quantitative-methods/descriptive-statistics.org

Quantitative Methods

• Title slide
• Title slide
• What do we aim to achieve in this course?
  • Make friends with numbers
  • Learn how to read numbers, how to present them, and how to write about them
  • Learn how to use computers to work with numbers
• Two Types of Statistics
  • Descriptive Statistics
    • Use summaries of data for the entire population to describe a population
    • Use summaries of sample data to describe a sample
  • Inferential Statistics
    • Use sample data to describe a population
• Descriptive Statistics
  • Frequency
  • Measures of Central Tendency
  • Measures of Position
  • Measures of Dispersion
    • Range and other measures based on positions
    • Range and other measures based on positions
    • Variance, Standard Deviation and Coefficient of Variation
• Graphical Displays of Quantitative Information: Common Pitfalls
  • Common uses of statistical graphics
  • Mis-representation
  • Mis-representation
  • Mis-representation
  • Mis-representation
  • Mis-representation
  • Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)
  • Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)
  • Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)
  • Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)
  • The problem multiplied with the coming in of spreadsheets
• Graphical Displays of Quantitative Information: Dispersion
  • Histogram
  • Histogram with relative densities
  • Boxplot
  • Boxplot of wheat yields
  • Violin plots
  • Boxplots: Useful to identify extreme values
  • Boxplots: Useful for comparisons across categories
  • Violin plots
• Paul Krugman on Fiscal Austerity
  • What does this graph show?
  • What did Paul Krugman say?
  • How could we better show the relationship between unemployment and fiscal austerity

Title slide   slide

(org-show-animate '("Quantitative Methods, Part-II" "Descriptive Statistics" "Vikas Rawal" "Prachi Bansal" "" "" ""))

Title slide

(org-show-animate '("Why do financial journalists need to know quantitative methods?" "" "" ""))

What do we aim to achieve in this course?   slide

Make friends with numbers

Learn how to read numbers, how to present them, and how to write about them

Learn how to use computers to work with numbers

Two Types of Statistics   slide

Descriptive Statistics

Use summaries of data for the entire population to describe a population

Use summaries of sample data to describe a sample

Inferential Statistics

Use sample data to describe a population

Descriptive Statistics   slide

• Frequency
• Measures of central tendency
• Summary positions
• Measures of dispersion

Frequency   slide

      library(data.table)
      data.table(names=c("Anil","Neeraj","Savita","Srimati",
                         "Rekha","Pooja","Alex","Shahina",
                         "Ghazal","Lakshmi","Rahul","Shahrukh",
                         "Naman","Deepak","Shreya","Rukhsana"
                         ),
                 salary=c(71,50,65,40,
                          45,42,46,43,
                          45,43,45,45,
                          850,100,46,48
                          )*1000,
                 sex=c("M","M","F","F",
                       "F","F","M","F",
                       "F","F","M","M",
                       "M","M","F","F"
                       ))->workers
      workers$sno<-c(1:nrow(workers))
      workers[,.(sno,names,sex,salary)]

sno	names	sex	salary
1	Anil	M	71000
2	Neeraj	M	50000
3	Savita	F	65000
4	Srimati	F	40000
5	Rekha	F	45000
6	Pooja	F	42000
7	Alex	M	46000
8	Shahina	F	43000
9	Ghazal	F	45000
10	Lakshmi	F	43000
11	Rahul	M	45000
12	Shahrukh	M	45000
13	Naman	M	850000
14	Deepak	M	1e+05
15	Shreya	F	46000
16	Rukhsana	F	48000

  workers[,.(frequency=length(sno)),.(sex)]

sex	frequency
M	7
F	9

sex	frequency
M	7
F	9

Measures of Central Tendency   slide

  workers[,.(mean_salary=round(mean(salary),1),
              median_salary=quantile(salary,prob=0.5))]

mean_salary	median_salary
101500	45500

  workers[,.(mean_salary=round(mean(salary),1),
             median_salary=quantile(salary,prob=0.5)),.(sex)]

sex	mean_salary	median_salary
M	172428.6	50000
F	46333.3	45000

Measures of Position   slide

• First quartile
• Second quartile (median)
• Third quartile
• Deciles
• Quintiles
• Percentiles

