Midterm--Research Methods

Why is a sample statistic only an estimate of a population parameter? That is, why do we need to calculate an error margin and a confidence interval? - correct answer It's only an estimate because it varies from sample to sample. By calculating an error margin and a confidence interval we can make a certain level of inference from our sampling results. List and define the levels of measurement - correct answer 1) Nominal- Strictly categorical; places units of analysis into mutually exclusive and exhaustive categories; these categories have no order. 2) Ordinal- Places the units of analysis in categories that have order; categories are mutually exclusive and exhaustive; you cannot measure distance; typical relations are "greater than, less than", "excellent" to "terrible. 3) Interval- Occurs when able to rank a set of observations and the exact distance between each is known; variables here have an arbitrary point; we can map out units of analysis onto the number line; we can measure distance, but we cannot calculate ratios. 4) Ratio- The level of measurement is isomoprhic to a positive number line and has a non-arbitrary 0 point. This level of measurement appropriate for variables having a natural, absolute, fixed zero point (think age, time, length, weight); allows for the most precise measuring instruments; can map units of analysis onto the positive number line. Provide an example of nominal measurement - correct answer Measuring by religion (ex. Catholic, Protestant, Jewish, Muslim, Other) Provide an example of ordinal measurement - correct answer Measuring by educational level (ex. H.S./GED, Some College, Undergraduate, Some Graduate, Graduate Degree) Provide an example of interval measurement - correct answer Measuring by NET gains/losses (ex. Income, intelligence, voter turnout, crime rates) Provide an example of ratio measurement - correct answer 1) Measuring by order (ex. Time, weight, age, length) Define central tendency and dispersion - correct answer 1) Central Tendency- Is the best summary value of the variable. 2) Dispersion- A measure indicating how spread out the values of the variable are. For each level of measurement what is/are the appropriate measure(s) of central tendency and of dispersion? - correct answer 1) Nominal: Central Tendency: Mode; Dispersion: Proportion of outside the mode. 2) Ordinal: Central Tendency: Median, Mode; Dispersion: Range Measures, Proportion of outside of the mode. Interval/Ratio: Central Tendency: Median, Mode, Mean; Dispersion: Variation, Variance, Standard Deviation, Range Measures, Proportion of outside of the mode Define internal (causal) validity and discuss the criteria for inferring causation - correct answer Defined as finding out Did X cause Y. There are three necessary criteria to show internal/causal validity: 1) Correlation/covariation (does X and Y move together); 2) Time order/temporal precedence (does X come before Y); 3) Non-spuriousness (X and only X causes Y). define external validity. what is required of a research design for it to have external validity? - correct answer External validity makes a generalization to the population and other settings. It attempts to take what has been measured in a certain setting and generalize it to other circumstances and to other populations. The research design must have a probability sample and a "real world" data collection setting to be externally valid. Discuss the internal and external validity of experimental designs, i.e. which type of validity (internal or external) is an experimental design strong on and which type of validity (internal or external) is it weak on, and why? - correct answer In experimental design, internal validity is strong due to the fact that researchers can establish the criteria for inferring causation, conduction comparisons between groups, and control/manipulate aspects of the experiment as they see fit. In experimental design, external validity is generally weak. There is no generalization or a probability sample. Additionally, it may not have a "real world" setting which in turn impacts the behavior of participants. Discuss how and why a correlational design differs from an experimental design in terms of internal and external validity? - correct answer Correlational Design looks at the relationship between property and disposition. Correlational design is weak when it comes to internal validity as non-spuriousness and time order are problematic. It has strong external validity as it has probability samples and a real world setting and data collection. give an example of a research question where you would use an experimental design - correct answer Does the use of sugar alcohols increase a person's risk for Type II diabetes? give an example of a research question where you would use a correlational research design - correct answer Does a person's income