Exploring the Gen Z Paradox: A Report

Introduction

Generation Z, born between 1997 and 2012, came into the world where the labor market had become increasingly quiet, offering less economic value than it used to: wage stagnation, high rates of unemployment among recent graduates, and a university degree that provides less financial security than what the generation before received from it. Yet, on the other hand, Gen Z is known for its lack of involvement in risky activities that have characterized the generations preceding it. Gen Z members drink less alcohol, engage in less sexual activity, and take fewer social risks than other generations.

These two observations are generally studied separately. Economists analyze the changes in the labor market, while sociologists examine behavioral changes among youth. However, examining these issues together poses an intriguing question that has yet to be addressed by either group: can these trends be correlated? Are these patterns of caution among Gen Zers a matter of character, culture, or simply a logical response to economic uncertainty?

This issue is crucial for understanding, teaching, using, or helping youths in any manner whatsoever. A shift towards lower levels of social risk behaviors could be understood in one particular light if it was largely cultural, while an economic explanation would point in a completely different direction (towards earnings and employment trends, as well as increasing debt loads compared to returns on investments in higher education). An incorrect analysis would mean a misinterpretation of an entire generation. The key research question we address in this paper is the following: To what extent can changes in risk behavior among Gen Zers be related to changes in outcomes for new graduates?

Data Sources

For this analysis, we utilize six data sources through two main sources.

The labor market data source is from the Federal Reserve Bank of New York, Labor Market for Recent College Graduates. This source provides data on labor markets nationwide for individuals aged 22 to 27 years. The four datasets used here include college_outcomes.csv, which compares outcomes by major; college_underemployed.csv, which examines the trend in long-term underemployment; college_unemployed.csv, which is mainly used in the unemployment trend; and college_wages.csv, which looks at the wage inflation.

On the other hand, the behavioral data source is from the CDC, Youth Risk Behavior Surveillance System (YRBSS). These sources provide data from 1991 to 2017, which has been narrowed down to participants from New York State. Two datasets utilized from the source are alcohol_and_other_drug_use.csv, which looks into substance abuse trends; and sexual_behaviors.csv, which studies changes in sexual behavior among the youth. Further, the visualizations were limited to participants from New York City.

Design Approach

Why Closeread?

Our main novel tool in use during this project is Closeread, an extension of Quarto that allows us to create a scroll-driven narrative. We looked into {ggiraph} initially, as it could have provided us with tooltips and hovering capabilities in our regular Quarto document. We ruled out this idea since {ggiraph} doesn’t enable the construction of a sticky-visual narrative flow. The logic of our reasoning is simple: for a good narrative, every visualization should be placed in its position until the text is analyzed.

We also thought about making a Shiny dashboard for the same reason of giving users some extra freedom to navigate through our data. However, it goes against our narrative logic, and the point is that the Gen Z paradox requires following a certain sequence of presenting data. The most convincing case for it can be done only by introducing the economic trends first and then the behavioral ones, with a combined overlay being presented in the very end. All visualizations in this project have been made using ggplot2 and plotly libraries with custom theme_gen_z().

Design Decisions

The choice of Federal Reserve Bank of New York’s data set was made in preference to Bureau of Labor Statistics’ offerings because the former data set was designed around the age group of interest. Within this data set, we chose to load up the underemployment variable along with unemployment, but we focused on unemployment rates and salaries. Underemployment seems to be a more problematic trend than unemployment; however, it can be more difficult to articulate in simple terms. Moreover, presenting the idea of underemployment to the public without appropriate background information could complicate our story.

We decided to introduce the wage stagnation graph first and not start the article with the graph of unemployment because, by doing so, we would shift the focus from the question about the availability of work positions to the question about whether the students receive what they were promised upon graduation. We originally included the graph that combines alcohol use rate and graduate unemployment rate at some point in the middle of the article; however, it was moved to the end.

In the peer review by Cozy Candles, it was pointed out that our initial draft had not made it clear on how the direction of causation ran. We have now revised our language to ensure that the entire discussion is consistent in treating the association as one of correlation rather than causation. In the peer review by Cozy Blankets, we were asked why we highlighted recent graduates and alcohol in particular. The reason why we selected recent graduates is that the dataset from the Federal Reserve is organized in such a way that recent graduates will be most impacted by the events we discuss. We focused on alcohol since it was the only one that saw a steady decline; marijuana was constant and hard drugs were already rare.

Key Findings

The median income of those with a Bachelor’s degree increased little over 35 years, from about $58,000 in 1990 to approximately $60,000 in present-day terms after adjusting for inflation. From the graph on unemployment, it can be seen that younger employees and college graduates suffer from higher rates of unemployment compared to other age groups and educational levels, especially during peaks in the years of 2008-2010 and 2020.

From the behavioral perspective, alcohol consumption among NYC adolescents decreased from nearly 40% in the late 1990s to less than 20% in the mid-2010s. Sexual engagement declined from 50% in the early 2000s to 30% by 2017. Consumption of marijuana and more potent drugs did not change much, which is quite interesting since it indicates that the decline in risky behaviors is specifically directed at drinking and sex.

When looking at both charts simultaneously, it can be observed that they are discussing each other. While the rate of graduate unemployment peaked and the salary growth slowed down, the rate of youth taking risks steadily decreased.

Limitations

The first one is that of confounders; smartphones, social media, new cultural trends, and better health education might be independent explanations for the observed behavioral changes. We have tried to tackle it by always using correlations rather than cause-and-effect relations while formulating each point of the study and highlighting this limitation in the conclusion.

There is also a real problem with a geographical mismatch of the two types of data, one being national labor market statistics and the other filtered to NYC. To try to reduce the significance of this problem, we chose New York City since it is among the most competitive labor markets in the United States, while it is also one of the most surveyed places in the YRBSS. It would not have made much sense to narrow down our national labor market data to the state level due to insufficiently detailed statistics.

The difference in time span covered by the behavioral data, which represents high school students, and that covered by the labor market data, which represents 22 years old and above graduates, creates a generational gap in terms of causality. In order to bridge this gap, we started our analysis of the behavioral data from the first YRBSS year, namely 1991, in order for the dataset to cover the formation years of the generation of individuals that entered the labor market during periods of heightened unemployment. The teens who reported having consumed alcohol during the late 1990s are basically the same group that graduated during 2008 to 2010.

As a final note, the end of YRBSS data series is in the year 2017, which means that we did not manage to cover the COVID-19 period through these surveys. It was discussed to incorporate another source with more recent data, but this was impossible due to lack of comparable datasets.

Conclusion

The paradoxical nature of Generation Z, which does less drinking, less sexual activity, and less risky behavior compared to previous generations, is commonly considered a cultural quirk. But this research offers a different way of seeing things, suggesting that economic factors that shape the labor market among graduating students might also shape behavior. As the consequences of making mistakes become greater, being cautious starts to feel more necessary than simply responsible.

Although we see this relationship now it does not mean it will last forever. But we can certainly see the consistency of the temporal trend, its alignment with the direction of behavioral trends, and the need to study the issue in a more sophisticated manner. The Scrollytelling form was selected as a means of presenting this argument since it requires gradual construction as part of a story rather than a correlation matrix. If the issue of the labor market persists for future graduates, the question of choice or necessity becomes especially important.