scale had optimal internal consistency. Additionally, the average variance extracted analysis result was 0.53, which was higher than the 0.5 recommended by Bagozzi and Yi [24]. In short, the analytical results showed that, overall, the questionnaire developed for the study had acceptable reliability and convergent validity.

In addition, confirmatory factor analysis was applied to test the construct validity of the questionnaire. The questionnaire was analyzed under the assumption of a one-factor model. The hypothesized model was determined and was found to fit the data adequately, with the specific goodness-of-fit indices being as follows: confirmatory fit index (CFI) = 0.92; root mean square error of approximation (RMSEA) = 0.02; and standardized root mean square residual (SRMR) = 0.09. A CFI value higher than 0.90 indicates acceptable goodness of fit [25]. With respect to RMSEA values, a value higher than 0.1 indicates poor fit, a value above 0.08 but below or equal to 0.1 indicates a mediocre fit, a value above 0.05 but below 0.08 indicates a fair fit, and a value below or equal to 0.05 indicates a good fit [26]. An SRMR value lower than 0.05 indicates a good fit, whereas one lower than 0.1 is a fair fit [25]. The results indicated that the goodness-of-fit indices were all acceptable, which also indicated that the questionnaire was unidimensional and exhibited construct validity. The pretest results were analyzed using the Rasch model of item response theory; the overall difficulty was between −3.157 and 1.564 and the questionnaire discrimination was between 0.097 and 0.507. In other words, the difficulty level of this questionnaire was medium to easy, which fit the original setting for questionnaire development.

In addition to the OL questionnaire, this study surveyed student background variances (i.e., gender and ATO) and school background variances (i.e., school size, school location, and marine education activities). The items for ATO were modified from the Trends in International Mathematics and Science Study (TIMSS) international survey [27]. The TIMSS international survey used three items (I enjoy learning science; science is boring; and I like science) to define science attitudes and explore the sources of variability in science performance. Similarly, the current study obtained attitudes regarding marine science by asking students to indicate agreement on the three revised items. The three revised statements were as follows: I enjoy learning marine knowledge; marine science is boring; and I like marine science. Each item is based on a four-point Likert scale (1 = strongly disagree; 2 = disagree slightly; 3 = slightly agree; and 4 = strongly agree). The second item was scored reversely. Higher scores indicated students had more positive attitudes toward marine science. In the Taiwan Social Change Survey [28], school location was categorized into six groups: 6 = core urban; 5 = general urban; 4 = emerging cities and townships; 3 = traditional industrial sector cities and townships; 2 = general townships and villages; and 1 = rural areas).

Table 3. Coding of all used variables.

Variables Name	Variable Description
Student-Level Factors
Gender	1 = girls; 2 = boys
ATO	Students were scored according to their degree of agreement with the following three statements. (1) I enjoy learning marine knowledge; (2) marine science is boring; (3) I like marine science 1 = strongly disagree; 2 = disagree; 3 = neither agree nor disagree; 4 = agree; 5 = strongly agree; The total score is 15.
PEL	1 = Junior high school or below; 2 = Senior high school; 3 = university degree; 4 = Graduate school or above.
ROBM	1 = never; 2 = rarely; 3 = sometimes; 4 = often; 5 = always.
School-Level Factors
School region	0 = coastal; 1 = noncoastal
Marine activities	The total marine activates per semester in school.
School size	1 = 6 classes or fewer; 2 = 7 to 48 classes; 3 = above 49 classes
School location	6 = core urban; 5 = general urban; 4 = emerging cities and townships; 3 = traditional industrial sector cities and townships; 2 = general townships and villages; 1 = rural areas

Note: PEL = parental education level; ATO = attitude toward ocean; ROBM = frequency of reading ocean-themed books or magazines.

presents the coding of all the variables used.

2.2. Study Procedure

To determine the student- and school-level factors that influenced OL, standardization of the research procedure was required. The questionnaire in this study employed the paper-and-pencil test method and targeted Taiwanese senior high school students as respondents, who had to complete the questionnaire and background variance survey within 50 minutes. Additionally, school principals were asked to help with answering the survey for the school (e.g., the number of marine education activities the school organized within the most recent year). The participating students first indicated their agreement to take part in the study, after which they then answered the questionnaire anonymously. The development of this questionnaire, including expert revisions, was conducted from March to December 2016. The pretest was set in May and June 2017. Pretest analysis was conducted immediately after and the results were used to revise the questionnaire items once more to finalize the official questionnaire. For the testing conducted with the official questionnaire, the examiners, all of whom were high school teachers who had undergone training in how to conduct standardized testing, were sent to the participating schools. In order to ensure the highest level of accuracy, this testing with the official questionnaire was conducted in May and June of 2018. The testing was conducted throughout four regions of Taiwan, specifically, the Northern, Southern, Eastern, and Central regions, with a stratified random sampling approach being used to obtain the samples. Following the testing, the survey results were first double-checked to confirm that the data provided were correct and then archived.

