Reading 7: Estimation and Inference

Book 1: Quantitative Techniques

Probability Sampling

selecting a sample when we know the probabiliyt of each observation

Random Sampling

Each item is assumed to have the same probability of being selected

Simple Random Sampling

each data point has an equal opportunity of being selected

Nonprobability Sampling

not based on probability, sampling technique guided by either low cost and easy access to data or on the judgment of the researcher

Types of Probability Sampling Methods

Systematic Sampling

Stratified Random Sampling

Cluster Sampling

Systematic Sampling

selecting every nth member of a population

Stratified Random Sampling

separating a population into smaller groups (stratums) based on one or more distinguishing features, and from each stratum a random sample is taken

Where is stratified random sampling done?

Bond indexing

Cluster Sampling

creating subsets of a population and then randomly selecting different clusters

One-Stage Cluster Sampling

a random sample of clusters is selected and all the data in those clusters is selected

Two-Stage Cluster Sampling

random samples from each of the selected clusters is taken

Major difference between Cluster and Stratified

Stratified: homogeneous; creating similar groups based on a characteristic

• Systematically splitting up a population

Cluster: heterogeneous; each group contains a mix of many different characteristics

• Randomly splitting up a population

Which has a greater sampling error—One-Sample or Two-Sample Cluster?

Two-Sample

Convenience Sampling

selecting data based on ease of access

Judgemental Sampling

data selected based on the researcher’s experience and judgment

What does the Central Limit Theorem (CLT) state?

For simple random samples, the sampling distribution approaches a normal probability as the sample size becomes large enough

What’s the magic number in CLT?

n ≥ 30

Conceptually, what is the standard error?

The standard deviation of the sample means

Why is it hard to estimate the standard error?

Because we almost never know the population standard deviation.

Alternative methods for estimating standard error

Jackknife

Boostrapping

Jackknife

calculates multiple sample means, and each time one of the observations is removed from the sample