news article about lack of education in the philippines

• High contrast
• OUR REPRESENTATIVE
• WORK FOR UNICEF
• NATIONAL AMBASSADORS
• PRESS CENTRE

Search UNICEF

Filipino children continue missing education opportunities in another year of school closure, together with four other countries, the philippines has kept its schools closed nationwide since the covid-19 pandemic.

MANILA,25 August 2021-- A child's first day of school—a landmark moment for the youngest students and their parents around the world—has been delayed due to COVID-19 for an estimated 140 million young minds, UNICEF said in a new analysis released as summer break comes to end in many parts of the world.

The Philippines is one of the five countries in the world that have not started in-person classes since the pandemic began, affecting the right to learn of more than 27 million Filipino students. While new variants are causing a rise of infections, UNICEF is advocating for a phased reopening of schools, beginning in low-risk areas. This can be done on a voluntary basis with proper safety protocols in place.

"The first day of school is a landmark moment in a child's life—setting them off on a life-changing path of personal learning and growth. Most of us can remember countless minor details—what clothes we wore, our teacher's name, who we sat next to. But for millions of children, that important day has been indefinitely postponed," said UNICEF Executive Director Henrietta Fore. "As classes resume in many parts of the world, millions of first graders have been waiting to see the inside of a classroom for over a year. Millions more may not see one at all this school term. For the most vulnerable, their risk of never stepping into a classroom in their lifetime is skyrocketing."

For an estimated eight million students around the globe—who should have been in the first grade— the wait for their first day of in-person learning has been over a year and counting, as they live in places where schools have been closed throughout the pandemic.

The first grade sets up the building blocks for all future learning, with introductions to reading, writing, and math. It's also a period when in-person learning helps children gain independence, adapt to new routines, and develop meaningful relationships with teachers and students. In-person learning also enables teachers to identify and address learning delays, mental health issues, and abuse that could negatively affect children’s well-being.

“In 2020, schools globally were fully closed for an average of 79 teaching days, while the Philippines has been closed for more than a year, forcing students to enroll in distance learning modalities.  The associated consequences of school closures – learning loss, mental distress, missed vaccinations, and heightened risk of drop out, child labour, and child marriage – will be felt by many children, especially the youngest learners in critical development stages,” UNICEF Philippines Representative Oyunsaikhan Dendevnorov says.

While countries worldwide are taking some actions to provide remote learning, at least 29 per cent of primary students are not being reached. In addition to lack of assets for remote learning, the youngest children may not be able to participate due to a lack of support using the technology, a poor learning environment, pressure to do household chores, or being forced to work.

Studies have shown that positive school experiences during this transition period are a predictor of children’s future social, emotional and educational outcomes. At the same time, children who fall behind in learning during the early years often stay behind for the remaining time they spend in school, and the gap widens over the years. The number of years of education a child receives also directly affects their future earnings.

Unless mitigation measures are implemented, the World Bank estimates a loss of $10 trillion in earnings over time for this entire generation of students. Existing evidence shows the cost of addressing learning gaps are lower and more effective when they are tackled earlier, and that investments in education support economic recovery, growth and prosperity.

UNICEF urges governments to reopen schools for in-person learning as soon as possible, and to provide a comprehensive recovery response for students. Together with the World Bank and UNESCO, UNICEF is calling for governments to focus on three key priorities for recovery in schools:

• Targeted programmes to bring all children and youth back in school where they can access tailored services to meet their learning, health, psychosocial well-being, and other needs;
• Effective remedial learning to help students catch up on lost learning;
• Support for teachers to address learning losses and incorporate digital technology into their teaching.

"Your first day of school is a day of hope and possibility—a day for getting off to a good start. But not all children are getting off to a good start. Some children are not even starting at all," said Fore." We must reopen schools for in-person learning as soon as possible, and we must immediately address the gaps in learning this pandemic has already created. Unless we do, some children may never catch up."

In the following weeks, UNICEF will continue to mobilize its partners and the public to prevent this education crisis from becoming an education catastrophe. Online and offline campaigns will rally world leaders, teachers, and parents around a common cause: reopen schools for in-person learning as soon as possible. The future of the world’s most vulnerable children is at stake.

Media contacts

About unicef.

UNICEF promotes the rights and wellbeing of every child, in everything we do. Together with our partners, we work in 190 countries and territories to translate that commitment into practical action, focusing special effort on reaching the most vulnerable and excluded children, to the benefit of all children, everywhere.

For more information about UNICEF and its work for children in the Philippines, visit www.unicef.ph .

Follow UNICEF Philippines on Facebook ,  Twitter and Instagram .

Related topics

More to explore, upholding the rights of every bangsamoro child.

UNICEF Philippines Chief of Advocacy and Communication Lely Djuhari meets children and families in Bangsamoro region

How daycare workers strengthen the Philippines’ human capita

UNICEF and partners empower child development workers through skills training

Fathers, we play a critical role in our child's first years of learning

Investing in day care workers is an investment in the future -UNICEF

• International edition
• Australia edition
• Europe edition

Crisis in Philippines as millions of children face second year of remote schooling

Frustration mounts as Rodrigo Duterte rejects a pilot reopening of schools for fear children could pass coronavirus to elderly relatives

• See all our coronavirus coverage

Classrooms in the Philippines were silent on Monday as millions of schoolchildren hunkered down at home for a second year of remote lessons that experts fear will worsen an educational “crisis”.

While nearly every country in the world has partially or fully reopened schools to in-person classes, the Philippines has kept them closed since the start of the coronavirus pandemic, the UN says.

President Rodrigo Duterte has so far rejected proposals for a pilot reopening of primary and secondary schools for fear children could catch Covid-19 and infect elderly relatives.

“I want to go to school,” seven-year-old Kylie Larrobis said, complaining she cannot read after a year of online kindergarten in the tiny slum apartment in Manila she shares with six people.

“I don’t know what a classroom looks like – I’ve never seen one.”

Larrobis, who enters first grade this year, cries in frustration when she cannot understand her online lessons, which she follows on a smartphone, said her mother, Jessielyn Genel.

Her misery is compounded by a ban on children playing outdoors.

“What is happening is not good,” said Genel, who opposed a return to in-person classes while the Delta variant spread through the country.

A “blended learning” programme involving online classes, printed materials and lessons broadcast on television and social media was launched in October last year.

It has been plagued with problems: most students in the Philippines don’t have a computer or internet at home.

More than 80% of parents are worried their children “are learning less”, said Isy Faingold, Unicef’s education chief in the Philippines, citing a recent survey.

Around two-thirds of parents support the reopening of classrooms in areas where virus transmission is low. “Distance learning cannot replace the in-person learning,” Faingold said. “There was already a learning crisis before Covid … it’s going to be even worse.”

Fifteen-year-olds in the Philippines were at or near the bottom in reading, mathematics and science, according to OECD data.

Most students attend public schools where large class sizes, outdated teaching methods, poverty, and lack of investment in basic infrastructure such as toilets have been blamed for youngsters lagging behind.

School enrolments fell to 26.9 million in September 2020 and have dropped a further 5 million since, according to official figures.

Faingold fears many students may “never return”. “We hope in the next days the enrolments continue to accelerate,” he said.

Remote learning is also taking a toll on children’s mental health and development.

Rhodora Concepcion of the Philippine Society for Child and Adolescent Psychiatry said: “Long-term social isolation is closely related to loneliness and physiological illness in children.”

“With the disruption of face-to-face learning and social interaction, regression in formerly mastered skills may be observed in children.”

‘Reading skills really deteriorated’

Petronilo Pacayra is worried about his sons, aged nine and 10. Like most children in the Philippines, they rely on the printed worksheets supplied by their school. “Their reading skills really deteriorated,” the 64-year-old single parent said in the cramped and dimly lit room they share.

Pacayra helps them with their school work in between doing odd jobs to make ends meet.

His youngest child, nicknamed RJ, who is starting second grade, said: “I don’t like reading, I prefer to play with my mobile phone.”

Their school principal, Josefina Almarez, claimed “no children were left behind” in the first year of remote learning. But she admitted some “need special attention”.

Younger children were especially affected by school closures, said Faingold, describing the early years of schooling as “foundational”.

“If you don’t have a strong basis in numeracy and literacy it’s going to be very difficult to learn the other subjects that are part of the primary, secondary or even tertiary education,” he said.

Mercedes Arzadon, a professor of education at the University of the Philippines said it was “ridiculous” to keep schools shut indefinitely when other countries, including virus-ravaged Indonesia, had shown it was possible to reopen them safely. “Our youth’s future and wellbeing are at stake, and so is national development,” Arzadon said in a statement.

An “optimistic scenario” was for schools to reopen next year, said Faingold. But that could depend on the pace of Covid-19 vaccinations, with only about 20% of the targeted population so far fully inoculated.

Children have not yet been included in the programme.

Jessy Cabungcal, whose seven-year-old daughter is enrolled in a Manila private school and uses an iPad and desktop computer for online learning, agrees with Duterte’s decision to keep classrooms shut.

She said: “You could see he is afraid because he cannot assure us that the children will not catch the virus.”

• Philippines
• Coronavirus
• Rodrigo Duterte

Most viewed

UNESCO and DepEd launch the 2020 Global Education Monitoring Report in the Philippines

MANILA, 25 November 2020. Along with government officials, international aid agencies, education and humanitarian experts, policymakers, teachers and learners, the United Nations Educational, Scientific and Cultural Organization (UNESCO) and the Department of Education (DepEd) launched the 2020 Global Education Monitoring (GEM) Report on 25 November 2020 virtually.

With the theme “Inclusion and education: All means All,” the national launch was organized to increase awareness of the Report’s messages and recommendations on inclusion in education with a wider education community, with those working on humanitarian responses, and with government officials and policymakers. The event was broadcasted live on the official Facebook of UNESCO Jakarta and the Philippines’ Department of Education.

As part of its progress towards achieving the Sustainable Development Goal 4 (SDG 4)and its targets, the 2020 GEM Report ( https://unesdoc.unesco.org/ark:/48223/pf0000373721 ) provides an in-depth analysis of key factors in exclusion of learners in education systems worldwide, such as background, identity and ability (i.e. gender, age, location, poverty, disability, ethnicity, indigeneity, language, religion, migration or displacement status, sexual orientation or gender identity expression, incarceration, beliefs and attitudes).

One of the numerous examples highlighted in the report is the gender-responsive basic education policy created by DepEd. The policy calls for an end to discrimination based on gender, sexual orientation, and gender identity by defining ways for education administrators and school leaders such as improving curricula and teacher education programmes with the content on bullying, discrimination, gender, sexuality and human rights.

The Report also identifies the heightening of exclusion during the COVID-19 pandemic, where it has shown that about 40% of low and lower-middle income countries have not supported disadvantaged learners during temporary school shutdown. The event featured speeches and presentations from experts on inclusion from both government and non-governmental organizations, policy makers and practitioners, including a message from UNESCO’s Global Champion of Inclusive Education, Ms Brina Kei Maxino, and performances by the world-renowned and 2009 UNESCO Artist for Peace, the Philippine Madrigal Singers.

The highlight of the event was the live discussion between DepEd Secretary, Professor Emeritus Leonor Magtolis-Briones, and the Director of UNESCO Jakarta, Dr Shahbaz Khan, as they explored the findings of the report and deliberated on issues such as inclusion and education and its implementation; adjustment on the school policies during Covid-19; a horizontal collaboration between government and non-government stakeholders; education budget and spending; grants for students; and, social programs to support education.

Alongside today’s publication, UNESCO GEM Report team has also launched a new website called Profiles Enhancing Education Reviews (PEER) that contains information on laws and policies concerning inclusion in education for every country in the world. According to UNESCO, PEER shows that although many countries still practice education segregation, which reinforces stereotyping, discrimination and alienation, some countries like the Philippines have already crafted education policies strong on inclusiveness that target vulnerable groups.