2.3. Participants

Taiwanese senior high school students were recruited to participate in the formal test. First, stratified random sampling was applied to the Northern, Central, Southern, and Eastern regions of Taiwan, recruiting a total of 99 schools and 2038 students to participate. Invalid responses and missing data were eliminated to obtain a total of 1994 valid samples, among which 1021 were from male students (51.2%) and 973 were from female students (48.8%). In addition, 600 students were from the Northern region (30.1%), 618 were from the Central region (31.0%), 504 were from the Southern region (25.3%), and 272 were from the Eastern region (13.6%).

2.4. Data Analysis

Observational data regarding educational situations usually exist in the form of nested data structures, with students being nested within schools or within classrooms. Therefore, neglecting the influence of factors from different structural levels on observations may cause mistakes in interpreting the statistical analysis and research results [22]. To mitigate the influence of multilevel data, hierarchical linear model must be employed to analyze data in a multilevel structure [22].

Regarding the data analysis in this study, descriptive statistics was applied to the surveyed data. Moreover, a two-level hierarchical linear model data analysis method, in which the students effectively provided the level-1 data and the schools provided the level-2 data, was employed. The hierarchical linear model analysis included four steps. In step one (model 1), the unconditional model (null model) with no predictors was utilized, with the two levels and the OL performance of the students being utilized as the dependent variables. This first step/model yielded a measure of the proportions of the variance in OL within and between the schools. In step two (model 2), the significance of the correlation between the OL and the variance was verified by inputting level-1 predictors into model 1. In the third step, relevant variances of level-2 data were introduced to model 2 to verify the influence of these variances on OL. In the final step, step four (the full model), all of the two-level variables were included in the model. To avoid bias in inferring the population characteristics, a multiple imputation procedure was employed when data were missing.

3. Results

Table 4. Descriptive statistics of the ocean literacy.

Variables Name	N of Case (%)	Mean Score	Standard Deviation
Student level factor
Gender
girl	973 (48.80)	25.05	4.53
boy	1021 (51.20)	24.38	5.74
Frequency of reading ocean books or magazines (ROBM)
never	199 (9.98)	21.97	6.49
rarely	1094 (54.86)	24.97	4.85
sometimes	559 (28.03)	24.99	5.01
often	126 (6.32)	25.58	5.02
always	16 (0.80)	24.25	7.46
Parental education level (PEL)
junior high school or below	129 (6.47)	22.74	6.18
senior high school	837 (41.98)	24.14	5.15
university degree	776 (39.07)	25.33	4.82
graduate school or above	252 (12.64)	25.68	5.31
School level factor
School Region
coastal	721 (36.16)	23.81	5.71
noncoastal	1273 (63.84)	25.21	4.81
School location
core urban region	122 (6.12)	26.40	4.57
general urban region	620 (31.09)	25.22	4.93
emerging cities and townships	437 (21.92)	25.12	4.89
traditional industrial sector cities and townships	243 (12.19)	24.34	5.26
general townships and villages	469 (23.52)	24.27	5.20
rural areas	103 (5.17)	20.73	6.32

presents the descriptive statistics of the student- and school-level factors and OL. This study recruited 973 female students (48.80%) and 1021 male students (51.20%) and revealed that female students scored higher than their male counterparts in OL. Nearly 64% of the students reported rarely reading ocean-themed books or magazines. Students who do so at a higher frequency had superior performance in OL. Similarly, higher parental education level resulted in greater OL performance. With respect to school-level factors, students in coastal schools had inferior performance in OL than did those in noncoastal schools. In total, 36.16% of participating schools were located in coastal regions. Additionally, students from schools located in more urban areas scored higher in OL with a highest score of 26.40 points, whereas students in schools located in rural areas scored lower in OL with a lowest score of 20.73 points.

3.1. Student-Level Influence

This study established its first model, a null model, in response to the first research purpose. The only dependent variable for this model was OL score. Neither student-level variables nor school level variables were included in this model. In other words, this model was equivalent to a one-way analysis of variance with random effect. This model was used to determine the number of OL variables that originated from students and in schools. Intraclass correlation (ICC) was employed to determine whether the data could be analyzed by hierarchical linear model. The research results are presented in Table 3. The ICC was calculated as follows:

$${\tau }_{00}/\left({\tau }_{00}+{\sigma }^2\right)=\frac{6.86}{6.86+21.18}=0.245$$

(1)

The results revealed that 24.5% of the overall variables in students’ OL originated in the differences between schools, whereas 73.5% of the variables were the result of individual differences between students. The ${\mu }_0$ in the null model was the variance of the mean OL score of each school compared with the overall mean score. The variance obtained was 6.86, which was statistically significant (p < 0.001). This result indicated that the variance significance of the mean OL score of each school and the overall mean OL score was not 0, which meant that the mean OL score of each school had significant variance. Moreover, these results proved that employing hierarchical linear model to analyze influencing factors on student OL was necessary.

