ch38 how economists use data

Data

Factual information often quantitative that provides the basis for reasoning and discussion

econometrics

Subfield of economics that develop tools to analyze data

Randomized control trial

Experiment where researcher randomly divides subjects into groups, treats groups differently and compare their outcomes

Experimental data

Data that comes from researchers running a randomized controlled trial

Observational data

Comes from researchers observing the world as it presents itself

Challenge 1 observational data: Confounding variable

An omitted variable that can mislead the researcher bc it is related to the variables of interest

Challenge 2: reverse causality

Situation where researcher confuses the direction of influence between 2 variables

What are 3 types of data

Cross sectional , time series, panel

Cross sectional data

Data that presents info about multiple subjects (such as people, firms, or nations) at a given time

Time series data

Data that present information about a single subject (such as a person, firm, or nation) at various times

Panel data

Data that presents info about multiple subjects (such as people, firms, or nations) at various times

What do economists do with data?

Describe the economy ; Quantify relationships ; Test hypothesis ; Predict the future

parameters

Numerical values that govern the strength of the relationships among variables in a model

Statistical model

Mathematical representation of the process that generates the data

Residual

"Represents the many forces that may influence DV but excluded from the model

This is assumed to be zero on average and uncorrelated with the independent variable"

Linear regression

A statistical model where the dependent variable is linearly related to one or more IV plus random residual

Ordinary least squares

Stat method for estimating parameter values by minimizing the sum of squared residuals (best fit line)

Economists find the data relevant for the issue at hand, posit a statistical model that can plausibly explain data, estimate the models parameters using methods such as ordinary least squares. Using estimated model they can reach quant conclusions

Sampling variation

Variability that arises bc diff random samples lead to somewhat diff estimates

Standard deviation

Measure of variability across observation

Standard error

Using sd and sample size, measure of uncertainty associated with a parameter estimate that results from sampling variation

Margin of error

a measure of uncertainty in a sample-based estimate, indicating the range within which the true population value likely lies

Multiple regression

Linear regression w more than 1 iv

When an omitted variable (ability as measured by IQ in our example) directly influences the DV (wages) and the omitted variable is correlated with the IV (schooling) OLS yields misleading results

The OLS estimate confounds the effect of the IV with the effect of the omitted variable. One way to deal with this is to include the previously omitted variable in a multiple regression

Natural experiment

A chance event that causes variation in the data similar to that generated by a randomized controlled trial

Statistical approach to measure causal effects in data from natural experiments

Instrumental variables methods ; Find some random variable called the instrument that meets two conditions: The instrument is correlated with the IV of interest and Instrument does not affect the DV other than through its effect on the IV