Scopes and Methods, Unit 2

Population

The entire group of people, places, events, or other cases that you want to study

People in a country, conflicts in a region, election officials in a government

Parameter

A fixed value that describes something about a population

Often the average, median, proportion, or standard deviation of a variable.

How can we measure a population’s parameter?

1). Measure each and every unit of the population (taking the average)

2). Measure some units in the population, then make an informed estimate about the rest

Sample

A selected set of units from a population collected with the goal of understanding more about the population

Statistic

A fixed value describing something about the sample with the goal of estimating a population’s parameter

What’s the goal?

Obtain a sample from a population in order to measure a statistic that lets us estimate some unknown population parameter

Reliable

Gets us a similar estimate every time we sample

Valid

Reflects the true population parameter, not bias

Law of large numbers

The more cases that we sample from a population, the closer the sample average will be to the true population average

Coverage bias

When your strategy for selecting units omits some of the population

Sampling frame

A list from which you select units to sample

• May not cover the whole population

Non-response bias

When certain demographics are less likely to answer

Types of Non-random sampling

• convenience sampling

• quota sampling

• snowball sampling

Random Sampling

• Simple random sampling

Convenience sampling

Just sample whoever you can reach

Low quality, bias, don’t trust

Quota sampling

Sampling whoever you can reach but attempting to make the sample resemble the population

Simple random sampling

Just sample from the full population at random

Can get a good estimate if you can truly sample at random and get a random sample size, but might be costly, finding the right frame is difficult, and what happens if there are large differences between population subgroups?

Stratified sampling

Divide the population into subgroups based on variables important to our analysis (age, race, gender, income), sample randomly from each subgroup

Cluster sampling

Divide population into clusters with heterogenous units, samples randomly