# 4 Random sampling, generalization, and statistical estimation

Far better an approximate answer to the right question … than an exact answer to the wrong question.18

Aside from hypothesis testing, one of the most common uses of statistical inference is to estimation of unknown parameters using sample data. Polling is one application where statistical estimation is used. For example, Gallup and the Pew Research Center are organizations that use statistical estimation to provide snapshots of public attitudes and opinions on topics from politics and the economy, to social awareness and health and well-being. The results of their polls are seen on a daily basis in almost every newspaper, news blog and website across the world.

Statistical estimation is used by more than pollsters. Biologists, social scientists, and medical researchers use statistical estimation to quantify population characteristics. For example, each year the Minnesota Department of Natural Resources estimates the populations of various species of animal, bird, and fish. These estimates are used to help set hunting and fishing regulations, as well as to allocate resources.19

### Outline and Goals of Unit 4

The following schematic outlines the course readings, in-class activities, and assignments for Unit 4.

Unit outline
4.1     🔨   Crazy in love
4.2     📖   External validity evidence and random sampling
4.3     📖   Validity evidence and inferences
4.4     🔨   Comparing study designs
4.5     🔨   Monday breakups revisited
4.6     📖   Estimating, compatibility intervals, and margin of error
4.7     🔨   CEO compensation
4.8     🔨   Comparing estimates from random samples
4.9     📖   Uncertainty and bias
4.10     🔨   Summative activity: Montana YRBS
4.11     📖   Unit 4 summary

In the readings, course activities, and assignments in Unit 4, you will learn why random sampling helps provide validity evidence for generalizing results to the population (external validity evidence). You will also learn how to make estimates about a population parameter using sample data, including accounting for uncertainty in the estimate.