Introduction
Online searches greatly impact student decisions. In 2019, Studyportals saw 36 million prospective students compared to 5 million enrolled globally. If one considers the correlation between digital programme searches and enrolment rates, rated from +1, to -1, to gauge this relationship, what is the link between the two variables?
Correlation measures the strength of association between two variables and the direction of the relationship. A value of +1 indicates a perfect association in the same direction. In contrast, a value of -1 indicates a perfect association in the opposite direction. Values closer to 0 indicate a weaker association with a value of 0 indicating no association. It is important to note that there is a difference between causation and correlation, which are distinct concepts in statistics and research. Causation suggests a direct cause-and-effect relationship between two variables, where changes in one variable could lead to predictable changes in the other. In contrast, correlation indicates a statistical association between variables, showing that they tend to move together in some manner but without implying causation.
While causation requires controlled experiments, correlation can be quantitatively measured but does not reveal the underlying reasons or direction of the relationship. Thus, causation cannot be assumed solely based on observed correlations, as other factors may be involved.
Over the years, Studyportals ACT has conducted several studies to observe the correlation between pageview traffic registered across our portals and enrolment data made available by a selection of public datasets. Doing so, each study showed the correlation between pageviews and enrolment on a country level by using Spearman’s rank correlation coefficient (ρ) to determine the relationship’s strength. This correlation ranges from “very strong” (ρ between ±0.80 and ±1.0) to “very weak” (ρ between ±0.00 and ±0.19).
Summary of key takeaways
Studyportals ACT has successfully examined and confirmed the strong correlation between the pageview data collected from Studyportals’ websites and public datasets related to enrolments and commencements in various destination markets. These comprehensive investigations have yielded the following outcomes:
- Studyportals pageviews and OECD/Eurostat/UNESCO enrolment data (2017): ρ = 0.85
- Studyportals pageviews and HESA enrolment data (2018): ρ = 0.79
- Studyportals pageviews and HESA enrolment data (2021): country-level (ρ = 0.92) and institution-level (ρ = 0.83)
- Studyportals and US enrolment data (2022): Ohio (ρ = 0.80), New York (ρ = 0.76), Texas (ρ = 0.79), Pennsylvania (ρ = 0.78), and Massachusetts (ρ = 0.77)
- Studyportals pageviews and US enrolment data (2023): ρ = 0.80
- Studyportals pageviews and Australian commencements (2023): ρ = 0.80
- Studyportals’ composite (private) enrolment dataset and UNESCO enrolment data (2023): UK (ρ = 0.91) and the Netherlands (ρ = 0.81)
Based on these studies, the correlation between Studyportals pageviews and student enrolments generally stands at ρ = 0.80 (or 80%), indicating that the relationship between these two variables is both positive and strong. Interestingly, the more student traffic Studyportals has from a specific source country, the more reliable the relationship between pageviews and the rank correlation coefficient is.
For instance, the correlation between pageviews and enrolments for the US (see Figure 5) revealed that France, Mexico, Colombia, Canada, and India had strong correlations, which is likely influenced by the large amount of pageviews generated by each source country. In relation to this, the analysis investigating the correlation between Studyportals private dataset with public datasets showed that study destinations with robust education systems, such as the UK and the Netherlands, were more likely to have stronger correlations, emphasising the importance of sample size to gain accurate results.
Lastly, the correlation analysis conducted between pageviews and HESA enrolments in 2021 yielded results that suggest the strong relationship between these two variables is not likely to be due to random chance. In other words, the strong connection between pageviews and enrolments appears to be a meaningful and genuine association rather than a coincidence.
Correlation studies
While a single pageview may not tell us much, when we gather millions of pageviews related to factors such as programmes, fields of study, or countries of origin, this becomes a robust benchmarking tool to compare interest against institutional, disciplinary, or country averages. By analysing this interest data over time, we can get hints about future growth possibilities and how current events and developments affect student recruitment.
What makes this student interest indicator particularly useful is its timeliness. Unlike other indicators like enrolment data, it is available in real time. Moreover, prospective students typically start researching their programme options 6-24 months before enrolling, with an average of around 15 months. Therefore, this indicator offers insights into the interest that may translate into enrolments 1-2 years later.
