Every year, we gallop toward a day of green, wearing it all over, dying various water features and/or rivers, and/or massive mugs of beer — all in the service of St. Patrick’s Day, March 17th. (Savannah in particular transforms itself into a kind of collective organism that gobbles up the city for 24 to 36 hours.)
That results in a comprehensive and widespread day of pain the following day.
I have always been curious about how pervasive this phenomenon is and how intense the pain is felt. Since I have too much time on my hands, I created two indexes based on real life metrics that align with hangovers.
We are going to create two indexes:
- A Hangover Participation Index (HPI). This measures the scale of participation in St. Patrick’s Day celebrations that result in an unfortunate combination of pain and regret.
- A Hangover Severity Index (HSI): This measures the intensity of pain within those hangovers and how wretched the aftermath becomes (e.g., medical, behavioral, safety).
Each index is a number based on a weighted combination of factors (variables) that are directly related to hangovers. Think of weights as an indication of how important the variable is within the index. Some variables will be included in both indexes, but weighted differently. This data is based on the behavior of New Yorkers from 2016-2024 (the data and years available).
Hangover Participation Index (HPI)
Remember, the HPI measures quantity, not quality, so the variables will be based on things that measure the participation in the parade and as such the higher probability of people getting tanked. Some of them measure the intensity of the hangovers, but at smaller weights they also indicate a count of overindulgence. All weights are represented as percentages. We also use both weather and day-of-week multipliers to factor in how these affect the party.
- Alcohol Sales (% increase), 30%:
Strongest single indicator of overall participation intensity. (Raiding the liquor store.) - Parade Attendance (millions), 20%:
Public engagement and turnout in a central event. (How many green claded maniacs.) - Subway Ridership (millions), 15%:
Proxy for foot traffic and mobility during celebrations. (And getting stuck in turnstiles.) - DUI Arrests (NYPD), 10%:
Risky participant behavior—indicates a rowdier party night. (Don’t be a nitwit.) - Rideshare Demand (% vs avg), 10%:
Reflects private travel due to nightlife and drinking. (And masochist Lyft drivers.) - Google Trends (e.g., “Hangover cure” searches), 10%:
Early post-party regret = indirect measure of how many overindulged. (If you can type.) - Emergency Calls (% increase), 5%:
Slight signal of public-level chaos or volume. (Let’s be more careful.) - Day of the week multiplier:
Reflects how the day of the week impacts how many people participated in the parade. (Begging for Friday.) - Weather Multiplier:
Reflects how weather conditions impact public turnout and the likelihood of getting wasted. (Not really sure how loaded leprechauns care about getting wet.)
When you combine these metrics together, they result in a single number that represents how many of us had our head in the toilet. Though, it doesn’t mean much without comparing it to something else. In this case, over time.
So what’s going on here? How often are we getting sloshed over the past 8 years?
Let’s take a look.
Why does the index move the way it does?
Each year’s hangovers are driven by different factors. Some more than others.
2016 – Classic Thursday Chaos:
St. Patrick’s Day on a Thursday brought people imbibing in droves, good weather, and jolly packed bars—classic NYC weekday partying with lots of sleeping-on-the-subway and high booze sales numbers.
2017 – Friday Green-Out:
The holiday landed on a Friday, giving New Yorkers license to go all-in. Pain numbers surged with high rideshare demand and a spike in desperate Google searches the next day.
2018 – Saturday Blowout:
A Saturday St. Pat’s and perfect weather created one of the highest number of hangovers in years, with slammed parades, all-day bar crawls from midtown to the upper east side (parade route), and far beyond.
2019 – Split Celebrations:
The parade fell on Saturday (March 16) while the holiday was Sunday. Sadly, this split celebrations and dulled turnout on the actual day, lowering the number of overall hangovers.
2020 – Pandemic Wipeout:
NYC shut down bars and canceled the parade on March 16. March 17 saw nearly zero public activity but at least we had Tiger King.
2021 – Party in Quarantine:
No parade, limited indoor dining, and lingering COVID concerns kept the parties quiet. Participation began to return, but stayed small and mostly private.
2022 – Return to the Green:
The parade returned, bars reopened, and St. Pat’s fell on a Thursday! While office crowds were still light, public celebration rebounded massively. Good things actually do come to those who wait.
