• Population: all subjects of interest
  • we use statistics to learn about population and the entire group of interest
• Sample: Subset of the population
  • Data collected for sample because we cannot measure all subjects in population

• Researcher observes response/explanatory variable.
• Nothing done to subjects

• assigns subjects to experimental conditions and observing outcomes based on response variable
• randomized experiments = subjects randomly assigned to treatmetns
• randomization reduce potential for lurking variable
• gives more control over outside influences
• only experiment can establish cause/effect
• not always possible due to time/ethical factors

• list of subjects in population from which sample is taken
• ideally lists the entire population of interest

• determines how sample is selected
• ideally gives each subject equal change of being selected to be in sample

• each possible sample of n size has same chance of being selected

• Number subjects in sampling frame using numbers of same length
• select numbers of same length from table of random numbers
• Include in the sample those subjects having numbers equal to the random numbers selected

• tell us how well the sample estimate predicts the population percentage
• Simple random sample of n subjects will have margine of error of 1/(√n) *100%

Convenience Samples (Poor ways to Sample)

Volunteer Sample

• Convenience sample: type of survey sample that is easy to obtain
  • Unlikely to be representative of population
  • Often severe biases result from sample
  • Results apply only to observed subjects
• Volunteer Sample: Most common form of convenience sample
  • Subjects volunteer for sample
  • volunteers do not tend to be representative of entire population

• Having sampling frame that does not represent some parts of population

• Bias resulting form sampling method
  • ex. nonrandom samples or undercoverage

• Occurs when some sampled subjects cannot be reached or refuse to participate or fail to answer some question

• Occurs when subject gives an incorrect response or question is misleading

Large sample does not guarantee unbiased sample

• ID population of all subjects of interest
• Construct sampling frame that attempts to list all subjects in the population
• Use random sampling design to select n subjects from sampling frame
• Be cautious of sampling bias due to nonrandom samples

• Subjects of an experiment
• entities that we measure in an experiment

• specific experimental condition imposed on subjects of the study
• Treatmetns correspond to assigned values of explanatory variable

• Defiens groups to be compared with respect to values on response variable

• Outcome measured on subjects to reveal effect of the treatments

• Control/comparison group
• Randomization
• Replication
• Blinding the Study

• Allows researcher to analyze effectiveness of primary treatment
• Control group typically receives placebo

• Eliminates possible research bias
• Balances comparison groups on known as well as on unknown variables

• process of assigning several experimental units to each treatment
• allows us to attribute observed effects to the treatments rather than ordinary variability

• subjects unaware to treatment received
• double blind experiment controls response bias from respondent and experimenter

• observed differences in experiment is larger than what would be expected just by chance

• Divide populations into separate groups called strata
• Select simple random sample from each stratum
• Combine samples from all strata to form complete sample
• Pro:
  • can include in sample enough subjects in each stratum you want to evaluate
• Cons:
  • must have sampling frame and know into which stratum each subject belongs

• Divide population into large number of clusters (ex. city blocks)
• select simple random sample of the clusters
• use subjects in clusters as sample
• Preferable when:
  • reliable sampling frame is unavailable
  • Cost of selecting SRS is excessive
• Disadvantage
  • need larger sample size than SRS in order to achieve particular margine of error

• Attempts to take a cross section of a population at the current time

• Observation study
• looks into past
• Case-control study

• Observational study
• follows subjects into the future

• uses single experiment to analyze effects of 2 or more explanatory variables on the response
• Categorical explanatory varibles in experiments are called factors

• subjecst receiving 2 treatments are somehow matched
• ex. same person, husband wife, 2 plots in same field

• Same individual used for 2 treatments

• Block = set of experimental units w/ one or more matched characteristics
• Randomized block design = random assignment of experimental units to treatments is carried out separately w/i each block. 
• elimiinates variability in response due to blocking variable
• matched pairs design = special case of RBD