The 2020 Global Education Monitoring (GEM) Report urges countries to focus on those left behind as schools reopen to foster more resilient and equal societies.

• Global Education Monitoring Report

Related items

• Education for sustainable development
• Country page: Philippines
• UNESCO Office in Jakarta and Regional Bureau for Science
• SDG: SDG 4 - Ensure inclusive and equitable quality education and promote lifelong learning opportunities for all
• See more add

This article is related to the United Nation’s Sustainable Development Goals .

More on this subject

Other recent news

The Current Education Issues in the Philippines — and How Childhope Rises to the Challenge

• August 25, 2021

Even before COVID-19 struck and caused problems for millions of families, the country’s financial status is one of the top factors that add to the growing education issues in the Philippines. Furthermore, more children, youth, and adults can’t get a leg up and are thus left behind due to unfair access to learning.

Moving forward, such issues can lead to worse long-term effects. Now, we’ll delve deep into the current status and how we can take part in social efforts to help fight these key concerns of our country.

Crisis in Philippine Education: How is It Really?

Filipinos from rich households or living in cities and developed towns have more access to private schools. In contrast, less favored groups are more bound to deal with lack of classrooms, teachers, and means to sustain topnotch learning.

A 2018 study found that a sample number of 15-year-old Filipino students ranked last in reading comprehension out of 79 countries . They also ranked 78 th in science and math. One key insight from this study is it implies those tested mostly came from public schools. Hence, the crisis also lies in the fact that a lot of Filipinos can’t read or do simple math.

Indeed, it’s clear that there is a class divide between rich and poor students in the country. Though this is the case, less developed states can focus on learning if it’s covered in their top concerns. However, the Philippines doesn’t invest on topnotch learning as compared to its neighbor countries. In fact, many public schools lack computers and other tools despite the digital age. Further, a shortfall in the number of public school teachers is also one of the top issues in the country due to their being among the lowest-paid state workers. Aside from that, more than 3 million children, youth, and adults remain unenrolled since the school shutdown.

It goes without saying that having this constant crisis has its long-term effects. These include mis- and disinformation, poor decision-making, and other social concerns.

The Education System in the Philippines

Due to COVID-19, education issues in the Philippines have increased and received new challenges that worsened the current state of the country. With the sudden events brought about by the health crisis, distance learning modes via the internet or TV broadcasts were ordered. Further, a blended learning program was launched in October 2020, which involves online classes, printouts, and lessons broadcast on TV and social platforms. Thus, the new learning pathways rely on students and teachers having access to the internet.

This yet brings another issue in the current system. Millions of Filipinos don’t have access to computers and other digital tools at home to make their blended learning worthwhile. Hence, the value of tech in learning affects many students. Parents’ and guardians’ top concerns with this are:

• Money for mobile load
• Lack of gadget
• Poor internet signal
• Students’ struggle to focus and learn online
• Parents’ lack of knowledge of their kids’ lessons

It’s key to note that equipped schools have more chances to use various ways to deal with the new concerns for remote learning. This further shows the contrasts in resources and training for both K-12 and tertiary level both for private and public schools.

One more thing that can happen is that schools may not be able to impart the most basic skills needed. To add, the current status can affect how tertiary education aims to impart the respect for and duty to knowledge and critical outlook. Before, teachers handled 40 to 60 students. With the current online setup, the quality of learning can be compromised if the class reaches 70 to 80 students.

Data on Students that Have Missed School due to COVID-19

Of the world’s student population, 89% or 1.52 billion are the children and youth out of school due to COVID-19 closures. In the Philippines, close to 4 million students were not able to enroll for this school year, as per the DepEd. With this, the number of out-of-school youth (OSY) continues to grow, making it a serious issue needing to be checked to avoid worse problems in the long run.

List of Issues When it Comes to the Philippines’ Education System

For a brief rundown, let’s list the top education issues in the Philippines:

• Quality – The results of the 2014 National Achievement Test (NAT) and the National Career Assessment Examination (NCAE) show that there had been a drop in the status of primary and secondary education.
• Budget – The country remains to have one of the lowest budget allotments to learning among ASEAN countries.
• Cost – There still is a big contrast in learning efforts across various social groups due to the issue of money—having education as a status symbol.
• OSY – The growing rate of OSY becomes daunting due to the adverse effects of COVID-19.
• Mismatch – There is a large sum of people who are jobless or underpaid due to a large mismatch between training and actual jobs.
• Social divide – There is no fair learning access in the country.
• Lack of resources – Large-scale shortfalls in classrooms, teachers, and other tools to sustain sound learning also make up a big issue.

All these add to the big picture of the current system’s growing concerns. Being informed with these is a great first step to know where we can come in and help in our own ways. Before we talk about how you can take part in various efforts to help address these issues, let’s first talk about what quality education is and how we can achieve it.

What Quality Education Means

Now, how do we really define this? For VVOB , it is one that provides all learners with what they need to become economically productive that help lead them to holistic development and sustainable lifestyles. Further, it leads to peaceful and democratic societies and strengthens one’s well-being.

VVOB also lists its 6 dimensions:

• Contextualization and Relevance
• Child-friendly Teaching and Learning
• Sustainability
• Balanced Approach
• Learning Outcomes

Aside from these, it’s also key to set our vision to reach such standards. Read on!

Vision for a Quality Education

Of course, any country would want to build and keep a standard vision for its learning system: one that promotes cultural diversity; is free from bias; offers a safe space and respect for human rights; and forms traits, skills, and talent among others.

With the country’s efforts to address the growing concerns, one key program that is set to come out is the free required education from TESDA with efforts to focus on honing skills, including technical and vocational ones. Also, OSY will be covered in the grants of the CHED.

Students must not take learning for granted. In times of crises and sudden changes, having access to education should be valued. Aside from the fact that it is a main human right, it also impacts the other human rights that we have. Besides, the UN says that when learning systems break, having a sustained state will be far from happening.

How Childhope KalyEskwela Program Deals with Changes

The country rolled out its efforts to help respond to new and sudden changes in learning due to the effects of COVID-19 measures. Here are some of the key ones we can note:

• Continuous learning – Since the future of a state lies on how good the learning system is, the country’s vision for the youth is to adopt new learning paths despite the ongoing threat of COVID-19.
• Action plans – These include boosting the use of special funds to help schools make modules, worksheets, and study guides approved by the DepEd. Also, LGUs and schools can acquire digital tools to help learners as needed.

Now, even with the global health crisis, Childhope Philippines remains true to its cause to help street children:

• Mobile learning – The program provides topnotch access to street children to new learning methods such as non-formal education .
• Access to tools – This is to give out sets of school supplies to help street kids attend and be ready for their remote learning.
• Online learning sessions – These are about Skills for Life, Life Skill Life Goal Planning, Gender Sensitivity, Teenage Pregnancy and Adolescent Reproductive Health.

You may also check out our other programs and projects to see how we help street children fulfill their right to education . You can be a part of these efforts! Read on to know how.

Shed a Light of Hope for Street Children to Reach Their Dreams

Building a system that empowers the youth means helping them reach their full potential. During these times, they need aid from those who can help uphold the rights of the less privileged. These include kids in the streets and their right to attain quality education.

You may hold the power to change lives, one child at a time. Donate or volunteer , and help us help street kids learn and reach their dreams and bring a sense of hope and change toward a bright future. You may also contact us for more details. We’d love to hear from you!

With our aim to reach more people who can help, we’re also in social media! Check out our Facebook page to see latest news on our projects in force.

Subscribe to our Mailing List

©1989-2024 All Rights Reserved.

• Geopolitics
• Environment

Philippines needs to improve its education system

The research arm of Switzerland-based business school, the International Institute for Management Development (IMD) recently released the results of its survey on the talent competitiveness of 63 countries from around the world. Based on the rankings, the Philippines managed to jump up to 49th place from 55th last year. 

Regardless of the jump, the ASEAN country, unfortunately, still performed the worst when compared to other bloc members. In order to understand how this happened, it’s important to look at what the study uses for its indicators.

The World Talent Ranking looks at three main factors when determining how to rank a country. The investment and development factor, which measures resources used to cultivate homegrown human capital;  the appeal factor, which evaluates the extent to which a country attracts and retains foreign and local talent; and the readiness factor, which looks at the quality of skills and competencies of a country’s labour force.

While a simple example of the investment and development factor would be whether a country is able to provide education to its citizens, the readiness factor seems to indicate that it is also important to look at the quality of the education provided. This, according to the study, is where countries like the Philippines fall short.

Quality of education

For the three factors, the Philippines scored 31st for appeal, 61st for investment and development, and 26th for readiness. Similarly, last year, the Philippines scored lowest for investment and development.

In 2018, the IMD World Competitiveness Center’s director, Arturo Bris told the media that the Philippines’ labour force is not as equipped with skills that firms are looking for.

He acknowledged that it was true that the Philippines was making progress in managing its talent pool and is, in fact, one of only two countries in Southeast Asia along with Malaysia which has improved government investment in education as a percent of gross domestic product (GDP).

“However, in 2018, The Philippines witnessed a deterioration of its ability to provide the economy with the skills needed, which points to a mismatch between school curriculums and the demands of companies,” he said.

But it isn’t just a Swiss business school that thinks the Philippines needs to improve its quality of education. In June last year, local media reported the Philippine Business for Education (PBEd) as saying that while the state of education nationwide has progressed in terms of accessibility, it still has a long way to go when it comes to delivery of quality learning for the success of every learner.

PBEd executive director, Love Basillote said this can be attributed to many factors such as prevalence of malnutrition and a shortage of appropriate learning tools, adding that many college graduates are not work-ready due to a lack of socio-emotional skills.

“Our recommendation is we focus on learning by starting early, monitoring learning, raising accountability and aligning actors,” she said, also suggesting that the country participate consistently in international learning assessments to make Filipino learners and graduates globally competitive.

The World Talent Ranking 2018 cited the country’s top weaknesses in the areas of total public expenditure on education, pupil-teacher ratio in primary and secondary schools, and remuneration in service professions and labour force growth.

Upgrading digital skills

The World Economic Forum (WEF) head of Asia Pacific and Member of the Executive Committee, Justin Wood noted that Industry 4.0, also known as the Fourth Industrial Revolution, was unfolding at accelerating speed and changing the skills that workers will need for the jobs of the future.

On 19 November  last year, a coalition of major technology companies pledged to develop digital skills for the ASEAN workforce. Being part of the WEF’s Digital ASEAN initiative, the pledge aims to train some 20 million people in Southeast Asia by 2020, especially those working in small and medium-size enterprises (SMEs).

The move is most welcomed especially due to the threat of huge job displacement across the region. Now, following results from the World Talent Ranking, it seems that this initiative would be much needed in the Philippines as well.

However, the Philippines must understand that the pledge will only go so far in ensuring that it has the right workforce for the new skills demands of companies. Improving the quality of education in the country is still critical and as 2019’s results highlight, the Philippines needs to continue working on this.

This article was first published on 5 December, 2019. 