2023 – Guinness Gone Wild:
Full recovery, packed parade, Friday holiday = the most intense blowout since 2018, with spikes in pricey rideshare, nodding off and missing subway stops, and bar traffic.
2024 – Sunday Surge, Monday Misery:
The parade was strong (held Saturday), but actual March 17 celebrations were quieter. Participation dipped slightly. (This is a puzzler. Doesn’t follow the model as much as we’d expect. But St. Pat’s may just be its own animal once in awhile.)
Key Drivers of HPI Changes Over Time
There are some common qualities that affect our boozing. We seem to be predictable.
- Day of the Week: Fridays and Saturdays supercharge our getting bombed; Sundays and weekdays dampen it.
- Weather: Good weather boosts parade and bar attendance. Cold/rainy years keep us more indoors, even with our broken green umbrellas.
- COVID-19: 2020–2021 brought shutdowns and severe restrictions, driving HPI to historic lows.
- Transit & Rideshare Use: Spikes in subway and Uber/Lyft demand mirror public mobility and dancing on bar tops.
- Parade Status: Obviously, full parade = big HPI. Cancellations or low turnout = weak participation.
Enough about how much we partied. How much pain and suffering did we inflict upon ourselves? Going to bed happy and waking up in the bag?
Let’s take a look.
Hangover Severity Index (HSI)
Remember, the HSI measures the severity of anguish, so the variables will be based on things that measure how tanked we get whilst staggering around in the parade.
- DUI Arrests (NYPD), 30% weight:
Severe legal consequence; risk to others. (C’mon, be smart.) - Alcohol-Related ER Visits (Est.), 25% weight:
Direct medical harm from overconsumption. (Ooof.) - Emergency Calls (% Increase), 20% weight:
Public-level crises – fights, unconscious people, etc. (You can be a grownup if you try.) - Google Trends (“Hangover cure” searches), 15% weight:
Mild-moderate hangover impact: self-care and suffering. (At least you can type.) - Over-the-Counter Pain Relief Sales (Est.), 5% weight:
Indicator of milder suffering across large population. (Four Advil, three Tylenol, every hour.) - Hydration Product Sales (Est.), 5% weight:
Indicator of moderate suffering and physical recovery needs. (Pedialyte anyone?) - Day of the week multiplier:
Reflects how the day of the week impacts likely differences in throbbing heads. (TGI Friday.) - Weather Multiplier:
Reflects how weather conditions impact boozy turnout and celebration behavior. (So many puddles.)
Look familiar? It’s moving in a similar direction as the HPI, though not perfectly.
Why are hurting so badly?
2016 – Balanced Aftermath:
Strong celebrations led to a noticeable hangover effect: moderate ER visits, DUI arrests, and Google searches for cures—all signs of a big but manageable night.
2017 – Mellow Recovery:
Despite heavy partying on Friday, the city avoided serious fallout—perhaps thanks to good transit use and awareness campaigns. Hangover severity stayed contained.
2018 – The Great Hangover:
Saturday partying + great weather = all-day drinking and night-time chaos. ER visits, emergency calls, and pain relief sales surged—a peak year of suffering.
2019 – Sunday Saves the Day:
With the holiday falling on a Sunday, fewer people overindulged and most stayed local. Consequences (DUI, ER visits) remained well below peak levels.
2020 – COVID Collapse:
With no public events and bars closed, medical and legal hangover data dropped off a cliff. Google searches for “hangover cure” were almost non-existent but we were introduced to Ted Lasso so still a win.
2021 – Pandemic Plateau:
A few small hangovers from at-home drinking, but still well below average. OTC sales and ER visits remained muted; the agony hadn’t returned yet.
2022 – The Pain Comes Back:
As bars reopened and ambitious boozers filled the streets, NYC’s hangovers made a strong return. Google searches, DUI arrests, and 911 calls all ticked upward again.
2023 – Full Force Fallout:
The first post-COVID Friday St. Pat’s brought both high participation and widespread regret—Google searches and painkiller sales hit near-record highs and all sorts of other idiotic behavior manifested in the form of a floundering leprechaun.