Using data from (inter)national enrolment databases poses a significant challenge because it has been inconsistent and limited, making it tough to compare across various providers. However, Studyportals has confirmed most of the assumptions and continues to research to find more matches and explore any unusual data points. It is important to note that for each research study outlined below, Studyportals ACT investigated the correlation between pageview traffic for Master’s programmes listed on Studyportals’ website and enrolments for Master’s programmes available in the selected public datasets.
List of correlation studies
Correlation between Studyportals pageviews and HESA enrolment data (2018) – Discipline level
Correlation between Studyportals pageviews and HESA enrolment data (2021) – Country-level
Correlation between Studyportals and US enrolment data (2022) – Institution level
Studyportals pageviews and US enrolment data (2023) – Discipline & country-level
Studyportals pageviews and Australian commencements (2023) – Discipline and country-level
Correlation between Studyportals pageviews and OECD/Eurostat/UNESCO enrolment data (2017) – Country & discipline level
On average two million visitors visited the Studyportals website monthly in 2017. The data generated by these users gave valuable insights into student preferences, but Studyportals lacked a way to confirm them. To validate this data, Studyportals ACT used public data on international mobile student enrolments for ‘ ‘Master’s programmes from three organisations: UNESCO, OECD, and Eurostat. However, due to varying data quality and availability within the public datasets from these organisations, some countries were excluded from the research. Austria, Belgium, Denmark, Finland, Germany, the Netherlands, Sweden, Switzerland, and the United Kingdom were included in the study.
The study verified Studyportals’ visitor data for Master’s programmes by comparing it with past enrolments in the same field, with the result showing that data from Studyportals closely matched actual enrolments in these countries, showing a strong correlation with a Spearman’s correlation coefficient of ρ = 0.85 (see Figure 1).
Furthermore, it is interesting to note the outlier in the upper left corner of the figure, which represents Medicine programmes offered in Belgium. Many international students, particularly Dutch students, enrol in this field due to the restrictions on medical student numbers in the Netherlands. Interestingly, Dutch students often do not use Studyportals to search for these programmes but instead rely on word-of-mouth recommendations, as Belgium is a well-known alternative for students who do not get accepted for Medicine programmes in the Netherlands. This, as such, highlights some of the disparities that can exist in student choices.
Figure 1: Correlation between pageviews and OECD, UNESCO, and EUROSTAT enrolment data
Correlation between Studyportals pageviews and HESA enrolment data (2018) – Discipline level
Online search has become integral to our decision-making process, from shopping to choosing where to study. With the global rise in smartphone usage, the line between digital and offline decisions is blurred. Young adults increasingly start their study searches online, raising the question: Can online search behaviour predict future enrolments?
To investigate this question, Studyportals analysed enrolment data for Master’s programmes from the Higher Education Statistics Agency (HESA) in the UK. The results showed a strong correlation (ρ = 0.79, or 79%) between online pageviews for ‘ ‘Master’s programmes and international student enrolments for a Master’s in the UK (see Figure 2). Disciplines like Business & Management exhibited a significant link between online interest and actual enrolments. In 65% of source countries for UK Master’s degrees, Business & Management was the top choice, matching pageview data in 68% of countries of origin.
Figure 2: Rank correlation between pageviews and HESA enrolment data (2013-2017)
Correlation between pageviews and HESA enrolment data for the destination the United Kingdom. Each dot represents a discipline.
This showed that online search behaviour, as observed on platforms like Mastersportal, can reliably predict future international student enrolments, particularly in specific disciplines and source countries. This highlights the growing impact of digital platforms on educational decisions worldwide.
Correlation between Studyportals pageviews and HESA enrolment data (2021) – Country-level
A more recent correlation report showcased the strong link (ρ = 0.92) at the country level and also at the institution (ρ = 0.83) between Studyportals pageviews and HESA enrolment data for Master’s programmes (see Figure 3). The fact that enrolments and pageviews are correlated on the institution, country, and discipline level means that these correlations are likely not spurious, meaning the connection between the two variables does not exist by chance. As a result, this connection offers valuable insights for higher education decision-makers, trend analysis, and accurate predictions.
Figure 3: Rank correlation between pageviews and HESA enrolment data (2019/2020)
Correlation between Studyportals and US enrolment data (2022) – Institution level
In 2022, Studyportals ACT also researched whether there was a strong link between enrolment data for Master’s programmes offered by US institutions and Studyportals pageviews for a Master’s in the US. The reason behind this investigation arose from the fact that the United States lacks a single place for higher education enrolment data, with each state having its department that reports these numbers. Thus, the research examined the connection between pageviews and enrolments in five populous states: New York, Ohio, Texas, Pennsylvania, and Massachusetts.
