Objective

Explain the objective of the original data visualisation and the targeted audience.

The following graph, available online from InformationIsBeautiful.net, focuses on a study conducted in December 2010 by D. McCandless et. al. The subject of this visualisation is regarding the potential tax revenues that could be generated if all illicit drugs were to be legalized and controlled within the United Kingdom.

The main objective of this data visualization is to present an argument as to the benefits associated with legalising illicit substances. This is depicted by providing an estimate of the potential tax revenues that could be generated and comparing this to other revenue sources and UK budget expenditures.

According to the underlying dataset of this graph, the legalization and taxation of all illicit drugs would result in an estimated additional revenue of 5 billion pounds per annum.

Original


Source: InformationIsBeautiful.net


The visualisation chosen had the following three main issues:

Issue 1:

Issue with Data Integrity:

The source of the data used to prepare the visualisation is from an original study undertaken by M. Atha in October 2004.

Upon review of the underlying data used in the visualisation, the figure used for VAT (UK’s equivalent to GST) has been calculated in a different way as provided by M. Atha, and includes a comment that the figure had been ‘corrected’.

The ‘Tax Revenue’ income used in the graph for each drug type is intended to depict a 50% duty tax on sales. For example, if the retail price is $50, then a further $50 is added to the price in the form of a duty tax. VAT would then be calculated by multiplying 100 by the VAT percentage of 17.5%. This was correctly calculated by M. Atha, however D. McCandless changed this formula to take 17.5% of just the duty taxes.

This resulted in the estimated average VAT income dropping from 654bn to 327bn pounds, which is a significant reduction in the true reflection of potential revenue as intended by M. Atha.

Issue 2:

Perceptual Issues:

The graph present by D. McCandless is difficult to understand and does not provide any form of graphical comparison for how the hypothetical tax revenue from illegal drugs compares to other revenue sources. Whilst the original study conducted by M. Atha aimed to show the benefit of additional tax revenue, the visualisation prepared by D. McCandless seems to otherwise focus on depicting the breakdown in revenue by each individual drug.

The breakdown of revenue by drug is depicted in a brightly coloured rainbow that takes up the majority of the graphic. This makes it so the actual comparison of the totals against other revenue streams and expenditures is pushed to the side and depicted in light grey coloured boxes.

A bar graph would better present the information in order to support the original intention of the study.

Issue 3:

Deceptive Methods:

In order to enhance the outcome of the study, figures have been included that have inflated the estimated benefit that could be derived by legalising illegal drugs. The most outlandish figure included in the 5bn revenue estimate is the 1bn included as ‘income tax from working users’.

This figure is calculated by the following: 1. Survey from 1994-1997 estimated that those with a drug conviction earn 1,000 per year less than those with no criminal record, multiplied by 2. The number of individuals with a drug conviction

Based on the estimated resulting benefit of 1bn, M. Atha’s study must have assumed that there are 1,000,000 individuals in the UK that have a drug conviction. M. Atha’s study also states in an early paragraph that “between 1945 and 2002, there were around 1,230,000 cannabis convictions in the UK”. Based on this comment, The figure of 1,000,000 used in the calculation therefore appears grossly exaggerated and is unaccompanied by any proper justification.

Secondly, the societal costs associated with illegal drugs such as death, overdose, hospitalisation, addiction and general loss of productivity are not at all considered in the study. There is the inclusion of ‘New Costs’, which subtracts 0.3bn, however this merely covers costs such as establishing a licensing authority and associated compliance costs. If all illegal drugs were to be made legal, the detrimental costs to the health of society wouldn’t simply vanish.

The culmination of these two items has greatly inflated the estimated benefits associated with the legalisation of illicit substances, which provides deceptively positive evidence to support the basis of the study. Therefore, for simplicity, the additional $1bn will be removed from the visualisation.

Reference

Data Visualisation:

Underlying Study:

Code

The following code was used to fix the issues identified in the original.

#Load Dataset
csv_path <- "drug_deal_data.csv"
df <- read_csv(csv_path, locale = locale(encoding = "UTF-8"))

#Subset to remove unnecessary columns & rows:
df <- df[-(1:10), !names(df) %in% c("Source", "Link", "Note")]
rows_to_delete <- c(5,7,9,10,15:24)
df <- df[-rows_to_delete, ]

#Convert Data Type to Numeric:
df$Average <- parse_number(df$Average)

names(df)[names(df) == "Type.of.Saving"] <- "Type"
df$Combined <- paste(df$Drug, df$Type, sep = " ")

#Split dataset to correct calculations
df <-
  data.frame(
    Type = c(df$Combined),
    Amount = c(df$Average)
  )
df_drug <- df[-(6:10),]
df_graph <- df[-(1:6),]

#Correct inflation row so it's not a cumulative figure:
df_drug <- rbind(df_drug, (c("Inflation Adjustment (10%)", sum(df_drug$Amount)*0.1)))

#Tidy numeric formatting:
df_drug$Amount <- as.numeric(df_drug$Amount)

#Multiply VAT by 2:
df_drug <- rbind(df_drug, (c("VAT", df_drug$Amount[2]*2)))

#Remove the Row "New Income Tax":
df_drug <- df_drug[-3,]

#Add revised Drug Revenue to other graph data:
df_graph <- rbind(df_graph, (c("Drug Tax Revenue", sum(as.numeric(df_drug$Amount)))))

#Tidy numeric formatting:
df_graph$Amount <- as.numeric(df_graph$Amount)
df_graph$Type[2] <- "Alcohol Tax Revenue"
df_graph$Type[3] <- "Tobacco Tax Revenue"
df_graph$Type[1] <- "University Deficit"
rownames(df_graph) <- NULL

#Divide values by 1bn to summarise output
df_graph$Amount <- df_graph$Amount / 1000000000

#Specify custom colour palate:
custom_palette <- c("#336699","#CC6666","#6a3d9a","#33a02c","#fdbf6f")

#Reprepare Plot
drug_plot <- ggplot(df_graph, aes(x = Type, y = Amount, fill = Type)) +
  geom_bar(stat = "identity", width = 0.6) +
  labs(x = "Revenue/Expense Savings", y = "Amount (Billions)", title = "Potential Tax Revenue from Legalising Ilicit Drugs in the UK") + 
  theme(axis.text.x = element_text(angle = 75, hjust = 1)) +
  scale_fill_manual(values = custom_palette)

Data Reference

Reconstruction

The following plot fixes the main issues in the original.