Draft paper

· Size

· Increase shrinks confidence interval

· A range in which we are confident that a certain percentage of surveys will yield accurate representativeness

· Increase increases confidence levels

· A confidence that a certain percentage of surveys will fall within a confidence interval

· 95% certain that 48 to 52 percent of voters will vote for Gibson

Sampling Designs

· Simple random

· Systematic

· Usually first chosen randomly, then a system is used

· Every tenth element, 10,15,25, 30,45,50

· Stratified

· Ensuring selection of subsets

· Equal number for Random selection of men, random selection of women

· Disproportionate

· Selecting more out of a more rare subset

· UW-P Need to select 70% of people of color, 15% of Whites

· Cluster

· Selection of groups or units, then selection of elements from each group

· Random selection of 12 cities. Random selection of officers from each of their departments.

Obtrusion

· Experiment

· Did we control all the variables necessary to isolate the dependent and independent variables?

· Natural Research

· Did we avoid introducing variables that lead to occurrence of dependent variable?

Elements of Classic Experiment

· Independent Variable (IV)

· Causal variable introduced by research

· Dependent Variable (DV)

· Variable measured for change caused by introduction of IV

· Experimental Group

· Subjects that experience IV

· Control Group

· Subjects that do not experience IV

· Random Selection – group equivalency

Classic Observations

· Measurement can be operationalized in any particular way

· Pretest

· Measurement made before IV introduction

· Posttest

· Measurement made after IV introduction

· Double-Blind

· Researchers are unaware of IV and/or observation test

Classic

Control Group O1 O2

Experimental Group O1 X O2

· Compare first observations: O1s Same?

· Introduce Independent Variable: X

· Compare second observations: O2s Different?

Threats to Internal Validity

· History

· Something happens during experiment

· Maturation

· Subjects just change, experiment or not

· Testing

· Subjects change due to observation

· Instrumentation

· Observation technique may differ

· Statistical Regression

· Extreme scores naturally normal out

· Selection Biases

· Subject selection not simple random

· Experimental Mortality

· Subjects drop out of experiment

· Causal Time Order

· Which variable did the causing?

· Diffusion or Imitation of Treatments

· Subjects share the treatment with control group

· Compensatory Treatment

· Control group unintentionally gets compensated in other ways

· Demoralization

· Control group members give up

· Compensatory Rivalry

· Control group tries harder

Threats to external validity

· Sample Size

· So small generalization is weakened

· Control vs Natural

· Setting is so unnatural, does the causation occur in a natural setting?

Variations from classic designs

· Posttest Only

· Factorial – multiple treatments/treatment groups

· Cohort -- Group selection based on natural grouping ie Class of 2014

· Time-series

· Simple interrupted

· Nonequivalent groups

· Removed treatment

· Switching replication