Predicting Students’ Dropout and Academic Success

The Dataset:

This dataset analyzes student academic outcomes. It was designed to help educators understand patterns within student performance by focusing on graduation, enrollment, and dropout rates. It is currently used to predict student outcomes and examine demographic variables, like gender, to inform decisions. In use, it would be helpful for intervening with a student predicted to drop out and bring about more educational equality.

Project Statement:

The dataset, Predicting Students Drop Out and Academic Success, was designed to identify factors that would impact a student's risk of dropping out. The dataset defines gender as male and female. What is the implication of such a narrow definition of a largely impactful socio-environmental factor within academic successes?

Research Elements:

Reading Databases Element: structural equality

This element will help address how a system with gender binary influences or enforces a binary within the education system.

Formal Anatomies Element: connotative reading

(Assesses the context of the data and how constantly shifting definitions for the dependent and independent variables impacts the represented data). This element will help to analyze how the meaning of "gender" has changes over time and been applied in various contexts.

Connotative reading: Gender vs. Sex:

The dataset approaches the definition of gender as gender essentialism, meaning it only accounts for two binaries, female and male. If we change the definition to fit more modern contexts, then it would be through a gender constructionist view, which makes room for more genders, more experiences, and widens the data pool to a larger sample for higher accuracy. Though the dataset defines gender as male and female, that definition is more adequately fit to the term sex. If we define this term separately from gender, it changes the data's meaning, as sex pertains to more biological aspects, and gender to social roles, expression, and constructs.

Dataset Analysis:

• Out of 1556 males: 33% graduated, 20% enrolled, and 45% dropped out
• Out of 2868 females: 58% graduated, 17% enrolled, and 25% dropped out

The drop out statistic is much higher than compared to US and EU averages which categorize by gender:

• US women: 4.3%
• US men: 6.3%
• EU women: 7.7%
• EU men: 11.3%

The exclusion of transgender, non-binary, and intersex students can skew datasets. We researched the rates for these students who were not accounted for and found that in one school year, 16% of transgender students failed to re-enroll. 32.4% of transgender students seriously considered dropping out, and in Australia, 19% of intersex students dropped out of secondary school, compared to the country's average of 2%.

Real World Examples:

We researched some active examples of how incorrectly defining gender and refusing to include others such as transgender, intersex, and non-binary people impacts datasets. Some prominent ones are: 

Collection of Covid-19 infection and mortality rates: 

Using only biological sex makes it difficult to determine if disparities in infection/mortality is due to biological or social factors

Drug trials and treatments: 

Drug trials usually focus only on biological males, which not only excludes the female sex but also excludes intersex people and non-cisgender individuals

Crime statistics:

Analysis on crime rates and stats are usually done with only biological sex, which neglects gender-based statistics that could predict crime patterns or provide insight to non-cisgender criminals and victims 

Employment and further education data: 

Many disparities in employment and education are gender based, not just sex based, which is not reflected when data analysis is done using only biological sex 

Explanation and Examples, Structural Equality:

Structural inequality refers to a systematic disadvantage, in which certain groups face challenges and unequal access to opportunities. This inequality doesn't come from individual biases but from society, its institutions, policies, and norms. Creating structural equality starts by looking at these institutions, policies, and norms to shape equal outcomes for all groups. 

When looking at the data set, gender only being male and female leaves out a broad group of individuals when looking at student dropout and success rates. 

Explanation and Examples, Connotative Reading:

Connotative reading refers to an interpretive reading of a datasets meaning, it is how the meaning has changed over time, because of what social interests, and in what context. 

The meaning of gender has changed significantly over time, changing from a biological and binary understanding to a more diverse spectrum of gender identities and expressions. This change has been a result of a broader cultural shift to be more inclusive and give more recognition non binary identities. 

Connection:

This all collides under the theme of bias & equality. Through connotative reading, we see how overtime as contexts change, bias develops in lieu of different definitions. This can cause education inequality when students are assessed only through one definition or lens. Through structural equality, we also see how education inequality forms. As long as there is inequality in how we conduct research and collect data by being biased towards and excluding people, the data will not be accurate. 

Conclusion: 

Reading our dataset elements shows how important context and definition is when collecting data, without proper and inclusive context and definitions it can leave out a big part of the population, affecting the results of data.