POS3713 Understanding Political Science Research Prof John B Ryan Midterm 2 February 24 2014 Study Guide Internal Validity A research design is internally valid if it provides convincing evidence of causality The experimental design rates highly on internal validity External Validity A research design is externally valid if it yields conclusions that can be generalized to other settings and individuals different from the ones studied The experimental design does not rate highly on external validity Good Research Designs Account for o The Unit of Interest Individuals states countries o Temporal spatial variation of interest over time between units Cross Sectional many units sampled over a single time period variation across space Time Series single unit sampled over many time periods variation over time o Threats to Causality History other events that affect the dependent variable besides the experimental stimulus Maturation processes within individuals that change the DV simply because time passes Testing the DV is measured at two points in time and the first measurement effects the second Selection each case has the same chance of being selected Regression to the Mean Experimental Designs o Field experiments Randomly assign people to groups but perform the manipulation in the real world Ex What is the best method of getting out the vote Door to door Phone calls Mail Random assignment of X but people s homes are not randomly distributed o Natural experiments Event outside the social scientist s control separates people into control and treatment groups Ex Newspapers and political information random assignment of X but people s homes are not randomly distributed Statistical Control Reliability the consistency or repeatability of a particular statistical measure Test Retest Alternative Form Split Halves Bias a form of inaccurate measurement in which the data consistently overestimate or underestimate the true value of an event Validity Most important feature of measurement An invalid measure measures something other than it was originally intended o Face Validity An indicator of a concept has face validity if it seems like it measures what it is intended to measure Content Validity Content validation is a rigorous process that forces the researcher to come up with a list of all of the critical elements that as a group define the concept we wish to measure Construct Validity The degree to which the measure is related to other measures that theory requires them to be related to Levels of Measurement the type of values that the variable takes on i e categorical ordinal continuous Nominal Ordinal Interval Ratio Interval is a variable that has numbers as its values the distance or interval between pairs of consecutive numbers is assumed to be equal Ordinal low medium high Clear ordering of variables Median Mode Ratio A variable that meets the criteria for interval variables but also has a meaningful zero point Nominal The Appropriate Descriptive Statistics for Each Level Mode The mode of a variable is the value with the greatest frequency Median The median is the value of the middle case when cases are ranked in order of their values on the variable The median can be measured only for variables measured at the ordinal interval or ratio level Mean The mean of a variable is the sum over all cases of the value of the variable divided by the number of cases The mean can be measured only for variables measured at the interval or ratio level Zero Sum Property o Least Squares Property o Expected Value Variance and Standard Deviation Standard Deviation Average difference between values of Y and the mean value of Y Square root of variance Variance The sum of the differences of the mean squared divided by the sample number minus 1 o Of both a population and a sample Distributions Bimodal distribution A distribution of opinions that shows two responses being chosen about as frequently as each other Unimodal distribution frequency distribution with one value clearly having a larger frequency than any other Skewed vs Normal
View Full Document
Unlocking...