strategic DESIGN + data analysis

Designing Smarter Policies with Data: A Gambling & Mental Health Case Study

ORGANIZATION

GambleAware
NSW

ROLE

Data analyst
Strategic Designer

EXPERTISE

Urban Planning, Policy and Design

Data Analysis

Urban Planning, Policy and Design
Data Analysis

YEAR

2024

Why This Matters

When designing policies for urban health, we need more than just good intentions—we need hard data to guide our decisions. This case study explores how analyzing the relationship between counseling and gambling behavior can lead to better strategies for public health interventions.

In this case study, I explored the relationship between gambling and mental health in NSW to uncover patterns that could inform better public policies. The goal? To bridge the gap between data and action, ensuring that policies are based on real evidence, not just assumptions.

Timeline

1 week

What’s Going On?

Gambling is a significant public health issue, often linked to financial hardship and mental health challenges. This study investigates the spatial distribution of gambling machines and mental health services (counseling centers) in Greater Sydney to explore whether access to mental health services correlates with lower gambling machine availability.

To make sense of the problem, I worked with datasets on gambling behaviors, mental health reports, and socioeconomic factors in NSW. Here’s where data analysis became crucial: trends aren’t always obvious, and raw numbers don’t speak for themselves.

Key questions I aimed to answer

Understand how the availability of mental health services correlates with the concentration of gaming machines across different regions.

Indicators

Gambling Machine (Clubs) Density v.s Mental Health Service Density

Mental Health Service Accessibility

Service Adequacy Ratio

🔬 Data Analysis Approach

I conducted a Zipf analysis to study how gambling machines and counseling services are distributed across Local Government Areas (LGAs) in Greater Sydney. Zipf’s Law helps us understand size distribution—in this case, how gambling machines and counseling services are ranked relative to population size.

The goal? Identify disparities and provide insights for better urban planning and policy development.

Data Source

Gambling machine distribution across Sydney LGAs
Number of available counselors per LGA
Population data (age 20+) per LGA

Zipf Analysis

For frequency distribution, identifying disparities, etc. more direct approach to examining the spatial distribution of gambling machines and counseling services

📌 Key Takeaway

Gambling machines are more concentrated in specific areas, while counseling services are distributed more proportionally.

This imbalance can create higher gambling exposure in vulnerable communities without adequate mental health support.

📊 🎨 Visualizing the Data

📊 🎨 Visualizing the Data

I merged the datasets and applied log-log transformations to compare population size against the number of gambling machines and counseling services. This method, similar to Zipf’s Law, allows us to see if resources are distributed fairly or disproportionately concentrated.

Using Seaborn and Matplotlib, I plotted:

Gambling machine density vs. population (log-log scale)
Counseling service density vs. population (log-log scale)
Trendlines to visualize disparities

# Load the data from CSV files
gambling_data = pd.read_csv("club_gambling_machine_numbers_LGA_Greater_Sydney.csv")
counseling_data = pd.read_csv("number_of_counsellors_LGA_Greater_Sydney.csv")
population_data = pd.read_csv("population_age over_20LGA_Greater_Sydney.csv")

# Merge datasets
merged_data = pd.merge(gambling_data, counseling_data, on='LGA_NAME21', suffixes=('_gambling', '_counseling'))
merged_data = pd.merge(merged_data, population_data, on='LGA_NAME21')

# Calculate log values for ranking
merged_data['Log Population'] = np.log(merged_data['Total Population Age over 20'])
merged_data['Log Gambling Rank'] = np.log(merged_data['Electronic Gaming Machine numbers'])
merged_data['Log Counseling Rank'] = np.log(merged_data['Number of Counsellors'])

🔬 Findings: Gambling Machines vs. Counseling Services

1️⃣ Gambling Machines Are Disproportionately Clustered

The trendline equation:
y = 1.51x - 10.94

A coefficient above 1 suggests that gambling machines are more concentrated in certain areas than expected based on population size.
This means some communities are exposed to significantly higher gambling risks than others.

2️⃣ Counseling Services Are More Evenly Distributed

The trendline equation:
y = 0.72x - 5.76

A coefficient below 1 suggests that counseling services are more evenly spread across LGAs.
This implies mental health resources are not disproportionately allocated, but some high-risk areas remain underserved.

# Fit trendlines
slope_gambling, intercept_gambling, _, _, _ = stats.linregress(
    merged_data['Log Population'], merged_data['Log Gambling Rank'])

slope_counseling, intercept_counseling, _, _, _ = stats.linregress(
    merged_data['Log Population'], merged_data['Log Counseling Rank'])

# Plot data
plt.figure(figsize=(12,6))
sns.scatterplot(data=merged_data, x='Log Population', y='Log Gambling Rank', label='Gambling Machines')
sns.scatterplot(data=merged_data, x='Log Population', y='Log Counseling Rank', label='Counseling Services')
plt.plot(merged_data['Log Population'], slope_gambling * merged_data['Log Population'] + intercept_gambling,
         color='purple', linestyle='--', label=f'Gambling Trendline: y = {slope_gambling:.2f}x + {intercept_gambling:.2f}')
plt.plot(merged_data['Log Population'], slope_counseling * merged_data['Log Population'] + intercept_counseling,
         color='blue', linestyle='--', label=f'Counseling Trendline: y = {slope_counseling:.2f}x + {intercept_counseling:.2f}')
plt.xlabel('Log Population (Age over 20)')
plt.ylabel('Log Rank')
plt.legend()
plt.show()

🏆 Ranking Areas in Sydney (LGAs): Who’s Doing Well and Who’s Struggling?