2024 – Monday Misery:
Though Sunday partying was lighter, Monday March 18th hangovers were brutal. Google Trends and hydration product sales soared, boosting HSI despite moderate turnout.
Key Drivers of Pain
- Participation Volume: More people partying = more hangovers and injuries—but not always proportionally.
- Day of the Week: Sunday St. Pat’s yields softer partying and fewer consequences; Friday/Saturday ramps up late-night fallout.
- Public Behavior vs. Medical Consequence: High turnout doesn’t always mean high severity—safer transit options or moderate drinking can decouple them.
- Pandemic Suppression: 2020–2021 saw near-zero ER visits, DUI arrests, and hangover behaviors due to shutdowns.
- Delayed Effects: When St. Pat’s falls right before a workday (e.g. Sunday or Thursday), the next-day misery (March 18) is amplified.
Participation vs Pain
Why the HPI and HSI Differ
Hangover Participation Index (HPI) measures how much people partied.
Hangover Severity Index (HSI) measures how much they suffered afterward.
Just because people went hard doesn’t always mean they woke up in the ER. And just because partying was light doesn’t mean a few overachievers didn’t have to Google “can you overdose on Guinness.”
Let’s outline more of the key reasons?
1. More Partiers ≠ More Pain
Example: 2017 or 2023:
- A Friday St. Patrick’s Day draws massive crowds (HPI), but good weather, transit use, and smart choices keep ER visits and DUI arrests in check (HSI moderate).
- Think: “We drank a lot, but we actually managed to act somewhat responsibly.”
2. Smaller Crowds Can Still Have Big Hangovers
Example – 2024:
- A Sunday holiday means fewer partygoers overall—but those who went out really went for it, and Monday was brutal.
- Painkiller sales and hangover cure searches spike even if participation wasn’t massive.
3. Public vs Private Consequences
- HPI measures what happened in public: alcohol sales, crowds, subway rides.
- HSI includes what happened later or behind closed doors: ER visits, OTC sales, online searches for “how to survive a hangover” or “how do I clean green beer off the carpet.”
4. COVID Years: Anomalous Decoupling
2020 & 2021:
- HPI: Basically zero (no parade, no bars, no fun).
- HSI: Also near zero, but not completely—some people still managed to create their own private zoom party disasters at home. Just fewer ambulances involved.
5. Weekday vs Weekend Dynamics
- A St. Pat’s on a Thursday or Friday inflates both HPI and HSI.
- A Sunday holiday brings a spike in HSI (Monday misery) without necessarily boosting HPI.
Bottom Line
They’re siblings. They’re related. But they don’t always ride the same subway.
HPI: “How much green did we spill in the streets?”
HSI: “How many aspirin did we crush the next day?”
Data Calculation
We don’t always have exact, year-by-year data for every variable in NYC tied to St. Patrick’s Day, so we use a combination of real anchor points, reputable data sources, and informed estimation methods to project those values consistently from 2016 to 2024.
General Methodology
Find Real-World Anchor Data:
- From public sources (NYPD, MTA, Google Trends, Nielsen, etc.) and proprietary benchmarks (Lyft data, Catalina sales reports, etc.).
Normalize Values:
- Data was scaled relative to the maximum observed or plausible value for each variable across all years.
Estimate Missing Years Using Logical Patterns:
- Based on day-of-week effects (e.g., Friday = more partying)
- COVID impact (2020–2021: major drops)
- Weather influence, when relevant
- Population and behavior patterns, e.g., NYC’s reliance on transit vs car usage
Variable-by-Variable Calculation
Alcohol Sales (% Increase)
- Anchor: NielsenIQ showed +174% beer sales on St. Pat’s nationally.
- Method: Used this as a ceiling for peak years (2016–2019, 2023–2024).
- COVID years (2020–2021): Reduced to 0–20% due to bar closures; used known lockdown dates.
Parade Attendance (Millions)
- Anchor: Parade Committee claims ~2 million spectators in normal years.
- 2020–2021: Officially canceled → attendance = 0.
- 2022: Parade returned but rain + remote work = ~0.05M (estimated from media).
Subway Ridership (Millions)
- Anchor: MTA published daily ridership totals (e.g., 5.57M in 2019).
