In these early stages, researchers are able to describe what is happening but cannot get into the objective specifics of how fast, how much, and other metrics with certainty. For example, a graduate student may have observed a relationship between the timing of lectures and student participation. But without running an objective test to determine whether this relationship exists, what they have is conjecture and possibly a hypothesis. As such, they cannot use this to make any recommendations to their institution, as they would have no basis for this, save for their own observations.
Luckily, researchers are able to get past these initial arguments and establish theories through quantitative modeling. Let us go through how these tools fill the gap between qualitative observations and objective findings across different fields.
What Gaps Does Quantitative Modeling Fill in Academia?
Quantitative modeling uses mathematical, statistical, and computational formulas to depict real-world settings, which enables researchers to derive insights from raw data that would otherwise not have been apparent. The models vary from statistical to financial and predictive models. But regardless of their setup, these tools serve various roles in academia, from analyzing trends to assessing risks. Most importantly, they address the following research gaps:
The Influence of Coincidence
Naturally, we are all subject to apophenia, which is the tendency to find patterns even when they do not exist. This subjectivity stems from our innate desire to seek order, but while this is a good thing, it can result in false conclusions. For example, a researcher may note that the divorce rate in Chicago aligns with the per-capita consumption of milk in the past ten years. And thanks to apophenia, they may assume that one causes the other and can come up with a hypothesis that states as much.
Given that apophenia can result in the misinterpretation of data or accidental p-hacking, where researchers manipulate data to find patterns, it is important for them to use quantitative models. These tools enable them to test the mathematical probability that a correlation happened by pure chance.
As such, they are able to not only establish existing relationships but also calculate the probability of an event based on the available data. In this way, their statements change from ‘It looks like X causes Y’ to ‘We are 95% confident that X causes Y.’
The Need for Isolation
While theories often assume that events take place in a vacuum, the reality is that in the real world, this could not be further from the truth. For example, on paper, a new education policy may look like it will boost students’ test scores because it will allow them to enjoy more tutoring. But in reality, this will not always translate into the desired results. After all, out in the world, we have to think about aspects such as school funding, the socioeconomic status of the students, the experience of the teachers guiding the students, and so on. And thanks to these nuances, the effect of the policy on student A could be quite different from that on student B.
But that is where quantitative models come in. You see, when using logic or just observation, researchers cannot isolate the impact of other factors on an existing relationship. For instance, with our education policy and test scores, we cannot simply isolate teacher experience by observation. But with a quantitative model, we can.
Researchers are able to mathematically hold every other factor constant so that they can isolate the impact of a single variable within the entire system. It is only in this way that they are able to tell whether their efforts will bear fruit, given the strength of the relationship between the key dependent and independent variables. They are also able to study the effects of the other variables, such as teacher experience and school funding, by including them as mediators and moderators in the relationship.
The Role of Prediction
Descriptive research plays many important roles in our society as it describes things as they are. For instance, a researcher may compile data on the unemployment rates in Atlanta over the past five years. So, they would map the trajectory from 3.9% in 2021 to 3% in 2022, 3.2% in 2023, 3.4% in 2024, and 3.4% in 2025. This information would enable them to write a report on this and possibly include the factors contributing to the fluctuation of this rate, including why it was so high in 2021.
But these numbers only tell us what has happened. Academia thrives on seeking solutions for the future. So, instead of researchers focusing only on the past, they may also want to look forward, and this is where quantitative models would play a part in their work. How?
Well, these models are able to extract patterns from historical data and use the same to simulate future scenarios. Thanks to this, researchers are able to predict the future, regardless of whether they are studying economic market crashes or consumer purchasing power. These predictions are vital to various stakeholders in society who can use this data to make necessary changes, e.g., investors who read predictive reports on a possible market crash can rearrange their asset portfolios to minimize their risk exposure.
The Scaling Power
In the world of research, what is good for the goose is not always good for the gander. Take altruistic punishment, for example. Many experiments conducted on this have shown that human beings are often willing to give up their own financial gain so as to punish someone who they deem to have acted unfairly. But when researchers replicate the same experiments in smaller nomadic communities, they find that this social theory is ineffective at predicting people’s behavior. As such, while a qualitative case study may give researchers insights into how one community behaves, they cannot assume that this will apply to other communities or even an entire nation.
While that may be so, there are many cases where what works for one community may work in another. A good example of this would be the trials conducted in the UK, which proved that small and low-cost approaches to public communication improved compliance rates. Other countries borrowed the idea and started implementing it in their own communications to boost participation in public programs such as public health outreach and environmental conservation. Even private organizations around the world have adopted the same techniques due to their high effectiveness.
But how can researchers tell what can carry over to another region successfully? They rely on quantitative models, which allow them to take sample data, verify the representation mathematically, and then scale the findings in order to make statements that can apply to larger populations who meet the set criteria. This approach bridges the gap between one community and the world as a whole while also ensuring that researchers account for key nuances that could affect the success of the programs.