The performance of 33 LGAs (Local Government Areas) in Sydney are compared by three key performance indicators:

1️⃣ Counselor to Gambling Machine Ratio – More counselors per machine = better support.
2️⃣ Counselor to Population Ratio – More counselors per capita = better mental health accessibility.
3️⃣ Gambling Machine to Population Ratio – More gambling machines per capita = higher risk

🔎 Inverse Relationship Between Gambling Machines and Mental Health Services

Areas with high counseling service density (e.g., Sydney CBD) have fewer gambling machines.

Regions with more gambling machines tend to have fewer counseling centers, suggesting an uneven distribution of mental health support.
Higher counselor-to-population ratios correlate with lower numbers of gambling machines, suggesting that increased access to mental health support may reduce gambling prevalence.

🏅 Best-Performing Areas

🥇 Ku-ring-gai – The best performer! It has the highest availability of mental health support relative to gambling exposure, with the lowest gambling machine density.

🥈 Hunters Hill – Strong counselor presence per gambling machine and per capita, with relatively low gambling exposure.

🥉 Woollahra – A good balance of counseling support and fewer gambling machines per capita.

😓 Worst-Performing Areas

😩 Fairfield – The highest gambling machine density in Sydney but one of the lowest counseling support levels.

🙁 Georges River – Ranked last in counseling access, meaning people here have the least support to deal with gambling harm.

😔 Canterbury-Bankstown – One of the worst counselor-to-gambling machine ratios, meaning people are much more likely to encounter a gambling machine than a mental health professional.

Correlation between Counselor to Gambling Machine Ratio and Population: -0.4398673850174663
Correlation between Counselor to Population Ratio and Population: -0.400337454983926
Correlation between Gambling Machine to Population Ratio and Population: 0.3965035273255982
                      Counselor_to_GamblingMachine  \
LGA_NAME21                                           
Ku-ring-gai                               0.189655   
Hunters Hill                              0.153846   
Woollahra                                 0.135593   
Wollondilly                               0.077844   
Blue Mountains                            0.067485   
Waverley                                  0.056911   
Mosman                                    0.053333   
Lane Cove                                 0.046875   
Sydney                                    0.039427   
Inner West                                0.035556   
The Hills Shire                           0.027244   
Strathfield                               0.027027   
Hawkesbury                                0.026415   
Canada Bay                                0.025391   
Parramatta                                0.022059   
Northern Beaches                          0.021881   
Willoughby                                0.021448   
Sutherland Shire                          0.020503   
Hornsby                                   0.019837   
Bayside (NSW)                             0.017958   
...
Liverpool                                           19.0  
Penrith                                              5.0  
Canterbury-Bankstown                                 3.0  
Georges River                                        4.0

Correlation between Counselor to Gambling Machine Ratio and Population: -0.4398673850174663
Correlation between Counselor to Population Ratio and Population: -0.400337454983926
Correlation between Gambling Machine to Population Ratio and Population: 0.3965035273255982
                      Counselor_to_GamblingMachine  \
LGA_NAME21                                           
Ku-ring-gai                               0.189655   
Hunters Hill                              0.153846   
Woollahra                                 0.135593   
Wollondilly                               0.077844   
Blue Mountains                            0.067485   
Waverley                                  0.056911   
Mosman                                    0.053333   
Lane Cove                                 0.046875   
Sydney                                    0.039427   
Inner West                                0.035556   
The Hills Shire                           0.027244   
Strathfield                               0.027027   
Hawkesbury                                0.026415   
Canada Bay                                0.025391   
Parramatta                                0.022059   
Northern Beaches                          0.021881   
Willoughby                                0.021448   
Sutherland Shire                          0.020503   
Hornsby                                   0.019837   
Bayside (NSW)                             0.017958   
...
Liverpool                                           19.0  
Penrith                                              5.0  
Canterbury-Bankstown                                 3.0  
Georges River                                        4.0

🧐 Observations & Unexpected Trends

Observations & Unexpected Trends

Data Suggests that Regions with more gambling machines tend to have fewer counseling centers. (more mental health support = less gambling)
Areas with high counseling service density (e.g., Sydney CBD) have fewer gambling machines. (less mental health support = more gambling)

However, there are exceptions…

❗ Areas where gambling machines are high, but counseling is low

Fairfield, Campbelltown, Canterbury-Bankstown – These LGAs have high gambling exposure but minimal mental health services.