Measures of Dispersion   slide

Range and other measures based on positions   slide

$range=max-min$

min_salary	max_salary	range
40000	850000	810000

    workers[,.(min_salary=min(salary),
                max_salary=max(salary),
                range=max(salary)-min(salary))]

Range and other measures based on positions   slide

• Distance between any two positions (Deciles, Quintiles, Percentiles) can be used as a measure of dispersion.

$inter.quartile.range=Q3-Q1$

  25%   75% 
44500 53750
  10%   90% 
42500 85500
   10%    95% 
 42500 287500
   25%    95% 
 44500 287500
   0%   75% 
40000 53750

##  summary(workers$salary)
  quantile(workers$salary,probs=c(0.25,0.75))
  quantile(workers$salary,probs=c(0.1,0.9))
  quantile(workers$salary,probs=c(0.1,0.95))
  quantile(workers$salary,probs=c(0.25,0.95))
  quantile(workers$salary,probs=c(0,0.75))

Variance, Standard Deviation and Coefficient of Variation

$variance=\frac{1}{n} \times \sum(x_{i}-x)^{2}$

$standard.deviation = \sqrt{variance}$

$cov=\frac{standard.deviation}{mean}$

  workers[,.(var_salary=round(var(salary),1),
             sd_salary=round(sqrt(var(salary)),1),
             cov_salary=round(sqrt(var(salary))/mean(salary),2))
          ]

var_salary	sd_salary	cov_salary
40075200000	200187.9	1.97

    students[,.(var_salary=round(var(salary),1),
                sd_salary=round(sqrt(var(salary)),1),
                cov_salary=round(sqrt(var(salary))/mean(salary),2)),.(sex)]

sex	var_salary	sd_salary	cov_salary
M	89680952381	299467.8	1.74
F	54500000	7382.4	0.16

Graphical Displays of Quantitative Information: Common Pitfalls

Common uses of statistical graphics   slide

• To show trends over time
• To show mid-point variations across categories
• To show composition
• (less commonly, though more usefully) to show/analyse dispersion

Mis-representation   slide

"and sometimes the fact that numbers have a magnitude as well as an order is simply forgotten"

Mis-representation   slide

Another example borrowed from Tufte

Mis-representation   slide

Tufte's graph on fuel economy of cars

Mis-representation   slide

Nobel prizes awarded in science (National Science Foundation, 1974)

Mis-representation   slide

Nobel prizes awarded in science (corrected by Tufte)

Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)   slide

Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)   slide

Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)   slide

Mis-representation: illustrations from Thomas Piketty's work (source Noah Wright)   slide

The problem multiplied with the coming in of spreadsheets   slide

Graphical Displays of Quantitative Information: Dispersion   slide

Histogram   slide

Histogram with relative densities   slide

Boxplot   slide

• Invented by John Tukey in 1970
• Many variations proposed since then, though the essential form and idea as remained intact.

Boxplot of wheat yields   slide

Violin plots   slide

Boxplots: Useful to identify extreme values   slide

Boxplots: Useful for comparisons across categories   slide

Violin plots   slide

Paul Krugman on Fiscal Austerity

What does this graph show?   slide

Source: https://www.nytimes.com/2018/11/02/opinion/the-perversion-of-fiscal-policy-slightly-wonkish.html

What did Paul Krugman say?   slide

"Here’s what fiscal policy should do: it should support demand when the economy is weak, and it should pull that support back when the economy is strong. As John Maynard Keynes said, “The boom, not the slump, is the right time for austerity.” And up until 2010 the U.S. more or less followed that prescription. Since then, however, fiscal policy has become perverse: first austerity despite high unemployment, now expansion despite low unemployment.

How could we better show the relationship between unemployment and fiscal austerity   slide