Related articles:

Critical thinking needed to upgrade skills

Time to invest in human capital

Tangshan And Xuzhou: China's Treatment Of Women

Myanmar's suu kyi: prisoner of generals, philippines ends china talks for scs exploration, ukraine war an ‘alarm for humanity’: china’s xi, china to tout its governance model at brics summit, wrist-worn trackers detect covid before symptoms.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 03 May 2023

Profiling low-proficiency science students in the Philippines using machine learning

• Allan B. I. Bernardo   ORCID: orcid.org/0000-0003-3938-266X 1 ,
• Macario O. Cordel II 1 ,
• Marissa Ortiz Calleja 1 ,
• Jude Michael M. Teves 1 ,
• Sashmir A. Yap 1 &
• Unisse C. Chua 1  

Humanities and Social Sciences Communications volume  10 , Article number:  192 ( 2023 ) Cite this article

25k Accesses

2 Citations

1 Altmetric

Metrics details

Filipino students’ performance in global assessments of science literacy has always been low, and this was confirmed again in the PISA 2018, where Filipino learners’ average science literacy scores ranked second to last among 78 countries. In this study, machine learning approaches were used to analyze PISA data from the student questionnaire to test models that best identify the poorest-performing Filipino students. The goal was to explore factors that could help identify the students who are vulnerable to very low achievement in science and that could indicate possible targets for reform in science education in the Philippines. The random forest classifier model was found to be the most accurate and more precise, and Shapley Additive Explanations indicated 15 variables that were most important in identifying the low-proficiency science students. The variables related to metacognitive awareness of reading strategies, social experiences in school, aspirations and pride about achievements, and family/home factors, include parents’ characteristics and access to ICT with internet connections. The results of the factors highlight the importance of considering personal and contextual factors beyond the typical instructional and curricular factors that are the foci of science education reform in the Philippines, and some implications for programs and policies for science education reform are suggested.

Similar content being viewed by others

Highly accurate protein structure prediction with AlphaFold

Impact of artificial intelligence on human loss in decision making, laziness and safety in education

Sleep quality, duration, and consistency are associated with better academic performance in college students

Introduction.

Global concerns such as the ongoing COVID pandemic and climate change crisis underscore the importance of science and technology for providing sustainable and responsible strategies for global development. Yet in many parts of the world, students’ interest and achievement in science continue to decline (Fensham, 2008 ). The Philippines is one of those countries where students are observed to have low levels of science literacy for many years now (Martin et al., 2004 ; Talisayon et al., 2006 ). This pattern was confirmed when the Philippines participated for the first time in the Program for International Student Assessment (PISA) in 2018, where the results found Filipino 15-year-olds near the bottom of the ranking among 78 countries and territories (Organisation for Economic Cooperation and Development [OECD], 2019 a, 2019 b). Some Philippine studies have tried to understand low science achievement by looking at the curriculum (Belmi and Mangali, 2020 ; Cordon and Polong, 2020 ) and instruction (Sumardani, 2021 ). In this study, we used machine learning approaches to determine the most accurate predictive models that can identify the poorest-performing science students in the PISA 2018 sample. For the variables in the predictive model, we consider a range of variables in the student questionnaire of PISA that refer to the student’s home and family background, beliefs, goals, attitudes, perceptions, and school experiences. We focus on non-instructional and non-curriculum variables with the view of understanding the variables that identify the Filipino students who are most vulnerable to poor science learning.

Filipino students’ science literacy in PISA

The Philippines participated in PISA for the first time in 2018, with students’ answering the assessments in reading, mathematics, science, and global competencies. For science literacy assessment, the PISA 2018 Framework broadly defines science literacy as “the ability to engage with science-related issues, and with the ideas of science, as a reflective citizen” (OECD, 2019 a, 2019 b, p. 100). According to the PISA science framework, scientific literacy relies on a combination of knowledge and competencies that are applied to different contexts. Student performance was reported using seven levels of proficiency, with Level 6 being the highest level of proficiency and Level 2 as the minimum level of proficiency. Students who achieve Level 2 proficiency are able to demonstrate the ability to use basic or everyday knowledge to explain scientific phenomena in familiar contexts and to interpret simple data sets. This level of proficiency serves as a baseline or minimum evidence for science literacy.

There were 7233 15-year-old Filipino students who participated in the PISA 2018 cycle (OECD, 2019 a, 2019 b), where the Philippines ranked as one of the poor-performing countries in science. The country had an average score of 357 which is significantly below the OECD average score of 489 with boys and girls performing similarly (355 and 359 average performance, respectively). Only about 22% of these students achieved Science Literacy scores at Level 2 or higher. In comparison, an average of 78% of students from OECD countries reached Level 2 or higher in the science literacy assessment. Students at Level 2 or higher can recognize the correct interpretation for familiar scientific phenomena and can use such knowledge to identify, in simple cases, whether a conclusion is valid based on the data provided. The poor performance of Filipino students is reflected in the fact that around 77% of them did not reach the minimum proficiency level. At the lowest proficiency levels (1A and 1B), students are only able to use everyday content and procedural knowledge to explain simple or familiar phenomena. Their ability to understand data and to design scientific inquiry is highly limited (OECD, 2019 a, 2019 b).

The pattern of Filipino students’ performance in PISA 2018 matches their achievement in another international assessment, the Trends in International Mathematics and Science Study (TIMSS). Similar to PISA, TIMSS measures students’ ability to apply their knowledge in different content areas of science. Performance was evaluated using benchmarks, each with a corresponding scale score: Low (400), Intermediate (475), High (550), and Advanced (625) (Mullis et al., 2020 ). Fourth-grade Filipino students who participated in the TIMSS 2019 cycle achieved an average scale score of 249, the lowest in 58 participating countries with an overall average score of around 491. Only 19% of Filipino students achieved scores in the Low benchmark or higher, which implies that the overwhelming majority of Filipino students “show limited understanding of scientific concepts and limited knowledge of foundational science facts” (Mullis et al., 2020 , p. 107).

Such consistently poor achievement levels in science are very likely the results of a wide range of interacting factors. Previous research using PISA data has attempted to identify important factors that differentiate the performance of high and low high and low scorers in PISA. For example, to determine which factors contribute to the gap between high and low PISA science scores, Alivernini and Manganelli ( 2015 ) considered factors coming from country, school, and student levels. They applied a classification and regression tree analysis to the PISA 2006 data from 25 countries to identify the factors that predicted high (above Level 4) or low (below Level 2) proficiency. The strongest country-level predictor was teacher salary. At the school level, parental pressure on the school’s standards (for low teacher salaries) and school size (for high teacher salaries) predicted students’ PISA performance. At the student level, science self-efficacy and awareness of environmental issues determined whether a student would be a low or high performance in the PISA science assessment.

In this study, we employ a similar approach to studying the variables that might explain the poor performance of most Filipino students. We compare the group of poor-performing students with the group of better-performing students and consider variables related to the student’s family/home backgrounds, beliefs, goals, attitudes, perceptions, and school experiences. Instead of using statistical approaches, we use machine learning approaches to test models that best identify and distinguish the group of poor-performing students from the better-performing ones. Machine learning approaches have been proposed as complementary to statistical approaches (Lezhnina and Kismihok, 2022 ), particularly for purposes of handling very large numbers of variables in high-dimensional datasets (like those in the PISA) while avoiding convergence problems and for developing multidimensional complex models that may feature nonlinear relationships (Hilbert et al., 2021 ; Yarkoni and Westfall, 2017 ). Such machine learning approaches have been used to study science achievement in PISA 2015 (Chen et al., 2021 ), but the study focused on identifying the top performers, not the poor performers. Such approaches have been used to study the PISA 2018 data in other countries like China (Lee, 2022 ), Singapore (Dong and Hu, 2019 ), and the Philippines (Bernardo et al., 2021 , 2022 ), but these studies focused on predicting either performance in reading, mathematics, or the average across domains, and none so far, have focused on the PISA 2018 science results. The analytic approaches are discussed in the methods section. But we first consider the range of possible predictor variables suggested by the relevant literature and that were available in the PISA student questionnaire the Filipino students answered.

Predictors of science learning and achievement

Most studies on science education in the Philippines have focused on curriculum (Balagtas et al., 2019 ; Ely, 2019 ; Morales, 2017b ), knowledge, beliefs, and practices of science teachers (Bug-os et al., 2021 ; Macugay and Bernardo, 2013 ; Orbe et al., 2018 ; Walag et al., 2020 ), and beliefs and perceptions of science learning (Alonzo and Mistades, 2021 ; Bernardo et al., 2008 ; Magalong and Prudente, 2020 ; Montebon, 2014 ); typically such studies do not empirically establish any relationship with Filipino students’ science learning or achievement. But there are some studies that do identify some predictors of Filipino students learning and achievement in chemistry, biology, physics, or some specific science lessons. And these typically fall into two types of inquiries: (a) those that investigate the learning outcomes of particular instructional strategies (Antonio and Prudente, 2021 ; Francisco and Prudente, 2022 ; Magwilang, 2016 ; Morales, 2016 , 2017a ; Orozco and Yangco, 2016 ), and (b) those that looked into student motivations and other non-cognitive student level variables as predictors of learning and achievement (Bernardo, 2021 ; Bernardo et al., 2015 ; Ganotice and King, 2014 ; King and Ganotice, 2013 , 2014 ). In this study, we worked with variables from the student self-report questionnaire of PISA 2018, so we could not study instructional strategies (i.e., the first set of studies above), but we are able to study student-level variables similar to the latter group of studies that include motivation, self-beliefs and a host of other variables that relate to students family and home backgrounds, perceptions and attitudes related to their classroom and school experiences, and their goals and aspirations for after they finish high school. We consider what the research literature suggests about such variables below, starting with student-level variables that were included in the PISA 2018 self-report survey and that were found to be important predictors of science literacy in previous PISA research in different countries.

Student factors

Certain student characteristics have been shown to influence their performance in science or scientific literacy. Gender appears to be associated with scientific literacy, with boys performing better than girls in the 2015 PISA cycle (OECD, 2016 ), but the results of numerous other studies are mixed (Cutumisu and Bulut, 2017 ; Lam and Lau, 2014 ; Sun et al., 2012 ). Affective and motivational factors seem to be important correlates of science achievement in PISA; these factors include students’ enjoyment of science and perceived value of science (Ozel et al., 2013 ), positive motivations, interest, more sophisticated epistemic beliefs (Hofverberg et al., 2022 ; She et al. 2019 ), self-efficacy, intrinsic and instrumental motivations for learning science (Kartal and Kutlu, 2017 ; Mercan, 2020 ), having a growth mindset (Bernardo, 2021 , 2022 ; Bernardo et al., 2021 ), among others. Other motivation-related processes are also associated with science literacy achievement. These include students’ projective self-assessments of their own abilities and their future aspirations (Lee and Stankov, 2018 ), perseverance and willingness to solve problems (Cutumisu and Bulut, 2017 ), and use of metacognitive strategies (Akyol et al., 2010 ; Callan et al., 2016 ). Interestingly, students’ reading skills and reading strategies have also been associated with science achievement (Barnard-Brak et al., 2017 ; Caponera et al., 2016 ). The role of reading strategies is proposed to be important as science learning depends to an extent on students’ comprehension of scientific text (Cano et al., 2014 ; Kolić-Vrhovec et al., 2011 ) and this association seems particularly important when the students are learning science in a second language instead of their home language (Van Laere et al., 2014 ), which is the case with the Filipino students who participated in PISA 2018.

Family and home factors

The socioeconomic status (SES) of students’ families has been a consistent predictor of scores in PISA (Lam and Zhou, 2021 ; Lee and Stankov, 2018 ), and this is true in the domain of science (Sun et al., 2012 ). This variable has been unpacked and many other factors associated with SES have been identified as predictors of achievement in PISA. These SES-related factors include the educational attainment and occupation of their parents (Chen et al., 2021 ; Schulze and Lemmer, 2017 ). In one such study, researchers found that parents’ education had the largest indirect effect on children’s PISA test scores (Burhan et al., 2017 ). The influence of each parent’s education, however, appears to differ. A study that analyzed the PISA 2000 performance of 30 countries found that the mother’s educational attainment had a greater impact on students’ scores than the father’s educational attainment (Marks, 2008 ). Similar to education, parents’ occupations also predicted students’ learning outcomes. Students whose parents had a higher level of occupation were found to have higher scientific competencies than students whose parents were low-skilled workers (Chi et al., 2017 ). Another variable related to SES is the students’ access to information and communication technologies (ICT) at home, particularly ICT with access to the Internet. ICT availability and use positively predicted performance in various PISA assessments (Hu et al., 2018 ; Petko et al., 2017 ; Yoon and Yun, 2023 ). We also note that studies indicate SES seems to be associated with some student-level factors. For example, SES is strongly associated with feelings of school belonging (King et al., 2022 ).

