Sampling

What was the purpose of the 1970 Military Draft Example?

Attempt to do randomness for sampling, but failed. Randomness was not ensure when all capsules were combined, because they were not mixed together. Just reverse order starting with December on top, January on the bottom.

Barriers of Bias in Sampling

Language barrier, culture barrier, politic barrier, and others..

Problems with Cluster Sampling?

Little diversity exists between participants. Goal is to not find a pattern which would prevent creating bias.

Example of Cluster Sampling

Maybe you want to find voting habits of undergraduates at the top 10 universities. First you would randomly choose 3 universities (Random Number Table), then you would cluster a group. From that group, you would use Random Number Generator for the number of students you want.

Cluster Sampling does not assume

complete sampling frame. That is why we use clusters first. No sampling frame.

Sampling at Work: 1970 Military Draft (Vietnam War - Not Voluntary)

A number (1-366) representing each day of the year was placed in a capsule, and placed in a larger bowl. Capsule was drawn our representing a day, and all men born on that day between 1944-1950 were drafted.

Literary Digest Poll on 1936

A poll that went wrong in history. Choose Participants from phone directories and automotive registries.

Probability Sampling

ALL RANDOM; methods of identifying study samples that adhere to assumptions of probability theory.

Non-Probability Sampling is defined as

Any sampling method that does not adhere to the assumptions of probability theory. Based on convenience, NOT RANDOM. Cannot be generalizable!

Example of Purposive/ Quota Sampling

Asking a question giving option A) and option B), then ask only select few from either group.

NP Samp: Volunteer Sampling

Asking individuals to volunteer as participants

Example of Snowball Sampling

Asking survey questions to a patient diagnosed with rare form of lukema. It may be hard to find people with this specific diagnosis, so patient may know friends / family with this specific diagnosis.

Example of Volunteer Sampling

Asking whoever wishes to reveal who they want to vote for, please stand up. Or raise your hand if you wish to participate in this survey sample.

Example of Confidence Interval

Between 75-85% of registered Democrats will vote for Hilary Clinton in the 2016 presidential election. [We want out range to be small].

Problems with Systematic Sampling

Bias Information

P Samp: Systematic Sampling

Choosing every Nth person in a population to create your sample.

Example of Convenience Sample

Completing a survey of your closest friends, or completing a survey of people sitting next to you, or professor picking students from the first two rows within the class.

Surveying Methods: Nationwide Survey

Conducted online with people who have been recruited by knowledge networks to participate in research.

P Sample: Cluster Sampling is defined as

clusters (locations) are randomly sampled first, with individuals then randomly sampled from the cluster.

P Samp: Stratified Random Sampling

Dividing the population along a chosen characteristic and then randomly sampling from each group.

Equal Probability of Selection means

Every individual in the population has an equal chance of being selected for the study.

cluster sampling

clusters are randomly sampled first, with individuals then randomly sampled from the cluster

Calculating Sampling Error: Confidence Level

How sure we are that our confidence interval is accurate.

Example of Confidence Level

I'm 95% sure that range 75-85% of registered Democrats will vote for Hilary Clinton in the 2016 Presidential Election. [We want out level to be big].

Example of Random Sampling

If you have a roster of the class, you would use a random numbers generator to pick participants.

Random Selection means

Individuals chosen by chance. Helps researchers avoid sampling biases.

What went wrong in the 1970 Military Draft?

Men with birthdays in December were drafted at much higher frequency. Why? Reverse order of months of years, so December was placed on top.

Problems with Purpive/ Quota Sampling

Not random, might be bias.

Surveying Methods: Random Digit Dialing

Random Digit dialing of all numbers to reach a specific ethnic group in a specific Los Angeles Community.

Problems with Convenience Sampling

Sample size, demographic bias, information inconsistent with population based on sample chosen.

P Samp: Simple Random Sampling

Sampling Units Randomly selected from a population; assumes the sampling frame contains all members of a population.

Example of Cluster Sampling 2

Say we want to sample 1000 people from the U.S. population on voting. Our cluster would be 50 states, which we would randomly select 5 states, followed by randomly selecting 200 people from each step.

Example of Multi-stage Sampling

Say we want to sample 1000 people from the U.S. population on voting. Our cluster would be from 50 states, which we randomly select 5 states (Random Number Generator). From those 5 states, we select 5 counties (Random Number Generator). From each county, we randomly select 5 cities or equivalent. From each city, we select 8 individuals.

NP Samp: Purposive / Quota Sampling

Segmenting sample into different kinds of people, and then selecting individuals who fit those characteristics.

NP Samp: Snowball Sampling

Start with one participant, and snowball / branch out to their friends/ references.

Sampling Interval

The standard distance between elements (Population Size/ Desired sample size). Ex: 240 students / 12 (sample size) = Every 20th person is a participant.

Example of Systematic Sampling

Trying to find out political views of families within the neighborhood. You can tally how many blocks / how many houses. Every 9th house, would be a participant.

Problems with Volunteer Sampling

Volunteers may have specific qualities that is different then the population.

Example of Stratified Random Sampling

Voting behavior at Rutgers University, separated by Race. Having a complete sampling frame detailing what percentage of race the population is at RU. Determine what your population size is -> Ex: 200 * each percentage (race), to determine how many of each race. Then take a random number generator to pick from the list. Different then quota sample bc - randomly selects.

When is Snowball Sampling useful?

When trying to get participants from difficult populations or invisible markets.

NP Samp: Convenience Sampling

Who is conviently available as participants

Calculating Sampling Error: Confidence Interval

a range of values within which our population parameter is estimated to lie.

Sample is defined as

a subset of a population.

Benefits of Probability Sampling

a) Enhanced likelihood of representatives b) Ability to generalize to the broader population c) Accomplished through our ability to estimate sampling error

What when wrong with Literary Digest Poll of 1936?

a) bias b) Last minute voters c) popular vote vs. electoral vote d) something about respondents

example of cluster sampling

big 10 institutions- randomly select undergrad students

Sampling Error is defined as

degree to which sampling characteristics differ from the population's characteristics.

Population is defined as

every possible element of pre-defined aggregate that could be studied.

Representativeness leads to ___________ and _____________

generalizability and validity (match data with truth)

Representativeness refers to

how closely a sample matches its population in terms of the characteristics we want to study.

Random Sampling replies on

random numbers to identify sample. The most straight forward, but not necessary the most accurate.

Sampling Frame is defined as

the 'list' from which all members of a population are sampled.

multistage sampling

sample locations/ groups at multiple points before finally sampling individuals

P Sample: Multi-stage sampling is defined as

sample locations/ groups at multiple points before finally sampling individuals. Similar to cluster sampling, but involves multiple clusters.

for probability sampling, max error depends on

sample size, diversity of population, and confidence level

example of multistage sampling

we want to sample 1000 people from the US population- select 5 states, from those 5 select 5 counties, each county, select 5 cities, etc.