Data Collection

Simply observes and measures a variable of interest, doesn't have any control of the situation.

Pros

• Can detect associations (might or might not have meaning)

Cons

• Can't establish a cause and effect relationship.

Designed experiments that deliberately impose a treatment on individuals and record their responses.

they compare the recorded responses vs. a control group, placebo, or other

Goals of this study:

• Replication
• Randomization
• Control of Error

Pros

• Can analyze cause variables and influence responses.

Cons

• Cannot be done when the variables cannot be controlled.
• Cannot apply to come studies due to ethical or moral reasons.

Simple Random Sampling (SRS)

• Everyone in the total population size N has an equal change of being selected for our sample n
• Subjects selected cannot be sampled twice
• Ex. SRS of size n=2 out of total population N=4

Systematic (Probability) Sampling

• Uses chance to select a sample, based on known selection probabilities.

Stratified Sampling

• Dividing a population into subgroups and then sampling equally from those subgroups

Cluster Sampling

• Dividing a population into clusters and then randomly selecting all individuals from them
• Generally the clusters already exist (Ex. the rows of seats in a classroom and who sits where)

• Factors - measured aspects that we are interested in
  • Levels - specific values of factors
  • Treatment combination (interactions) - with multiple factors, a specific combination of levels
• Bias
  • measurement -- issues with the tools
  • sampling bias -- not sampling the right group
    • Voluntary response samples -- samples where the subject can select whether or not to answer
    • Convenience sampling - only sampling people that are coinvent
    • Incentivized sampling
• Populations and Samples
  • Population - consists of all subjects or items of interest. It is the group being studied.
  • Sample - a group selected of the population to participate in our experiment or to gather data
• Blinding - the subjects of the study don't know if they are a control or a treatment group
  • Double Blinding - neither the subject nor the admins know which group is which
• Time Terms
  • Retrospective - a study that analysis past data
  • Prospective - a study that deals with current data
  • Longitudinal - study that last a long period of time
• Types of studies
  • Cross sectional study - surveys that collect data at one point in time
  • Case-Control
  • Cohort study