Other than SES-related factors at home, parental involvement and family investment in children’s education also appear to influence students’ academic performance (Ho and Willms, 1996 ). Using data from a national survey of Chinese students’ science literacy, Wang et al. ( 2012 ) found that students’ high scores were associated with parents’ investment in their children’s education through the purchase of educational materials and other resources at home. A study of ninth-grade students in South Africa found that family experiences, such as the learning environment at home, were related to the student’s motivation to learn science (Schulze and Lemmer, 2017 ).

School factors

The school characteristics that have been shown to influence students’ scientific literacy performance include SES (or SES composition), school enrollment size, and location. Wang et al. ( 2012 ) found that certain school characteristics, namely school standing, having libraries and computer laboratories, good relationships between teachers and students, and funding for teacher training were associated with higher science literacy scores. School SES composition was found to be strongly associated with high scientific literacy scores of Australian students (McConney and Perry, 2010 ). Analysis of Hong Kong students’ PISA scores revealed that school SES composition partly explained differences in science achievement (Sun et al., 2012 ). Class size (Bellibaş, 2016 ; although see Lam and Lau, 2014 ) and school location (Topçu et al., 2014 ) are also predictors of science achievement.

Other than these school characteristics, students’ experiences and perceptions of their classroom and school environments also predict their achievement in PISA. In a study of the performance of Chinese students in the 2015 PISA, Huang ( 2020 ) found that reported experience of bullying in school was associated with achievement scores, and this relationship was medicated by the student’s sense of belonging in school. School disciplinary climate and quality of student-teacher relationship were significant predictors in particular countries (Shin et al., 2009 ); with the effect of disciplinary climate possibly having a more positive impact on students from low SES groups but the evidence across countries is mixed (Chi et al., 2018 ; Scherer, 2020 ).

The current study

The extant literature suggests that a wide range of factors at the student, family/home, and school level are likely predictors of science literacy, although some of these factors were shown to be important predictors in some countries but not all. In this study, we explore a range of such factors to inquire which best identifies the poor-performing Filipino students in contrast to the better-performing ones. The factors explored in the study are among those in the PISA 2018 student self-report survey that Filipino students answered.

Most education research that examines relationships among such variables applies statistical approaches. In such studies, correlations can show the linear relationship between each variable of poor and better-performing groups. In the current Philippines PISA 2018 dataset where we examine 85 variables as predictors, the possible correlations are over 7000 in number. For a more complex, nonlinear system with hundreds of variables that are not independent, we believe that it is best to use machine learning models. In contrast to the standard statistical approach, machine learning models capture the high-dimensional, possibly nonlinear, interrelations among a very large number of predictors (Hilbert et al., 2021 ; Yarkoni and Westfall, 2017 ), while identifying those most relevant to prediction. And, in order for the analysis to be more valid, we argue that the model should be optimal, in this case, the model with the best accuracy. For this study, we try out different machine-learning approaches to determine the best model to uncover the relationships between these variables.

The specific objective is to use machine learning approaches to determine the most accurate model that best identifies the Filipino students who performed at the lowest levels in the science domain of PISA 2018. We sought the best model that will indicate the factors that identify the students who are vulnerable to poor learning in science in the hope that the model will call the attention of Filipino educators to the non-instructional and non-curricular factors that contribute to poor learning in science among Filipino learners. The variables that were considered included student factors (e.g., motivations, self-beliefs, goals, aspirations), family/home factors (e.g., family SES, parents’ education and occupations, learning resources at home), and classroom/school factors (e.g., instruction time for science, teacher behaviors, perceived school environment, self-reported social experiences in school).

Our methodology for determining the best model that features the most important variables that identify the low-performing Filipino student in science is summarized in Fig. 1 , which shows the different phases of our data analysis. The first step is data preparation which entails data cleaning, that is, removal of variables with 100% missing data, identification, and imputation of entries with missing values, and variable scaling. Next is feature selection which involves the careful refinement of the list of variables that may contain negative suppressors. Then, machine learning model training follows to search for the best nonlinear prediction model. Finally, the machine learning model evaluation describes quantitatively the model performance and reports variable importance.

To find the optimal computational model, the whole data processing pipeline is performed for different sets of hyperparameters, for different machine learning approaches.

The dataset

The data we used in the analysis were from the Philippine sample in the PISA 2018 data (publicly accessible at https://www.oecd.org/pisa/data/2018database/ ). PISA 2018 test items for the science subject measure the ability to engage with science-related issues as a thoughtful citizen (OECD, 2019 a, 2019 b). To assess this, the questions given are related to contexts , e.g. personal, local and global issues, both current and historical that require understanding in science and technology; to knowledge , e.g. content, procedural, and epistemic; and to competencies that exhibit the ability to explain phenomena scientifically, evaluate and design scientific inquiry, and interpret data and evidence scientifically. In addition to these, students answered background questionnaires about themselves, their homes, and the school and learning experiences. As discussed, these variables were considered in this study. The performance of students is estimated and reported as 10 plausible values with 0.88 reliability for the Philippine science data.

A two-stage stratified sampling design was followed to obtain the nationally representative sample: (a) 187 schools were randomly selected from the country’s 17 regions, with the number of schools selected proportional to the regional distribution of schools, (b) students were then randomly selected from each school. As mentioned in the introduction, the final sample was 7233 15-year-old students. From the database, we identified 85 variables that referred to student, home/family, classroom/school factors suggested by the extant literature as possible predictors of science literacy, which we measured using the first plausible value of science literacy (PV1SCIE).

Data preparation

As reported earlier, over 80% of the Filipino students who participated in PISA 2018 were found to have less than Level 2 proficiency in science. The detailed distribution of participants across the different proficiency levels is shown in Fig. 2a . Because our goal is to identify the variables that are potentially influential in identifying the extreme poor performers in science, we decided to train a binary classification model that identifies these students and to study the variables that are important in this model prediction. For the binary classification, the sample data was divided into two categories; the (1) poor-performing students, who have proficiency at Level 1b and below, and (2) better-performing students, who have proficiency at Levels 1a, 2, and above. The data distribution for the two groups is shown in Fig. 2b .

Normalized Science proficiency level distribution of students ( a ) and distribution of students with poor and better performance in science ( b ). Poor performance category is for those students who belong to Level 1b and below proficiency levels, otherwise, the students are assigned to the better performance category.

The samples were further trimmed down as part of the data preparation. Students with more than 50% of the total variables missing were dropped from the dataset, obtaining the final data distribution in Table 1 . Sampled randomly, around 80% of the data were used as the training samples for the model training while the unseen or remaining data were used as the test data.

To avoid bias in training the model, data balancing was conducted by oversampling the poor-performance samples and undersampling the better-performance samples. For the poor-performing category with 2419 samples, the Synthetic Minority Oversampling Technique or SMOTE algorithm (Chawla et al., 2002 ) was applied to increase the samples. The SMOTE first chooses a random sample from the minority class, for example, sample A. Then, it looks for its nearest neighbor of the same class, for example, sample B. The algorithm performs a convex combination of the two samples to produce the synthetic sample. For the better-performing category with 3297 samples, the Tomek Links (Tomek, 1976 ) algorithm is used to undersample the majority class. The algorithm removes the ambiguous samples from the majority class, which is the data from the majority class that is closest to a minority class data. The final number of training samples for each of the poor and better performance categories is 3214.

The list of variables was further refined by removing variables with 100% missing values (i.e., the questions were not included in the set of questions asked for Filipino students). Those remaining variables with missing values were imputed using the k -nearest neighbor algorithm, where k  = 7. Also, initial experiments showed the occurrences of negative suppressors. To minimize the number of suppressors, we removed variables with high correlation with other factors, i.e. | ⍴ | > 0.75. Finally, normalization by scaling was performed per variable. In summary, 13 variables were removed because these variables have missing values only and 20 variables were removed because they have high correlation with other variables. The final number of variables is 72 plus the scientific literacy score of PV1SCIE.

Machine learning modeling

Our approach to determining the key variables that identify Filipino students with poor performance in science used machine learning, aiming to come up with a computational model that relates the input variables to the target variables. The design for the computational model is evaluated in terms of training and test accuracy to measure the model performance in both seen and unseen data and the Area under the Region-of-Convergence curve (AUC) to determine how well the separation of data is in the model space.

Exhaustive hyperparameter search on the following computational models: support vector machines (SVM), logistic regression, multilayer perceptron (MLP), decision tree, and random forest (RF). Performance in terms of accuracy revealed that the best model is the RF classifier, having 500 estimators, maximum decision tree depth equal to 20, and maximum features equal to ceiling(log 2 71) = 7 variables per individual tree, which is the best classifier. (Please refer to Supplementary File for the performance summary of the machine learning models considered.) To illustrate the RF model, please refer to Fig. 3 . The RF model is composed of several independent decision trees that are trained independently on a random subset of data. To measure the quality of a split, entropy is used to measure the information gain.

It is composed of n  = 500 decision trees called estimators with a maximum tree depth of 20. Each input to the estimator uses only a subset of variables equal to ceil(log 2 71) or 7 variables. This minimizes the model overfitting due to the original large number of variables.

Model performance

The summary of the model performance is shown in Fig. 4 . The positive class for this study refers to the poor-performing class while the negative class refers to the better-performing class. Since the test dataset is not balanced, three performance metrics were observed: classification accuracy, precision, and recall. Accuracy is the ratio of correctly classified students, whether poor-performing or better-performing students, over the total number of students. Precision is defined as the ratio of the number of correctly predicted poor-performing students and the total number of predicted poor-performing students. Recall is the ratio of correctly predicted poor-performing students and the total number of poor performing. High precision and recall show that the model is returning accurate results (high precision), and returning a majority of all positive results (high recall).

The confusion matrix ( a ) and the ROC ( b ) summarizing the performance of the RF model in classifying the PISA 2018 Science Proficiency of Filipino students. The average accuracy is 0.74 and the area under the ROC being equal to 0.83.

The RF Classifier returned a good balance of precision and recall on the training data with values equal to 0.74, and 0.79, respectively. In addition to this balance, among the different classifiers considered, the grid-search accuracy (see Fig. 5 ), shows that the RF classifier returned the best performance with final accuracy equal to 0.74, considering the precision and recall balance. The final precision, recall and accuracy using the test data are 0.73, 0.66, and 0.74, respectively. The area under the receiver operating curve (AUC) is 0.83 which implies a fairly good-fit model. A perfect classifier has AUC = 1.0 which implies that the model was able to separate the two classes, i.e. positive and negative, of data. The worst classifier, i.e. chance level accuracy, has AUC = 0.5.

The scatterplot illustrating the range of test accuracies during the cross-validation on best machine learning models shows that the RF returned the best accuracies.

Model Interpretation

We investigated the feature importance learned during training by the RF classifier. We used Shapley additive explanations (SHAP) which is a scheme based on cooperative game theory to interpret the contributions of features in the prediction. For the RF classifier in this study, these top 15 key features or variables are shown in Fig. 6 . Footnote 1 The important variables can positively affect or negatively affect the prediction of poor performance class ( y  = 1). Particularly, one student with higher values for the variables BELONG, WORKMAST and BEINGBULLIED will negatively affect the prediction of identifying the poor performers in science. Similarly, for students with high ST164Q05IA, BSMJ, and HISEI values, the prediction of identifying poor performers in science is higher since these values positively impact this prediction. We describe the 15 variables in meaningful groupings below.