More specifically, Figure 4 displays that the correlation in Ohio (ρ = 0.80) is “very strong,” while the correlations for New York (ρ = 0.76), Texas (ρ = 0.79), Pennsylvania (ρ = 0.78), and Massachusetts (ρ = 0.77) are considered to be “strong.”
Figure 4: Correlation between pageviews and enrolments for institutions in New York, Ohio, Pennsylvania, Texas, and Massachusetts
Studyportals pageviews and US enrolment data (2023) – Discipline & country-level
In relation to the previous correlation, Studyportals ACT recently conducted another study to analyse the relationship between pageview data for Bachelor’s and Master’s programmes collected in 2020 and US enrolment data from Open Doors and based on 24 origin countries for the academic year 2021/2022. The study correlates data for 2020 from Studyportals with 2021 US data because it is estimated that students search for programmes on Studyportals between 6 to 18 months before enrolling at a university. Therefore, 2020 pageviews should correlate with 2021 enrolments.
The study used Spearman’s Rank Correlation, not Pearson’s, because the data lacked normal distribution. Spearman’s assesses the relationship between rankings, focusing on discipline order rather than the specific pageview or enrolment numbers. We calculated discipline correlation per country, weighting each by pageviews. Following this, we found the weighted average correlation across all countries, which revealed the positive influence of countries with higher pageviews. Please see the Appendix section for a more detailed explanation of the methodology.
As shown in Figure 5, the correlation analysis showed that Studyportals pageview data is representative of international student enrolments per discipline in the United States, with a strong weighted correlation average of ρ = 0.80 (80%). Interestingly, the top 5 countries with the strongest correlation include France, Mexico, Colombia, Canada, and India.
Figure 5: Correlation between US enrolments and Studyportals pageviews (2020-2021)
Studyportals pageviews and Australian commencements (2023) – Discipline and country-level
ACT also investigated the correlation between Studyportals pageviews for ‘ ‘Master’s programmes between 2019-2020 and commencement data for Bachelor’s, Master’s, and PhD programmes in Australia from the dataset by Austrade for 2020 and 2021. The data for the investigation was based on the top 100 countries according to the number of pageviews generated, ensuring that a credible sample size was being used.
Once again, Figure 6 indicates that the correlation analysis resulted in a “very strong” Spearman’s correlation of ρ = 0.80 (80%), meaning that Studyportals pageview data is representative of commencement in Australia. The results also highlighted a particular trend, in that the more search traffic generated by a particular origin country, the more reliable the pageview data and commencement correlation would be.
Figure 6: Correlation between Australian commencements and pageviews (2019-2021)
Correlation between Studyportals’ composite (private) enrolment dataset and UNESCO enrolment data (2023) – Country-level
As previously mentioned, Studyportals provides a comprehensive database of English-taught tertiary programmes to help students find suitable study options based on their interests and qualifications. Student enrolments form a part of this database, and to validate this part of the database, Studyportals conducted correlation research with the public enrolment dataset made available by UNESCO and explored the impact of factors such as commercial bias, differing data sources, and evolving data standards.
Doing so, Studyportals sought to provide students with more precise information about available study options and enhance Studyportals’ credibility. Once again, Spearman’s correlation was used to investigate how the rankings of variables were related to each other in relation to Master’s programmes, not their exact values, and helped us understand the relationship between variables based on their order.
By combining correlation rates and sample sizes, a better understanding of the link between private and public enrolments was achieved, helping to identify meaningful correlations while considering data reliability.
This dataset revealed insights into the strength of the relationship between private and public enrolments, considering both correlation rates and sample sizes. The analysis showed that larger countries with substantial education systems tended to have stronger correlations, underlining the importance of sample size for an accurate interpretation. Below are two examples to show the result from the Spearman correlation for the UK (see Figure 7) and the Netherlands (see Figure 8).
Figure 7: Rank correlation for destination UK
In 2019, the UK was found to have a correlation coefficient of ρ = 0.91 (91%).
Figure 8: Rank correlation for destination Netherlands
In 2019, the Netherlands was found to have a correlation coefficient of ρ = 0.81 (81%).