Methods of Data Collection
What type of sample collection includes surveying everyone in the population of interest?
Census Survey
What is one of the first things that should be determined when planning a survey?
Identifying the population of interest
What are experimental units in an experiment?
Who or what treatment is imposed on
What two types of sampling methods lead to the most biased responses?
Convenience sampling and voluntary response sampling
What is an observational study?
A type of research strategy where data collected through a survey is observed and analyzed without imposing any treatment to determine a correlation or cause and effect relationship.
What's the difference between stratified random sampling and cluster random sampling?
Stratified: population of interest is grouped into homogenous groups, and the some of each group is (randomly) selected to take the sample
Cluster: population is grouped into heterogeneous groups, and all of some of these groups are (randomly) selected to take the sample
After identifying the population of interest (who it consists of, how large it is, etc), what is the next step to take when planning to conduct a survey?
Deciding which survey method to use (SRS, stratified, clustered, systematic, etc)
What is it called when either the subjects or the experimenters don't know about the treatments in an experiment?
Single Blinding
What type of bias is being described here?
To research student opinions on the lunch options at their school, a group of students interviewed student government members from each grade.
Undercoverage bias because some people (people who are not members of student government) are less likely to be chosen--in this case there's no chance of them being chosen at all.
Why is random sampling so important in experiments AND observational studies?
Experiments: it allows us to conclude that a treatment causes changes in a response variable
Observational studies: it allows us to generalize our conclusions to the population we sampled from
What type of sample is being described here: all members of a population of interest are numbered, and a number is randomly selected. Starting at that number, every 10th person is selected to take a survey.
Systematic Random Sampling
How do you create an SRS (simple Random sample)
1. Label individuals
2. Randomize (using a random number generator or some other method of randomization)
3. Select individuals using this method of randomization
What are the characteristics of a well-designed experiment?
- comparison (2 or more treatments assigned)
- randomization
- replication
- control (other variables are constant)
What is an example of nonresponse bias?
When some of the people included in a sample either decline to participate or are unable to participate
True or false:
A placebo effect can occur in an observational study.
False - placebo effects occur when a fake treatment works. Therefore it would be impossible for this to occur in an observational study since no treatment can be imposed (or it would be an experiment)
What is the most common type of sampling used?
SRS - simple Random sampling is one of the easiest type of sampling to conduct that still reduces bias
What is perhaps THE MOST important thing to do when creating and distributing a survey?
Incorporating + ensuring randomization to lower bias
How do you incorporate random assignment in an experiment?
First label subjects / units, then using a RNG (random number generator) or a different means of randomization, assign treatment to units
What type of bias occurs when there are issues in the data gathering process?
Response bias--this can occur when people lie in their response or if the wording of the question itself is biased.
What is being described here: an experiment or an observational study?
A researcher wants to determine whether students at a certain school prefer the SAT or the ACT. She randomly selects 35 seniors and 35 juniors to survey on this topic.
Observational study because no treatment is imposed - it's simply a study to determine if a behavior is prevalent in a population
What is being described here: an experiment or an observational study?
To determine whether or not drinking coffee increases productivity, a group of scientists selected 20 volunteers, and randomly placed them into one of two groups. The first group of 10 was given one cup of coffee to drink every morning for a week, and their productivity was measured. The other was restricted from drinking coffee for a week, and their productivity was also measured.
An experiment because a treatment is imposed
What is important to a sampling method in order for it to produce the best estimates?
Low bias and low variability
What is it called when there's a variable that is related to the explanatory variable that influences the response variable?
For example: there is a high correlation between ice cream sales and shark attacks, but high temperatures lead to more ice cream sales and to more people swimming in the ocean. (Hint: what is the temperature in this situation?)
Confounding variable--it's important to remember that these exist when conducting experiments, so therefore we cannot always draw causation from strong correlation.
What (of course) would lead to the most bias occurring in an experiment if it is not incorporated?
Randomization!
What is the main difference between an observational study and an experiment?
Treatment is imposed in an experiment, while no treatment is imposed in an observational study.