❗ Areas where counseling is high, but gambling is also high

Hawkesbury & Sutherland Shire – Despite having more counselors, they still have a relatively high number of gambling machines.

📌 Unexpected Trend Explained

LGAs with unexpected patterns are investigated individually, such as:

Hawkesbury has a historical horse racing culture dating back to 1829. This could be driving gambling demand despite counseling availability.

Sutherland Shire has higher median incomes, meaning people have more disposable income for gambling activities.

🤔 Causation vs. Correlation

While the correlation is strong, the underlying causes remain complex:

Regulatory Policies: LGAs with more mental health services may have stricter gambling regulations, reducing machine density.
Socioeconomic Factors: Lower-income areas may have more gambling machines and fewer mental health resources due to funding disparities.
Young males under 35 with higher incomes and education levels are more likely to gamble online
LGAs with high non-English-speaking populations report higher gambling-related harm
Public Health Impact: Increased access to mental health services may help reduce problem gambling, leading to fewer machines being used.

🧐 How accessible are GambleAware Clinics in Sydney?

Having counseling services is one thing—but are they conveniently located for people to use?

I analyzed GambleAware Clinics, which provide free counseling for gambling harm.

✅ Most clinics are within a 10-minute walk from a train station, making them accessible.

❌ BUT some high-risk areas lack clinics!

Hornsby, Hawkesbury, and Northern Beaches have many gambling machines but no clinics nearby.

                Location Nearest Train Station  Travel Time (minutes)
0               Ashfield      Ashfield Station                    4.0
1              Bankstown     Bankstown Station                    3.0
2                Belmore       Belmore Station                    1.0
3                 Bowral        Bowral Station                    2.0
4           Campbelltown                  None                    NaN
5             Camperdown       Newtown Station                    7.0
6   Fairfield (Smart St)     Fairfield Station                    2.0
7    Fairfield (Dale St)     Fairfield Station                    0.0
8               Lewisham      Lewisham Station                    2.0
9              Liverpool     Liverpool Station                    4.0
10            Parramatta    Parramatta Station                    6.0
11               Penrith       Penrith Station                    2.0
12           Seven Hills   Seven Hills Station                    6.0
13               Tahmoor       Tahmoor Station                    3.0

📌 Key Takeaway

While clinic locations are strategic, some at-risk communities remain underserved.

Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.
side-by-side comparison
In areas where the counselor to person ratio is lower, indicating fewer counselors per person, there tends to be a higher concentration of gaming machines per person(Penrith, Campelltown, Randwick, Burwood, InnerWest, etc.).
Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.
side-by-side comparison
In areas where the counselor to person ratio is lower, indicating fewer counselors per person, there tends to be a higher concentration of gaming machines per person(Penrith, Campelltown, Randwick, Burwood, InnerWest, etc.).
Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.
side-by-side comparison
In areas where the counselor to person ratio is lower, indicating fewer counselors per person, there tends to be a higher concentration of gaming machines per person(Penrith, Campelltown, Randwick, Burwood, InnerWest, etc.).
Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.
side-by-side comparison
In areas where the counselor to person ratio is lower, indicating fewer counselors per person, there tends to be a higher concentration of gaming machines per person(Penrith, Campelltown, Randwick, Burwood, InnerWest, etc.).

Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.
Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.
Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.
Less counselling service = more gambling machines
Areas with fewer counseling services, like Penrith and Blacktown, tend to have more gambling machines, suggesting a link between mental health support and gambling prevalence.

⚖️ Designing Policies

Counseling Service Gaps

More counseling centers should be established in high-risk areas (e.g., Liverpool, Blacktown, and Fairfield).
Online counseling services should be expanded to meet the growing demand for remote mental health support.
Face-to-face counseling remains crucial, particularly for LGAs with fewer counselors but a growing need for intervention.

Targeted Support for Vulnerable Groups

Young adults (18-24) and non-English speakers are at higher risk of gambling harm.
LGAs like Fairfield and Liverpool need more multilingual counseling services.

Alternative Recreational Facilities

Areas with a high concentration of gambling machines may benefit from investment in alternative entertainment options. (LGA Camden is a good example this works)
Expanding sports and community centers can provide healthier alternatives to gambling.

GambleAware Clinics Expansion

More clinics should be placed in high-population areas with few counselors.
Confidential and multilingual online support should be prioritized.

💡 Final Thoughts: Why Data Analysis is Crucial in UX & Policy Design

This project shows how data analysis helps shape better policies by identifying hidden disparities in resource allocation.

Using these insights, policymakers can refine strategies to reduce gambling-related harm, ensuring equitable access to support services, and be precise in resources allocation.

For policymakers

These insights can guide funding decisions to make mental health support more accessible.

For designers

It’s a reminder that data informs strategy, whether you're working in public health, urban planning, or UX research.

Let's Connect!

Case studies

Let's Connect!

Case studies