- Method: Applied known weekday/weekend averages, adjusted for pandemic collapse in 2020–21.
- 2022–2024: Rebounded based on MTA recovery stats (~3–4M typical range).
DUI Arrests (NYPD)
- Anchor: 54 DUI arrests reported by NYPD on St. Pat’s 2016.
- Method: Used 30–60 range in normal years, depending on day-of-week. Near-zero in 2020–21.
Rideshare Demand (% vs. Avg Day)
- Anchor: Lyft blog (2023) reported +23% national spike.
- Method: Estimated NYC-specific increases (10–30%) in peak years. Large decline in 2020 due to lockdowns.
6. Google Trends (“Hangover cure”)
- Anchor: Google Trends provides a 0–100 index.
- Method: Used March 18 search spikes for “hangover cure” relative to the New Year’s Day benchmark (100). Lower index in 2020–21 (little partying), highest in 2019 and 2024 (Monday misery).
7. Emergency Calls (% Increase)
- Anchor: Based on anecdotal FDNY and NYPD activity spikes around major holidays.
- Method: Assumed +10–12% increase in high-activity years, -20% during pandemic.
8. ER Visits (Raw Count)
- Anchor: ERs report 200–400 alcohol-related visits in NYC on heavy holidays.
- Method: Estimated based on participation intensity. Peak in 2018 and 2023 (~350–400), bottomed out in 2020 (~50).
OTC Pain Relief Sales ($)
- Anchor: Catalina reported +24% OTC “hangover remedy” sales on St. Pat’s.
- Method: Scaled national market to NYC population, projected ~$100k–$650k in sales depending on year.
Hydration Product Sales ($)
- Anchor: Nielsen & IRI data; Pedialyte sales increasingly tied to adult hangovers.
- Method: NYC estimates range from ~$50k to ~$180k on March 18 depending on severity of prior day.
Bottom line Summary
All variable values were anchored in real-world data, then estimated across years using consistent logic, adjusted for context like COVID, calendar effects, and known NYC behavior patterns (e.g., reliance on transit, late-night partying).
Raw Data Sources
- NielsenIQ. (2023). St. Patrick’s Day Alcohol Sales Analysis. Retrieved from https://nielseniq.com
- Union POS Data. (2022). On-Premise Alcohol Sales Trends (Internal Report).
- Catalina Shopper Intelligence. (2022). Retail Sales During Holiday Weeks.
- National Retail Federation. (2023). St. Patrick’s Day Spending Statistics. Retrieved from https://nrf.com
- NYC St. Patrick’s Day Parade Committee. (2024). Official Parade Data & Archives. Retrieved from https://nycstpatricksparade.org
- MTA. (2024). Daily Turnstile Counts. Retrieved from https://data.ny.gov
- NYPD. (Various Years). St. Patrick’s Day Arrest Reports (via local news outlets).
- New York State Police. (2024). STOP-DWI Enforcement Results. Retrieved from https://troopers.ny.gov
- Lyft. (2023). Most Irish Cities: St. Patrick’s Day Blog. Retrieved from https://lyft.com/blog
- Google Trends. (2024). Search Interest in “Hangover Cure”. Retrieved from https://trends.google.com
- FDNY & NYC 311. (2023). Emergency Call Statistics (modeled from dispatch data and news reports).
- CDC AHEAD Tool. (2024). Alcohol-Related Hospitalizations & ER Visits. Retrieved from https://ahead.cdc.gov
- NYC ER Physician Commentary. (2019–2023). St. Patrick’s Day Case Reports (quoted in NYPost, Gothamist).
- FDNY & NYC 911 Dispatch. (2024). Emergency Response Volume During Holidays (via internal reports & news coverage).
- NYPD & NYS STOP-DWI Campaign. (2023). Holiday Arrest Summaries. Retrieved from https://stopdwi.org
- NielsenIQ. (2023). OTC and Health Product Retail Trends.
- Consumer Healthcare Products Association (CHPA). (2023). Annual OTC Pain Relief Market Report.
- IRI. (2023). Hydration and Rehydration Product Market Trends.
- Pedialyte. (2023). Brand Marketing & Consumer Use Studies (internal).