Blue bars represent variables that negatively affect the prediction of poor-performing students while red bars indicate that a variable positively affects the prediction of poor-performing students.

The largest cluster of variables relates to students’ metacognitive awareness in reading or their perceptions about the usefulness of particular metacognitive strategies when reading texts in their classes. These variables are related items, where students were asked to indicate whether the indicated strategy is useful for understanding and memorizing the texts they read. Three of the variables positively identified the poor-performing students: (ST164Q05IA) “I summarize the text in my own words,” (ST164Q04IA) “I underline important parts of the text,” and (ST164Q02IA) “I quickly read through the text twice.” These three reading strategies involve relatively low metacognitive skills and are often ineffective, and poor-performing Filipino science students tend to see them as useful. On the other hand, two of the variables negatively identified poor-performing students: (ST164Q01IA) “I concentrate on the parts of the text that are easy to understand,” and (ST164Q03IA) “After reading the text, I discuss its content with other people.” The poor-performing Filipino science students tend to perceive these strategies as not useful.

The next largest cluster of variables relates to the student’s classroom and school experiences. Sense of belonging (BELONG) and perceived cooperation among students (PERCOOP) both negatively identify poor-performing students; that is, students who perform poorly in science report a low sense of belonging and perceive less cooperation among students. These two variables suggest negative social relations experienced by poor performers in science. Fortunately, self-report of experiencing bullying (BEINGBULLIED) was also negatively identified as the poor performers in science, so they tended to report less experiences of bullying in school. The last variable related to classroom experiences was how often “Students don’t listen to what the teacher says” (ST097Q01TA), which negatively identified the poor performers in science. The poor-performing science students were less likely to say that students often do not listen to the teacher. We should clarify that the item refers to teachers who use English in their classes, which refers to teachers in several subjects including science, mathematics, and English.

Three variables relate to the students’ affective or motivational experiences. The student’s motivation to master assigned learning tasks (WORKMAST) negatively identify poor-performing students, which means they tend to have low work mastery motivation. On the other hand, the student’s expected occupational status (BSMJ) and feeling proud about the things they accomplished (ST188Q02HA) both positively identified the poor-performing students. So the students who scored very low scores in science also tended to report higher job aspirations and being proud of their accomplishments compared to others. It seems that the student’s low achievement in science is unrelated to their future occupational plans and their present sense of accomplishment.

Finally, the remaining variables relate to the student’s family and home learning resources. Having smartphones with internet access at home (ST012Q05NA) negatively identified the poor-performing students, which means they were less likely to have this learning resource. But interestingly, the mother’s education (ST005Q01TA) negatively identified the poor-performing students, but the parents’ occupational status (HISEI) positively identified the students. This means that having mothers with lower educational attainment but having parents with high-status occupations also identified the students who were performing poorly in science. We could be seeing a pattern where low achievement in science is probably not viewed or experienced as a hindrance to higher-status professions. We explore this point and other results in the discussion section.

We used machine learning approaches to explore the best model for identifying the poorest-performing Filipino students in science using the PISA 2018 data. The Random Forest model was found to have the highest accuracy performance and the SHAP analysis indicated 15 variables that identified the poorest-performing science students.

Caveats and limitations

Before we discuss the meaning and implications of the details of the results, we need to underscore some important limitations in our study. First, our study cannot speak to the instructional and curricular factors that are typically the subject of discussions on improving science education in the Philippines. Second, the predictors in the model were limited to the variables in the PISA student self-report questionnaire. While there was a wide range of variables in the student questionnaire, many of the questions referred to reading (because the 2018 cycle of PISA was focused on reading), and thus, could not be included in our study. We also did not include variables from the school-head questionnaire about school characteristics, resources, and practices; nor could we include other potentially important predictors of science achievement that were not included in the PISA. Thus, there are possibly other variables that identify poor-performing students that are beyond the scope of this inquiry.

One important caveat relates to the predictive nature of the machine learning approaches, which treat variables equally without any theoretical presuppositions. Machine learning approaches focus on prediction accuracy and is not used to test explanatory models that specify theoretical relationships among variables (Shmueli, 2010 ; Yarkoni and Westfall, 2017 ). As such, the variables identified in the most accurate model may not have any obvious theoretical connection. These caveats notwithstanding, there are useful insights revealed by the analysis, which we discuss below.

Reading strategies for learning science

Metacognitive awareness regarding five different strategies identified the poor performers in science. It may seem surprising that reading strategies play an important role in identifying poor science performers, but the results make sense if one considers that much of science learning might be based on reading science textbooks (instead of doing laboratory experiments or field projects). Research with Italian students, for example, showed a difference in science achievement between good and bad readers, regardless of whether the science items involved low or high reading demand (Caponera et al., 2016 ). There were similar associations between reading comprehension and science achievement in a study of Spanish (Cano et al., 2014 ) and Filipino students (Imam et al., 2014 ). We note that our results do not actually involve reading comprehension, but metacognitive awareness of reading strategies, similar to a study of Croatian students that established a relationship between students’ reading strategies and comprehension of scientific texts (Kolić-Vrhovec et al., 2011 ). It is plausible that poor achievers in science are those that might be adopting the wrong reading strategies in reading their science textbooks.

Families’ and students’ resources and aspirations

High social, cultural, and economic resources in the students’ families (Lam and Zhou, 2021 ; Sun et al., 2012 ) and higher professional aspirations (Lee and Stankov, 2018 ) are typically associated with better achievement. But in our results, the poor-performing students were identified by higher job aspirations and stronger pride about one’s achievements. It is as if low achievement in science was not a consideration when students think about their future occupations nor when they assess their self-worth and pride. If we consider that the lower educational attainment of the mothers and higher occupational status of parents also predicted the poor-performing, it may be that students view their poor achievement in science as not relevant to their future occupational prospects, as their parents enjoy good occupations, even if their mothers are not highly educated. This interpretation asserts that science achievement might not be valued in pragmatic terms by the students based on what they see in their elders, which might also explain the role of low work mastery in school in identifying poor-performing students. Indeed, it is possible that many high-status occupations in the Philippines do not require knowledge of science, and as such, persevering and doing well in science might not be an important motivation among the students. This interpretation will need to be verified in future studies.

Negative social experiences

It was interesting to note that experience of bullying was a negative factor in the model, so it was not the case that experience of bullying was positively linked to poorer science achievement, as was found in Chinese students (Huang, 2020 ). However, two factors that indicate relational issues in school are identified with the poor performers: reporting a low sense of belonging and low cooperation among students in school. These factors suggest that a lack of connectedness and a collective spirit might be associated with poor science performance. Trinidad ( 2020 ) found that school-level and student-level measures of school climate were predictors of Filipino students’ mathematics achievement; such social factors might also have similar roles in Filipino learners’ poor science achievement.

Access to ICT for learning

One factor that may be increasingly important in identifying poor science achievers is access to ICT devices with internet access. Studies on Filipino students; PISA achievement in reading (Bernardo et al., 2021 ) and mathematics (Bernardo et al., 2022 ) also found the same factor as a predictor of achievement, consistent with much of the research in other countries (Hu et al., 2018 ; Petko et al., 2017 ; Yoon and Yun, 2023 ; but see Bulut and Cutumisu, 2018 ). Presumably, access to the internet outside the school environment has become an important resource for learning science; perhaps not just for accessing relevant scientific knowledge available online but also as a means of communicating with classmates for information sharing, collaboration in learning activities, and supporting each other’s motivations and engagement in science learning. Filipino students without such access are disadvantaged in the domain of science.

Practical implications: Focusing on the lowest achievers

The current study provides some analysis that could inform reform efforts in the domain of science learning, particularly as it concerns the lowest-achieving Filipino students in science. The results and discussion focus on factors that seem to characterize these lowest-achieving science students, and as such provide entry points to identifying these students and designing interventions for this particular group of students. Our approach focuses on the sizable proportion (over 35%) of Filipino students who have been assessed as demonstrating extremely low competencies (levels 1b and below) in science. The Philippine educational system does not lack programs for the more gifted students in science such as special science schools (Faustino and Hiwatig, 2012 ), competitions, scholarships, and other forms of support for students pursuing advanced studies and careers in science (De La Cruz, 2022 ). But there is not much that is documented about what is being done for the students like the 35% who are demonstrating extremely low levels of scientific literacy, even if they have progressed to the high school levels of the country’s formal education system. The first important implication of our findings is that these students need to be identified and understood before their science learnings can be addressed.

We should clarify that the characterization of poor-performing Filipino students in science should not be interpreted as the opposite characterization of better- or high-performing students. It is likely that there are qualitative differences between the experiences of poor and better science learners that are not captured by simply assuming a linear relationship between the factors that predict science learning. Indeed, if our machine learning approach was applied to identify the high-achieving students (i.e., Levels 4–6), it is likely that a different set of variables will be in the best machine learning model (and that can be explored in a different study). But by implication, the characterization of the poor-performing students in the results does not point to simple instructional or curricular interventions, and we do think there are some important policy implications that can be considered by stakeholders who are concerned with improving science education achievement among Filipino learners.

Instructional programs for poor achievers

Science educators have long noted that there are profound diversities in students of different ability levels, that simply assuming that one form of good teaching fits all types of learners is no longer tenable (Ault, 2010 ; Lynch, 2001 ; Yang et al., 2019 ). In this regard, the science education reform community of stakeholders should consider moving away from a one-size-teaching-fits-all approach that tends to be designed for students in the middle range of abilities using whole class instruction, and instead, move towards approaches that consider diverse adaptive learning approaches (Yang et al., 2019 ) and differentiated instruction (Pablico et al., 2017 ) that might be more responsive to (or at least that might not simply ignore) the needs of the low achievers.

Ensuring reading skills

There is a lot of evidence that good reading strategies and reading comprehension are strongly associated with science achievement (Cano et al., 2014 ; Caponera et al., 2016 ; Kolić-Vrhovec et al., 2011 ), but Filipino learners on average have extremely poor reading skills in English (Bernardo et al., 2021 ), which is the medium-of-instruction in science. Presumably, there are science learning activities that are more experiential and discovery-oriented and less dependent on students’ reading textbooks; but a previous study of students’ perception of science classes revealed a trend of decreasing science inquiry activities accompanied by an increase in self-learning, presumably involving reading textbooks and learning modules from Grade 5 to 10 (Bernardo et al., 2008 ). If Filipino science learners will be expected to do much of their learning through textbooks and learning modules in English, there should be strong efforts to strengthen the reading strategies and skills of Filipino learners.

Science in future professions and Philippine society

We interpreted part of the results as being associated with the view that science learning and achievement are irrelevant to higher future occupational aspirations. While these interpretations are speculative, there is probably a strong basis for the view that one does not need science to attain respectable occupations in the Philippines. There are many models of successful Filipino professionals and individuals who do not seem overtly display knowledge and use of science. In this regard, efforts to improve the science achievement of Filipino students might need to reckon with the perceived irrelevance of science in Philippine society. Scholars have problematized the lack of a science culture in the Philippines (Pertierra, 2004 ), perhaps vividly displayed in the recent COVID-19 pandemic, when there was widespread uncritical sharing of misinformation on vaccines, false cures, and other scientific matters through social media and social networks (Amit et al., 2022 ) and when scientific advice on pertinent issues was diluted and filtered before decisions were made by national leaders (Vallejo and Ong, 2020 ). Beyond schools, there should be efforts to change public perceptions of the importance of science in Filipinos’ social mobility and Filipino society’s development.

Improving school climate

The poor-performing students in science were identified by reports of a low sense of belonging in school and low perceived cooperation among students. These social experiences may be associated with lower achievement as they indicate a lack of meaningful sense of connectedness with students and teachers in the school, which is associated with lower engagement in the science classes, even if the social experiences are not specifically confined or referring to the science classes. The factors that contribute to these negative social experiences might vary across schools and communities and should be understood in proper contexts. Once the nature and causes of these social experiences are better understood, appropriate contextualized interventions can be developed.

