ECON 257D1 Midterm 1

Main types of data

Primary, secondary

Secondary data

information collected by someone other than the user, sometimes for another purpose

Primary data

information collected by the user for a specific purpose

Correlation

two variables move together

Causation

movement of one variable forces movement of another

Reasons why there can be correlation without causation

Third variable problem, directionality (reverse causation), coincidence

Third variable problem

omitted variable bias, outside variable forcing movement of both

Chicken and egg (reverse causality)

Which causes which

How to prove causation

randomized controlled trials (where all variables except one are constant, hard to do in economics in the real world), temporal sequence (prove one happened before the other)

GDP

gross domestic product, market value of all final goods and services per year, q x p

PPP

purchasing power parody, eg Big Mac index, currency translation more accurate than exchange rate

real vs. nominal GDP

real GDP uses an index year/holds inflation stable, nominal is in current year

continuous variables

variables (like income) that can be any value

discrete variables

1/0, eg yes/no

ȳ

mean of data set

comparison of means

ȳ-ȳᶜ

population variance equation

𝜎²=1/n Σ(yᵢ-ȳ)²

sample variance

s²=1/n-1 Σ(yᵢ-ȳ)²

standard deviation equation

√var Y

standard deviation intervals

if normal, 34.1 is 1SD 13.6 is 2SD 2.3 is 3SD

z-score, units

(yᵢ-ȳ)/SD, unitless

shaky

lots of spread

good estimator is

unbiased, efficient

unbiasedness

ȳ is centred on μ, E(ȳ)=μ

population mean

μ, E(y)=μ

ū

average deviation

efficiency

ȳ is close to μ

what is 𝜎² the population variance of

population variance of yᵢ

population variance of ȳ (variance of means)

𝜎_ȳ², E((y-μ)²)

covariance

2E(x-μ_x)(y-μ_y)

covariance is negative, positive, 0 if …

positive if upward correlation, negative if downward, 0 if cloud no correlation independent

𝜎²_x+y

𝜎_x²+𝜎_y²

what to always do before starting a do file

clear

rename a variable

ren varog varnew

label a variable

label variable var1 “"

get rid of rows of data

drop in #/#

revalue variable

replace var=”” if var==””

convert variable from strings to numbers

destring var, replace

red data

strings

black data

numeric

find median

centile var, centile (50)

sort variable into above and below median

gen var2=# if var1>= `r(c_1)’&var1!=.

replace var2=# if var1<=`r(c_1)’

missing value

black .

Tell me the quantities of each value and percents

tab var

Tell me the mean, min/max, SD, # of observations, variance, median, quarters, skew

sum var, detail

save something

save “\” , replace

format for data, graph, table

.dta, gph, .rtf

make a graph

collapse (mean) mean_var1=var1, by var2

graph twoway (bar mean_var1 var2, barw(#)), xlabel(# ““ etc) xtitle(““) ylabel (# ““ etc) ytitle(““) title (““)

graph save

make a table

eststo var1:estpost summarize var2 if var1==#

esttab var1 var2 using “\.dta”, plain cells(“mean(label(Mean) fmt (0))” “sd(label(SD) fmt(0))” “count(label(N) fmt(0))”) mtitles(“ “ “ “) noobs replace label title()

import first row as variables

import ““, firstrow

how to import sas

import sas using ““

browse a variable

browse var

number of observations

count var

general stats about data (string/numbers, missing)

codebook var

look at distribution

hist var

sort variable from smallest to largest

sort var

bias directionality

can be product positive if all possible third variables end up positive or vice versa, or product neutral, it is positive and overstates bias if a third variable causes both things to happen more, negative understates bias and causes one, but reduces the other