The second research question was posed in order to investigate how level-1 student factors influence the OL performance of students. Therefore, on the basis of the null model, level-1 variables, namely gender, parental education level, reading ocean-themed books or magazines, and ATO, were introduced to the model and the results are presented in

Table 5. Effects of students and school-level factors on OL.

	Null Model	Model-1	Model-2	Model-3
Intercept	24.46*** (0.28)	18.82*** (0.92)	28.93*** (3.25)	23.39*** (3.29)
Student level factor
Gender		0.37 (0.25)		0.36 (0.25)
PEL		0.08 (0.12)		0.05 (0.12)
ROBM		0.52*** (0.13)		0.52*** (0.13)
ATO		0.37*** (0.05)		0.36*** (0.05)
School level factor
School region			1.29* (0.59)	1.21* (0.59)
School size			−0.26 (0.56)	−0.11 (0.57)
Marine activities			−0.74 (1.12)	−0.95 (1.14)
School location			−0.65*** (0.19)	−0.63*** (0.18)
Variance components
Intercept 1 (μ0)	6.86	6.86	5.40	5.47
Level-1 (γ)	21.18	20.10	21.17	20.08

Note: *p < 0.05; **p < 0.01; ***p < 0.001. PEL = parental education level; ATO = attitude toward ocean; ROBM = frequency of reading ocean-themed books or magazines.

. The school level variance component in this model was 6.86, which indicated significance (p < 0.001); in other words, significant variances were observed from school to school. Calculations revealed that 25.45% and 74.55% of the variables for student OL were from school- and student-level variables, respectively. In addition, the effects of reading ocean-themed books or magazines and ATO on OL were significant and their fixed effect values were 0.52 and 0.37, respectively. In other words, for each point added to reading ocean-themed books or magazines, the mean OL score increased by 0.52 points. Similarly, for each point added to ATO, the mean OL score increased by 0.37 points. These results indicated that the mean OL score of students who scored higher in ATO was higher than that of students who scored lower in ATO. In contrast, the results did not indicate that gender or parental education level were associated with any significant differences in OL.

3.2. School-Level Influence

The third research question was posed in order to investigate how level-2 school factors influence the OL performance of students. In this stage, the student-level variables introduced in the previous stage, namely gender, parental education level, reading ocean-themed books or magazines, and ATO, were replaced by school-level variables, namely school region, school size, school location, and frequency of marine education activities, to understand the influence of these four variables on students’ OL performance. Analysis results are shown in Table 5. The school-level variance component was reduced to 5.40 in this model but the test results still showed significance (p < 0.001), which means that significant variances were observed from school to school. ICC calculation results indicated that 20.32% and 79.68% of the variables for student OL were from school- and student-level variables, respectively. In addition, the effect of school region and school location on students’ OL was significant with coefficients of 1.29 and −0.65 and standard deviations of 0.59 and 0.19, respectively. The coefficient value of school region was positive, which means that students of noncoastal schools had superior OL performance compared with those of coastal schools, and the mean difference was 1.29 points. The coefficient value of school location was negative, which means that students in rural areas scored lower in OL. In particular, when the school was one unit away from the city center, students’ OL score decreased by 0.65 points.

Model 3, also called the full model, was the final model. All the factors were added to this model simultaneously in order to evaluate the correlations among all the predictors at the school and student levels. Table 3 presents the results, which indicated that student and school levels accounted for 21.41% and 78.59% of the total variance in OL, respectively. The results were consistent with models 1 and 2. At the student level, the frequency of reading ocean-themed books or magazines and ATO had statistical significance with the students’ OL, whereas at the school level, school location and noncoastal schools affected students’ OL.

4. Discussion

In the present study, a two-level hierarchical linear model was utilized to conduct an analysis and evaluation of how both the personal characteristics of students and school-level factors affect the OL of Taiwanese senior high school students. The results revealed that approximately three-quarters of the overall variance were attributed to differences between students and a quarter was attributed to differences between schools. Regarding student-level factors, ATO and reading ocean-themed books or magazines significantly affected students’ OL performance, and regarding school-level factors, school region and school location were substantial predictors of students’ OL.