Access to ICT devices and connectivity

Previous studies have documented how ICT availability and use positively predicted student achievement (Hu et al., 2018 ; Petko et al., 2017 ; Yoon and Yun, 2023 ), and similar results were also found in Filipino students’ achievement in reading (Bernardo et al., 2021 ), mathematics (Bernardo et al., 2022 ), and now in science. Together with improving access to the internet, there should be an effort to train teachers and students how to more effectively use the internet to deepen their learning of science concepts and processes, and in ways that adapt to students’ diverse abilities, interests, motivations, and circumstances (Yang et al., 2019 ).

Conclusions

Based on the assumption that science-for-all requires all Filipino citizens to acquire the scientific literacy required to effectively engage with and contribute to Philippine society in the 21st Century, we focused on the Filipino students with the lowest levels of science achievement in PISA 2018. We used machine learning to explore the variables that best identify the poor-performing Filipino students, as these variables could be used to better track and understand their learning needs. Our study points to a cluster of variables related to the student’s reading strategies, occupational aspirations, social experiences in school, and access to ITC and the internet. The variables depart from the typical focus of reform efforts on teachers’ competencies, curriculum, and instruction. But if we truly want to improve Filipino students’ science literacy, we need to understand the experiences of students who are failing to do so, as these point to problems that need to be addressed in their learning experiences in Philippine schools.

Data availability

The data analyzed in this study are available on the PISA 2018 Database page on the website of the Organisation for Economic Co-operation and Development at https://www.oecd.org/pisa/data/2018database/ , accessed on 17 Feb 2020.

For completeness, we also conducted a SHAP analysis for the best algorithm for each of the other machine learning approaches. A comparative summary of the top 15 variables that feature in the prediction models is shown in Supplementary File.

Akyol G, Sungur S, Tekkaya C(2010) The contribution of cognitive and metacognitive strategy use to students’ science achievement. Educ Res Eval 16(1):1–21. https://doi.org/10.1080/13803611003672348

Article   Google Scholar  

Alivernini F, Manganelli S (2015) Country, school and students factors associated with extreme levels of science literacy across 25 countries. Int J Sci Educ 37(12):1992–2012. https://doi.org/10.1080/09500693.2015.1060648

Alonzo SMD, Mistades VM (2021) Students’ conceptual understanding and problem-solving of the Work-Energy and Impulse-Momentum Theorems in a flipped classroom. J Phys Conf Ser 1882(1):012003. https://doi.org/10.1088/1742-6596/1882/1/012003

Amit AML, Pepito VCF, Sumpaico-Tanchanco L, Dayrit MM (2022) COVID-19 vaccine brand hesitancy and other challenges to vaccination in the Philippines. PLoS Glob Public Health 2(1):e0000165. https://doi.org/10.1371/journal.pgph.0000165

Article   PubMed   PubMed Central   Google Scholar  

Antonio RP, Prudente MS (2021) Metacognitive argument-driven inquiry in teaching antimicrobial resistance: Effects on students’ conceptual understanding and argumentation skills. J Turk Sci Ed 18(2):192–217. https://doi.org/10.36681/tused.2021.60

Ault CR (2010) One size fits none? J Sci Teacher Educ 21(1):1–5. https://doi.org/10.1007/s10972-009-9156-5

Balagtas MU, Garcia DCB, Ngo DC (2019) Looking through Philippine’s K to 12 curriculum in mathematics and science vis-a-vis TIMSS 2015 Assessment Framework. Eurasia J Math Sci T 15(12):1788. https://doi.org/10.29333/ejmste/108494

Barnard-Brak L, Stevens T, Ritter W (2017) Reading and mathematics equally important to science achievement: results from nationally-representative data. Learn Individ Differ 58:1–9. https://doi.org/10.1016/j.lindif.2017.07.001

Bellibaş MŞ (2016) Who are the most disadvantaged? Factors associated with the achievement of students with low socio-economic backgrounds. Educ Sci-Theor Pract 16(2):691–710. https://doi.org/10.12738/estp.2016.2.0257

Belmi RM, Mangali GR (2020) PISA 2018 science framework vis-a-vis the Philippine Kto12 science curriculum. In: Balagtas MU, Montealegre MC (eds) Challenges of PISA: the PNU report 100. Philippine Normal University & Rex Institute for Student Excellence, Manila, pp. 101–141

Google Scholar  

Bernardo ABI (2021) Socioeconomic status moderates the relationship between growth mindset and learning in mathematics and science: evidence from PISA 2018 Philippine data. Int J Sch Educ Psychol 9(2):208–222. https://doi.org/10.1080/21683603.2020.1832635

Bernardo ABI (2022) Growth mindset and reading proficiency of ESL learners: examining the role of students’ socioeconomic status using PISA 2018 Philippine data. Eur J Psychol Educ https://doi.org/10.1007/s10212-022-00629-6

Bernardo ABI, Cordel MO, Lapinid MR, Teves JMM, Yap SA, Chua UC (2022) Contrasting profiles of low-performing mathematics students in public and private schools in the Philippines: insights from machine learning. J Intell 10(3):61. https://doi.org/10.3390/jintelligence10030061

Bernardo ABI, Cordel MO, Lucas RIG, Teves JMM, Yap SA, Chua UC (2021) Using machine learning approaches to explore non-cognitive variables influencing reading proficiency in English among Filipino learners. Educ Sci 11(10):628. https://doi.org/10.3390/educsci11100628

Bernardo ABI, Ganotice FA, King RB (2015) Motivation gap and achievement gap between public and private high schools in the Philippines. Asia-Pac Educ Res 24(4):657–667. https://doi.org/10.1007/s40299-014-0213-2

Bernardo ABI, Limjap AA, Prudente MS, Roleda LS (2008) Students’ perceptions of science classes in the Philippines. Asia Pac Educ Rev 9(3):285–295. https://doi.org/10.1007/BF03026717

Bug-os MAAC, Walag AMP, Fajardo MTM (2021) Science teacher’s personal and subject-specific self-efficacy in teaching science: The case of El Salvador City, Philippines. Sci Int 33(3):179–186

Bulut O, Cutumisu M (2018) When technology does not add up: ICT use negatively predicts mathematics and science achievement for Finnish and Turkish students in PISA 2012. J Educ Multimedia Hypermedia 27(1):25–42

Burhan NA, Yunus MM, Tovar ME, Burhan NM (2017) Why are cognitive abilities of children so different across countries? The link between major socioeconomic factors and PISA test scores. Pers Individ Differ 105:95–106. https://doi.org/10.1016/j.paid.2016.09.043

Callan GL, Marchant GJ, Finch WH, German RL (2016) Metacognition, strategies, achievement, and demographics: relationships across countries. Educ Sci-Theory Pract 16 (5) https://doi.org/10.12738/estp.2016.5.0137

Cano F, García Á, Berbén ABG, Justicia F (2014) Science learning: a path analysis of its links with reading comprehension, question-asking in class and science achievement. Int J Sci Educ 36(10):1710–1732. https://doi.org/10.1080/09500693.2013.876678

Caponera E, Sestito P, Russo PM (2016) The influence of reading literacy on mathematics and science achievement. J Educ Res 109(2):197–204. https://doi.org/10.1080/00220671.2014.936998

Chawla NV, Bowyer KW, Hall LO, Kegelmeyer WP (2002) SMOTE: synthetic minority over-sampling technique. J Artif Intell Res 16:321–357. https://doi.org/10.1613/jair.953

Article   MATH   Google Scholar  

Chen J, Zhang Y, Wei Y, Hu J (2021) Discrimination of the contextual features of top performers in scientific literacy using a machine learning approach. Res Sci Educ 51(1):129–158. https://doi.org/10.1007/s11165-019-9835-y

Chi S, Liu X, Wang Z, Won Han S (2018) Moderation of the effects of scientific inquiry activities on low SES students’ PISA 2015 science achievement by school teacher support and disciplinary climate in science classroom across gender. Int J Sci Educ 40(11):1284–1304. https://doi.org/10.1080/09500693.2018.1476742

Chi S, Wang Z, Liu X, Zhu L (2017) Associations among attitudes, perceived difficulty of learning science, gender, parents’ occupation and students’ scientific competencies. Int J Sci Educ 39(16):2171–2188. https://doi.org/10.1080/09500693.2017.1366675

Cordon JM, Polong JDB (2020) Behind the science literacy of Filipino students at PISA 2018: a case study in the Philippines’ educational system. In Sci Ed J 1(2):70–76. https://doi.org/10.37251/isej.v1i2.59

Cutumisu M, Bulut O (2017) Problem-solving attitudes and gender as predictors of academic achievement in mathematics and science for Canadian and Finnish students in the PISA 2012 assessment. In: Johnston J (ed) Proceedings of EdMedia. Association for the Advancement of Computing in Education, Washington, DC, pp 728–738

De La Cruz RJD (2022) Science education in the Philippines. In: Huang R, et al., (eds) Science Education in Countries Along the Belt & Road. Springer, Singapore, pp. 331–345

Chapter   Google Scholar  

Dong X, Hu J (2019) An exploration of impact factors influencing students’ reading literacy in Singapore with machine learning approaches. Int J Engl Linguist 9(5):52–65. https://doi.org/10.5539/ijel.v9n5p52

Ely LL (2019) Mastery learning of chemistry competencies through the spiral progression approach in curriculum. Int J Educ Sci Res 9(5):9–28

Faustino JB, Hiwatig ADF (2012) Special science elementary school: Project and prospects for gifted education in the Philippines. J Sci Educ Japan 36(2):131–141

Fensham P (2008) Science education policy-making. UNESCO, Paris

Francisco Jr LM, Prudente MS (2022) Improving students’ conceptual knowledge in online distance learning through the use of micro-lectures: A photosynthesis example. Innov Technol Manag J 5:1

Ganotice FA, King RB (2014) Social influences on students’ academic engagement and science achievement. Psychol Stud 59(1):30–35. https://doi.org/10.1007/s12646-013-0215-9

Hilbert S, Coors S, Kraus E et al. (2021) Machine learning for the educational sciences. Rev Educ 9(3):e3310. https://doi.org/10.1002/rev3.3310

Ho SC, Willms JD (1996) The effect of parental involvement on the achievement of eighth grade students. Sociol Educ 69(2):126–141

Hofverberg A, Eklöf H, Lindfors, M (2022) Who makes an effort? A person-centered examination of motivation and beliefs as predictors of students’ effort and performance on the PISA 2015 science assessment. Front Educ 6. https://doi.org/10.3389/feduc.2021.791599

Hu X, Gong Y, Lai C, Leung FK (2018) The relationship between ICT and student literacy in mathematics, reading, and science across 44 countries: a multilevel analysis. Comput Educ 125:1–13. https://doi.org/10.1016/j.compedu.2018.05.021

Article   ADS   Google Scholar  

Huang L (2020) Exploring the relationship between school bullying and academic performance: the mediating role of students’ sense of belonging at school. Educ Stud 48(2):216–232. https://doi.org/10.1080/03055698.2020.1749032

Article   MathSciNet   Google Scholar  

Imam OA, Mastura MA, Jamil H, Ismail Z (2014) Reading comprehension skills and performance in science among high school students in the Philippines. Asia Pac J Educ Educ 29:81–94

Kartal SK, Kutlu Ö (2017) identifying the relationships between motivational features of high and low-performing students and science literacy achievement in PISA 2015 Turkey. J Educ Train Stud 5(12):146–154. https://doi.org/10.11114/jets.v5i12.2816