The Role of Quantitative Modeling in Research
Beyond what we discussed in the previous section, quantitative models also fill the gap between theory and falsifiability. You see, before researchers can accept a theory as scientifically valid, it must be falsifiable, which is to say that there must be a way to prove it wrong. After all, when a theory is vague, this flexibility allows the researcher to change it to fit the outcome they desire. But if they translate this theory into a quantitative model, they must define its parameters such that if the model’s predictions do not match the empirical data, the theory can be debunked objectively during the verification process.
Thanks to this, quantitative modeling is used for the following reasons in academia:
To Train Graduates
Graduate students enroll in academic institutions with the ability to review existing literature, as they usually have done so during their undergraduate studies. But as graduates, they are expected to move from just consuming research and instead learn how to analyze datasets and come up with statistically sound findings that add to the existing body of research. Their professors thus teach them how to use quantitative models.
To Test Hypotheses and Defend Dissertations
Researchers and graduate students must substantiate any claims they make, be it in journal articles or their own dissertations. Otherwise, their work cannot be accepted for publishing as it would be deemed as anecdotal. Quantitative models enable these researchers and students to test their hypotheses against real-world data to determine whether there is any objectivity to their claims. Moreover, they enable them to provide a framework through which other researchers can replicate the same research and verify the conclusions, which is integral in dissertation panels and journal peer reviews.
To Seek Funding
Academic research is not only time-consuming but also costly. Therefore, academic institutions often rely on external organizations to fund their research, both in terms of personnel and tools. But these organizations will rarely grant money to researchers based on hunches and will often want to see rigorous research designs that demonstrate the feasibility of the proposed research projects. Quantitative models allow researchers to do this.
For example, researchers can use a power analysis model to prove the number of subjects they need to sample to get a statistically significant result, which would serve as proof to the funding body that the requested amount was required to complete the project. In other cases, researchers use previous findings from smaller sample sizes to demonstrate the impact of a larger project. Either way, the models make the case for the funding.
To Collaborate with Other Fields
While universities may share common missions and visions, they are often highly compartmentalized such that different fields focus on different aspects. Even so, you find that these departments often work together on groundbreaking research. But how are they able to communicate effectively, given the differences in how they look at the world? Quantitative models!
You see, while researchers from different fields may have different outlooks, they understand how research models work, and as such, these models act as a bridge that allows departments to work on the same study. After all, the models standardize the data formats and analysis techniques, and with these out of the way, the researchers can focus on the output.
Quantitative Modeling in Different Fields
While quantitative modeling is used in various fields in academia, you find that it is heavily used in the following departments for the most part due to their unique research challenges:
Economics
Research in economics centers on relationships at the micro and macro levels, which are often influenced by factors such as government policies and market forces. As such, researchers often run into the challenge of endogeneity and reverse causality, where they run the risk of confusing causality and effect. Moreover, they grapple with unobserved heterogeneity where missing background factors, such as regional work ethics, can skew the results.
Economists rely on R, Python, and Stata for the most part. R, for example, enables them to run fixed-effect regressions to account for unobservable traits in the datasets by adding a dummy variable for every subject.
Political Science
Quantifying aspects such as voting behaviors and state interactions is not an easy task. To further complicate this research, researchers in this field often encounter challenges such as confounding factors and strategic selection biases that may not always be readily observable. For instance, wars and treaties are often not random and are instead the result of self-selection based on factors that researchers may not observe readily.
Like economists, political scientists rely on Python and R to account for these biases. Logit and probit regressions, for instance, which are often used in R and Stata, account for the fact that political outcomes are often binary and allow researchers to break down the possibilities while providing the maximum likelihood estimations for binary choices.
Finance
When dealing with volatile markets, finding the structural order is quite the undertaking. And when you account for possibilities such as volatility clustering, non-constant variance, and asset mispricing, the number of research challenges increases. A good example of this would be in time series data, where the variance of errors is not constant over time. Big shocks can follow other big shocks, and this would defy the assumptions in standard linear regressions and would thus make the standard error calculations incorrect.
With tools like generalized autoregressive conditional heteroscedasticity (GARCH), which are available in R, researchers are able to model variance as a dynamic variable. Python is also frequently used in finance research, especially when cleaning huge datasets or when researchers want to run Fama-French multi-factor regressions that can control for specific market anomalies like size.
Social Sciences
Whether researchers are delving into education, sociology, or even demographics, they are always uncovering relationships that exist within our society. Given the complexity of our societal structures and systems, researchers find themselves juggling challenges such as latent constructs and hierarchical nesting. Take the example of the hierarchical nesting, which results from our natural drive to group based on shared characteristics. If researchers fail to account for this when conducting research, they can produce distorted statistical results that would not be replicable in another environment without similar nesting.
Seeing as researchers cannot measure abstract ideas like social cohesion or ignore the groupings in society, they must use statistical tools that can handle such variables. Examples include structural equation modeling with confirmatory factor analysis, which analyzes the shared variance among survey items to separate the underlying signal from individual measurement error before running the causal paths. Others include mixed effects models and hierarchical linear modeling, which are available in R and Python.
Usually, departments use one or two of the following software: R, Python, Stata, and SPSS, based on the size of the datasets and the required insights.