Among the student-level factors, ATO and reading ocean-themed books or magazines were considerably influential, which supports the results of relevant studies [8,9,10]. Students with a more positive ATO also scored higher in OL. Teacher or parental efforts in nurturing student learning attitudes remarkably improve the learning outcomes of various subjects. Enhancing students’ positive ATO eliminates any fear of the ocean they might have and increases their OL performance. Therefore, the outcome of nurturing positive learning attitudes in students is exceptional. Furthermore, the results revealed that reading ocean-themed books or magazines is a crucial factor that affects students’ OL; students reading ocean-themed books or magazines frequently outperformed peers who did so less frequently.

Regarding school-level factors, school region and school location were substantial predictors for OL. However, our results were inconsistent with those of other studies in terms of school region [7,16]. Steel et al. revealed that US citizens living in coastal states had more marine and coastal knowledge as well as higher OL scores [16]. Moreover, the results of Chang showed that junior high school students who attended coastal schools scored significantly higher than those who did not [7], whereas the present study revealed that the OL performance of students from coastal senior high schools was inferior to that of students from noncoastal schools. Another factor influencing students’ OL performance is school location. This study’s results revealed that OL performance declined as school location moved toward rural areas. In other words, senior high school students in schools closer to core urban areas scored higher in the OL questionnaire. These results were inconsistent with the Nippon Foundation Ocean Policy Research Foundation study [21]. Considering the abovementioned two indicators, the authors speculated that the geographic environment of Taiwan is closely related to these results. As an island nation, Taiwan is surrounded by the ocean. Therefore, even if a school is not located in a coastal region, students can still reach the coast in two or three hours by public transport or with their parents by car. However, the urban–rural gap has a greater effect, particularly because in Taiwan schools closer to urban areas receive more educational resources than do those located in rural areas. Moreover, coastal regions mostly feature fishing villages, which are more likely to be classified as general townships and villages or rural areas. As a result, even if a school was located in a coastal region, marine education resources for its students are fewer than those for students in schools in core urban areas. Therefore, marine knowledge acquisition and students’ OL performance were affected by school location rather than school region. The OL survey results obtained were consistent with relevant studies on student learning outcomes [29,30], which indicated that Taiwanese students’ performance diminished from cities, towns, villages, to remote villages.

The questionnaire used in this study was developed through a rigorous process that ensured its compliance with various psychometric criteria (e.g., reliability and validity). Therefore, the study can help teachers to understand their students’ ocean literacy levels, and to adopt effective measures that cater to the varying levels of ocean literacy among their students. In this manner, teachers will be able to provide the appropriate guidance and assistance, such that the educational objective of teaching according to individual differences is met. The study limited its discussion to student- and school-related factors since it only focused on the effects of these factors on ocean literacy. However, there are in fact a variety of other factors that also influence the performance of students. Many studies have indicated that the teachers who provide instruction, particularly those with teaching experience and relevant majors, can significantly influence the learning outcomes of students. For this reason, future research should simultaneously investigate student-, teacher-, and school-related factors and utilize three-level HLM to analyze the data collected, so as to understand the influence of three-level variables on ocean literacy. The study’s second limitation relates to the use of Taiwan-based marine education topics as the testing framework for ocean literacy, and the collection of data only from Taiwanese students, which meant that it was not possible to compare the ocean literacy performance of students from different countries. However, given that the National Marine Electronics Association (NMEA) has already published the definition of ocean literacy and the ocean literacy principles for grades K–12, future research could compare the content differences between Taiwan’s marine education curriculum and the NMEA’s ocean literacy principles, and seek to develop assessment tools for both areas that are similar and different. This will facilitate future cross-border comparisons of students’ ocean literacy. The third research limitation has to do with the fact that this study was a cross-sectional study. Going forward, longitudinal studies could be conducted to explore the long-term development of ocean literacy among students. This will not only enable us to further understand the development of ocean literacy among senior high students, but also allow us to form a more accurate picture of the development process that these students undergo with respect to ocean literacy. Furthermore, such efforts will also contribute to the generation of normative developmental data regarding the ocean literacy of students, which will be useful to front-line educators.

5. Conclusions

The main finding of this study is that for senior high school students in Taiwan, individual factors are the primary drivers of OL. In other words, student-level factors accounted for a larger share of the total variance in the OL of students than did school-level factors. However, the school-level factors affecting students’ OL performance cannot be ignored. In addition, we found that at the student level, ATO and frequency of reading ocean-themed books or magazines were predictors of OL, whereas at the school level, school region and school location were the crucial influencing factors for OL. Marine education has always been a very important issue in sustainable environmental education. The ocean-literacy-construct can be a part of sustainable education. It also could embed the ocean-literacy-construct in the framework model of sustainability competencies developed by Waltner et al. [31].