King RB, Chiu MM, Du H (2022) Greater income inequality, lower school belonging: Multilevel and cross-temporal analyses of 65 countries. J Educ Psychol. https://doi.org/10.1037/edu0000736

King RB, Ganotice Jr FA (2013) Student motivation as hierarchical and multidimensional: Cross-cultural validation of personal investment theory in the Philippines. Univ Psychol 12(3):685–698. https://doi.org/10.11144/Javeriana.UPSY12-3.smhm

King RB, Ganotice FA (2014) The social underpinnings of motivation and achievement: Investigating the role of parents, teachers, and peers on academic outcomes. Asia-Pac Educ Res 23(3):745–756. https://doi.org/10.1007/s40299-013-0148-z

Kolić-Vrhovec S, Bajšanski I, Rončević Zubković B (2011) The role of reading strategies in scientific text comprehension and academic achievement of university students. Rev Psychol 18(2):81–90

Lam TYP, Lau KC (2014) Examining factors affecting science achievement of Hong Kong in PISA 2006 using hierarchical linear modeling. Int J Sci Educ 36(15):2463–2480. https://doi.org/10.1080/09500693.2013.879223

Lam SM, Zhou Y (2021) SES-achievement gaps in East Asia: Evidence from PISA 2003–2018. Asia-Pac Educ Res 1-20. https://doi.org/10.1007/s40299-021-00620-7

Lee H (2022) What drives the performance of Chinese urban and rural secondary schools: A machine learning approach using PISA 2018. Cities 123:103609. https://doi.org/10.1016/j.cities.2022.103609

Lee J, Stankov L (2018) Non-cognitive predictors of academic achievement: Evidence from TIMSS and PISA. Learn Individ Differ 65:50–64. https://doi.org/10.1016/j.lindif.2018.05.009

Lezhnina O, Kismihók G (2022) Combining statistical and machine learning methods to explore German students’ attitudes towards ICT in PISA. Int J Res Method Educ 45(2):180–199. https://doi.org/10.1080/1743727X.2021.1963226

Lynch S (2001) “Science for all” Is not equal to “one size fits all”: Linguistic and cultural diversity and science education reform. J Res Sci Teach 38(5):622–27

Macugay EB, Bernardo AB (2013) Science coursework and pedagogical beliefs of science teachers: the case of science teachers in the Philippines. Sci Educ Int 24(1):63–77

Magalong SJM, Prudente M (2020) Exploring students’perceptions and conceptual understanding in a next generation blended learning (NXGBL) physics class. Innov Technol Manag J 3:35–44

Magwilang EB (2016) Teaching chemistry in context: Its effects on students’ motivation, attitudes and achievement in chemistry. Int J Learn Teach Educ Res 15(4):60–69

Marks GN (2008) Are father’s or mother’s socioeconomic characteristics more important influences on student performance? Recent international evidence. Soc Indic Res 85(2):293–309. https://doi.org/10.1007/s11205-007-9132-4

Martin MO, Mullis IVS, Gonzalez EJ, Chrostowski SJ (2004) TIMSS 2003 international science report. TIMSS & PIRLS International Study Center

McConney A, Perry LB (2010) Science and mathematics achievement in Australia: the role of school socioeconomic composition in educational equity and effectiveness. Int J Sci Math Educ 8(3):429–452. https://doi.org/10.1007/s10763-010-9197-4

Mercan FC (2020) Control-value theory and enjoyment of science: A cross-national investigation with 15-year-olds using PISA 2006 data. Learn Individ Differ 80:101889. https://doi.org/10.1016/j.lindif.2020.101889

Montebon DT (2014) K12 science program in the Philippines: Student perception on its implementation. Int J Educ Res 2(12):153–164

Morales MPE (2016) Exploring indigenous game-based physics activities in pre-service physics teachers’ conceptual change and transformation of epistemic beliefs. Eurasia J Math Sci Technol 13(5):1377–1409. https://doi.org/10.12973/eurasia.2017.00676a

Morales MPE (2017b) Transitions and transformations in Philippine physics education curriculum: A case research. Issues. Educ Res 27(3):469–492

Morales MPE (2017a) Cultural historical activity theory (chat): influenced case research of a Philippine physics class. Asia-Pac Educ Res 26(1):85–96. https://doi.org/10.1007/s40299-017-0329-2

Mullis IVS, Martin MO, Foy P, Kelly DL, Fishbein B (2020) TIMSS 2019 international results in mathematics and science. TIMSS & PIRLS International Study Center, Boston College and International Association for the Evaluation of Educational Achievement, Chestnut Hill, MA

Orbe JR, Espinosa AA, Datukan JT (2018) Teaching chemistry in a spiral progression approach: Lessons from science teachers in the Philippines. Aust J Teach Educ 43(4):17–30

Organisation for Economic Cooperation and Development (2016) PISA 2015 results (vol I): excellence and equity in education. OECD Publishing, Paris

Organisation for Economic Cooperation and Development (2019a) PISA 2018 assessment and analytical framework. OECD Publishing, Paris

Organisation for Economic Cooperation and Development (2019b) PISA 2018 results (Volume I): What students know and can do. OECD Publishing, Paris

Orozco JA, Yangco RT (2016) Problem-based learning: effects on critical and creative thinking skills in biology. Asian J Biol Educ 9:3

Ozel M, Caglak S, Erdogan M (2013) Are affective factors a good predictor of science achievement? Examining the role of affective factors based on PISA 2006. Learn Individ Differ 24:73–82. https://doi.org/10.1016/j.lindif.2012.09.006

Pablico JR, Diack M, Lawson A (2017) Differentiated instruction in the high school science classroom: Qualitative and quantitative analyses. Int J Learn Teach Educ Res 16(7):30–54

Pertierra R (2004) Is there a cultural of science in the Philippines. Agham-Tao 13(6):1–19

Petko D, Cantieni A, Prasse D (2017) Perceived quality of educational technology matters: a secondary analysis of students’ ICT use, ICT-related attitudes, and PISA 2012 test scores. J Educ Comput Res 54(8):1070–1091. https://doi.org/10.1177/0735633116649373

Scherer R (2020) The case for good discipline? Evidence on the interplay between disciplinary climate, socioeconomic status, and science achievement from PISA 2015. In:Frønes TS, Pettersen A, Radišić J, Buchholtz N (eds) Equity, equality and diversity in the Nordic model of education. Springer, Cham, pp. 197–224

Schulze S, Lemmer E (2017) Family experiences, the motivation for science learning and science achievement of different learner groups. S Afr J Educ 37(1) https://doi.org/10.15700/saje.v37n1a1276

She HC, Lin HS, Huang LY (2019) Reflections on and implications of the Programme for International Student Assessment 2015 (PISA 2015) performance of students in Taiwan: the role of epistemic beliefs about science in scientific literacy. J Res Sci Teach 56(10):1309–1340. https://doi.org/10.1002/tea.21553

Shin J, Lee H, Kim Y (2009) Student and school factors affecting mathematics achievement: International comparisons between Korea, Japan and the USA. Sch Psychol Int 30(5):520–537. https://doi.org/10.1177/0143034309107070

Shmueli G (2010) To explain or to predict. Stat Sci 25:289–319. https://doi.org/10.1214/10-STS330

Article   MathSciNet   MATH   Google Scholar  

Sumardani D (2021) Philippines: Strength and weakness of science curricula. Sci Educ J 5(2):99–106. https://doi.org/10.21070/sej.v5i2.1507

Sun L, Bradley KD, Akers K (2012) A multilevel modelling approach to investigating factors impacting science achievement for secondary school students: PISA Hong Kong sample. Int J Sci Educ 34(14):2107–2125. https://doi.org/10.1080/09500693.2012.708063

Talisayon VM, Balbin CR, De Guzman FS (2006) Predictors of student achievement in TIMSS 1999 and Philippine results. In: Howie SJ, Plomp T (eds) Context of learning mathematics and science: Lessons learned from TIMSS. Routledge, London. p 225–242

Tomek I (1976) Two modifications of CNN. IEEE Trans Syst Man Cybern 6(11):769–772. https://doi.org/10.1109/TSMC.1976.4309452

Topçu MS, Arıkan S, Erbilgin E (2014) Turkish students’ science performance and related factors in PISA 2006 and 2009. Aust Educ Res 42(1):117–132. https://doi.org/10.1007/s13384-014-0157-9

Trinidad JE (2020) Material resources, school climate, and achievement variations in the Philippines: Insights from PISA 2018. Int J Educ Dev 75:102174. https://doi.org/10.1016/j.ijedudev.2020.102174

Vallejo Jr BM, Ong RAC (2020) Policy responses and government science advice for the COVID 19 pandemic in the Philippines: January to April 2020. Prog Disaster Sci 7:100115. https://doi.org/10.1016/j.pdisas.2020.100115

Van Laere E, Aesaert K, van Braak J (2014) The role of students’ home language in science achievement: A multilevel approach. Int J Sci Educ 36(16):2772–2794. https://doi.org/10.1080/09500693.2014.936327

Walag AMP, Fajardo MTM, Guimary FM, Bacarrisas PG (2020) Science teachers’ self-efficacy in teaching different K to 12 science subjects: The case of Cagayan De Oro City. Philippines. Sci Int 32(5):587–592

Wang S, Liu X, Zhao Y (2012) Opportunities to learn in school and at home: how can they predict students’ understanding of basic science concepts and principles? Int J Sci Math Educ 34(13):2061–2088. https://doi.org/10.1080/09500693.2012.709335

Yang S, Tian H, Sun L, Yu X (2019) From one-size-fits-all teaching to adaptive learning: the crisis and solution of education in the era of AI. Int J Phys Conf Ser 1237(4 Jun):042039. https://doi.org/10.1088/1742-6596/1237/4/042039 . IOP Publishing

Yarkoni T, Westfall J (2017) Choosing prediction over explanation in psychology: Lessons from machine learning. Perspect Psychol Sci 12(6):1100–1122. https://doi.org/10.1177/1745691617693393

Yoon M, Yun H (2023) Relationships between adolescent smartphone usage patterns, achievement goals, and academic achievement. Asia Pac Educ Rev 24:13–23. https://doi.org/10.1007/s12564-021-09718-5

Download references

Acknowledgements

This research was funded by a grant to the third author from the De La Salle University-Angelo King Institute for Economic and Business Studies (AKI Research Grants 2020–2021 Project No. 500-139), and a Research Fellowship to the first author from the National Academy of Science and Technology, Philippines.

Author information

Authors and affiliations.

De La Salle University, Manila, Philippines

Allan B. I. Bernardo, Macario O. Cordel II, Marissa Ortiz Calleja, Jude Michael M. Teves, Sashmir A. Yap & Unisse C. Chua

You can also search for this author in PubMed   Google Scholar

Contributions

Conceptualization: ABIB, MOC II, MOC; Machine learning methodology, MOC II; Machine learning modeling and evaluation: JMMT; Data preprocessing and feature engineering: SAY; Data visualization: UCC; Writing—original draft preparation, review, and editing: ABIB, MOC II, MOC; Project administration: MOC II, MOC; Funding acquisition: MOC, MOC II, ABIB.

Corresponding author

Correspondence to Allan B. I. Bernardo .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Ethics approval

The study involved secondary analyses of the officially published PISA 2018 dataset; as such ethics review and approval and informed consent does not apply. This dataset was downloaded as a public use file from the OECD website ( https://www.oecd.org/pisa/data/2018database/ , accessed 17 Feb 2020).

Informed consent

This paper involved no primary data collection on the part of the authors. However, PISA 2018 data collection procedures indicate that parents of the students who participated in the assessment provided informed consent.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary file, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Bernardo, A.B.I., Cordel, M.O., Calleja, M.O. et al. Profiling low-proficiency science students in the Philippines using machine learning. Humanit Soc Sci Commun 10 , 192 (2023). https://doi.org/10.1057/s41599-023-01705-y

Download citation

Received : 12 September 2022

Accepted : 18 April 2023

Published : 03 May 2023

DOI : https://doi.org/10.1057/s41599-023-01705-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Improving quality of basic education becomes a priority national concern

For more than two years, grade school and high school students could not attend face-to-face classes and had to be taught through a combination of distance learning methods that were extremely challenging for them, as well as for their teachers and parents.

The Covid-19 pandemic wiped out the painstaking gains in improving the quality of education achieved over the previous decades. In September 2021, the National Economic and Development Authority (NEDA) estimated that one year of shutdown would result in ₱11 trillion in productivity losses over an individual’s 40-year lifespan as a member of the labor force.

As the Marcos administration endeavors to put the educational system back on track, it must deal with the continuing lack of classrooms and basic facilities. Senator Sherwin Gatchalian, Senate education committee chair, cited the ₱430 billion budget allocated for meeting the shortage of 167,901 classrooms nationwide based on the 2019 National School Building Inventory.

An even more basic problem is ensuring universal access to basic education. Thanks to the government’s integrated development approach – as exemplified by the Pantawid Pamilyang Pilipino Program – children from the neediest families were able to enroll and sustain school attendance through a combination of nutrition, maternal health and livelihood programs augmented by direct cash grants.

The closing of private schools during the pandemic aggravated the overcrowding problem in public schools. Quality of education also affected the teacher-to-pupil ratio. In 2017, there were 29 high school students per teacher; the ratio was 26:1 in the primary grades – the highest in the region.

Vice President Duterte, who is also the DepED secretary, deplores that Filipino students are “not academically proficient,” emphasizing the need to refocus the curriculum toward raising aptitude levels in reading, writing, and mathematics. Latest surveys show that the Philippines as the tailender among ASEAN countries, behind Vietnam, Cambodian, and Myanmar based on standardized testing scores.

A good place to start would be to increase the country’s education spending to approximate the United Nations’ standard of six percent of GDP. From a peak of almost four percent in 2017, it dropped to three percent. From 16 percent of the national budget, it dropped to 14 percent.

Speaking at the summit, President Marcos said: “We have failed our children and let us not keep failing them anymore. Otherwise, we will not allow them to become the great Filipinos that we know they can be.” He vowed: “We will build infrastructure that will provide our learners, teachers, and the entire academic sector with a healthy and safe environment that is conducive to learning.”

By providing an email address. I agree to the Terms of Use and acknowledge that I have read the Privacy Policy .

12-year data track poor training of PH teachers

INQUIRER.net stock photo

The quality of education that future teachers are getting needs a lot of improvement as more than half of the schools training them fare poorly in the annual licensure exams for educators, a private sector-led advocacy group said on Thursday.

The Philippine Business for Education (PBEd), founded and financed by the country’s top business leaders, highlighted in a media briefing the need to upgrade teacher quality in the country given the direct correlation of teaching to the performance of students.

PBEd executive director Justine Raagas cited the 2022 research done by the government’s socioeconomic-policy think tank Philippine Institute for Developmental Studies (PIDS), which stated that low teacher qualification was a major factor in the low-quality education and poor performance of students.

A team of PBEd researchers analyzed 12 years’ worth of licensure examination for teachers (LET) data from the website of the Professional Regulation Commission (PRC) and merged these with other data points from the Commission on Higher Education (CHEd).

Based on their findings, Raagas said that 56 percent of the schools nationwide offering teacher education had below-average passing rates in the licensure exam since 2010.

“Poor performance (in LET) puts teacher quality into question,” Raagas said.

Diane Fajardo, PBEd deputy executive director, said they had done a similar study back in 2013 which yielded the same results after nine years.

“Nothing changed … [the] PRC never tracks [the performance] so hopefully, we can turn this over to them and then yearly they can track it,” she said.

The study also found that the overall passing rate of LET examinees was lower compared to that of other courses, such as architecture, nursing, civil engineering and accountancy.

Moreover, only 2 percent of schools offering teacher education were classified as high-performing, PBEd said, noting that the government should consider closing down the education programs of low-performing schools.

The study defined high-performing schools as those with LET passing rates of at least 75 percent.

Complacency

Of the 111 schools under the classifications of Center of Excellence (COE) and Center of Development (COD), more than 81 percent of COEs and 91 percent of CODs were not high performing in the licensure exams.

“Contrary to expectations … high passing rates do not seem to be sustained by all COEs and CODs,” the study said.

One of CHEd’s criteria to accredit schools as COE or COD was to rank in the top 10 and top 20 in the licensure exams, respectively, for three consecutive years.

“By being accredited, COEs and CODs are already assumed to have met the standards imposed by CHEd, which might have encouraged complacency,” the research said.

However, these institutions were not able to sustain high passing rates, it said, citing Zamboanga City State Polytechnic College, a COD that recorded an average overall passing rate of only 13.8 percent for secondary teacher licensure exams from 2010 to 2022.

The results also suggested that first-time takers had higher passing rates than repeaters.

“Alternatively, this means that there is a higher likelihood of failure among those who retake the exam,” the study said.

Regional disparities

It further noted the regional disparities in passing rates, wherein institutions from Luzon performed better compared to their counterparts in Mindanao that had the lowest regional passing rate, driven by the extremely low score of schools from the Bangsamoro Autonomous Region in Muslim Mindanao (BARMM).

It said that 45 percent, or 98 out of 217 schools in Mindanao, were “low-performing,” while first-time takers in the BARMM performed “just as bad as repeaters.”

Action plans

“We want to improve teacher quality both in preservice, or while they are still studying, and in-service, when they are already teaching,” Raagas said.

PBEd policy and advocacy manager Jose Andoni Santos laid down their policy recommendations, which included focusing on building more and maintaining COEs per region and strictly monitoring their performance.

Santos said that the curriculum of teacher education and questions in licensure exams must also be reviewed in relation to the Department of Education’s professional standards for teachers.

He added that the government could also consider closing down the education programs in low-performing schools, particularly those that consistently performed poorly year after year, and implement a “three-strike rule” for repeaters.

Santos also highlighted the urgent need to develop and publish data that would help measure the performance of schools for those taking up education.

PBEd officials expressed hope that their study could act as a catalyst to improve teacher education in the Philippines.

“Our teachers are heroes but we just can’t expect them to be resilient and to push on despite the lack of support. We have to give them the proper support, ensure that they will possess the skills that they will need and that they will have the materials that they need,” Santos said.

He also noted that PBEd’s study showed only a small part of what was needed to improve teacher quality.

Subscribe to our daily newsletter

“There’s really a lot of teacher quality interventions needed and we hope that this is the start,” he said.

News that matters

Disclaimer: Comments do not represent the views of INQUIRER.net. We reserve the right to exclude comments which are inconsistent with our editorial standards. FULL DISCLAIMER

© copyright 1997-2024 inquirer.net | all rights reserved.

We use cookies to ensure you get the best experience on our website. By continuing, you are agreeing to our use of cookies. To find out more, please click this link.

• Share full article

Advertisement

Supported by

Philippines Closes Schools as Heat Soars to ‘Danger’ Level

Scorching temperatures coincided with a nationwide strike of jeepneys, the main mode of public transport in the country.

By Jason Gutierrez

Reporting from Manila

The Philippines closed all public schools on Monday and Tuesday because of dangerously high temperatures, moving classes online in a country where schools are typically shut because of tropical storms.

Over the past week, average temperatures in many parts of the country topped 40 degrees Celsius, or more than 100 degrees Fahrenheit. Extreme heat is forecast this week to blanket almost the entire country, with the heat index in some regions rising to at least 42 degrees Celsius, or “danger” level, according to the Philippine Atmospheric, Geophysical and Astronomical Services Administration. That designation is the second highest on the agency’s heat index scale. It advised people to avoid exposure to the sun or risk heat stroke, heat exhaustion and cramps.

In metropolitan Manila, where the heat index is forecast to hit 45 degrees Celsius early this week, residents in overcrowded slums have been cooling off by setting up colorful inflatable pools on busy roads. Others in this megacity have been dipping into Manila Bay, flouting rules that prohibit swimming in its polluted waters.

In its advisory on school closures, the Department of Education on Sunday said the extreme weather coincided with a nationwide strike of jeepneys, the colorful, open-air vehicles that are the main mode of public transportation in the Philippines. Jeepney drivers are protesting a government plan to phase out their rides — which trace their origins to U.S. military jeeps — and replace them with modern, more energy-efficient minibuses.

The extreme heat had already forced some schools to cancel classes before the government’s call for closures. The Jesus Good Shepherd School in Imus, a city south of Manila, last week sent students back home because of soaring temperatures, even though the private institution is among the small minority of schools in the country that has an air conditioner in every classroom.

“It is hard for the students and teachers alike to concentrate, because the air-con is struggling, too,” said Ana Marie Macarimbang, a fifth-grade teacher at the school who has taught for nearly two decades. “We are in a tropical country, yes, but the heat now is more intense than I can remember.”

Weather-related school closures in the Philippines have historically been more common during the typhoon season, which peaks between July and October . The current closures, teacher’s groups have contended, could have been avoided had the authorities not changed the school calendar after the pandemic. The school year now runs from August to May, roughly, rather than the former June-to-March schedule.

President Ferdinand Marcos Jr. has said that he has no objections to readjusting the school calendar, and blamed climate change for the extreme heat. The government “really didn’t expect it to be like this,” Mr. Marcos said earlier this month.

Extreme temperatures are also disrupting everyday life in other parts of Asia, including Cambodia and Vietnam . Earlier this month, a heat wave forced schools in Bangladesh and India to close.

Explore Our Weather Coverage

Extreme Weather Maps: Track the possibility of extreme weather in the places that are important to you .

Tornado Alerts: A tornado warning demands instant action. Here’s what to do if one comes your wa y.

Flash Flooding: Fast rising water can be deadly. Here’s what to do if you’re caught off guard , and how to prepare for a future flooding event.

Evacuating Pets: When disaster strikes, household pets’ lives are among the most vulnerable. You can avoid the worst by planning ahead .

Climate Change: What’s causing global warming? How can we fix it? Our F.A.Q. tackles your climate questions big and small .

Heat wave in Southeast Asia closes schools, triggers health alerts

• Medium Text

DEHDRATION, HEAT SHOCK

Sign up here.

Reporting by Neil Jerome Morales in Manila; Panarat Thepgumpanat in Bangkok, Khanh Vu in Hanoi, Xinghui Kok in Singapore, Danial Azhar in Malaysia and Kate Lamb in Jakarta; Editing by John Mair and Raju Gopalakrishnan

Our Standards: The Thomson Reuters Trust Principles. New Tab , opens new tab

World Chevron

Lithuanian presidential hopefuls vow to stand up to Russian threat

Lithuanians vote on Sunday in a presidential election expected to hand a new term to incumbent Gitanas Nauseda, a staunch supporter of Ukraine in its two-year war with Russia, following a campaign focusing on security concerns in the Baltic states.

• Subscribe Now

education in the Philippines

The green report: it’s getting hotter – now what.

PMA Top 10 over the years: How a harmless event became a cause célèbre

UP Mindanao bares medicine, other new programs in Davao by 2025

Philippines will not use water cannons against Chinese vessels – Marcos | The wRap

Marcos says gov’t eyeing shift to old academic calendar in 2025

DLSU releases DCAT 2024 results

[Rappler Investigates] It’s too darn hot!

In Philippine classrooms, weather’s too hot to handle

DepEd wants to fast-track return to old academic calendar by March 2025 | The wRap

‘Aggressive’ revert to old academic calendar eyed for school year 2024 to 2025

Checking your Rappler+ subscription...

Upgrade to Rappler+ for exclusive content and unlimited access.

Why is it important to subscribe? Learn more

You are subscribed to Rappler+