Public Policy Analysis MPA 404 Lecture 9 Previous

Public Policy Analysis MPA 404 Lecture 9

Previous Lecture Ø Quantitative methods for analyzing a policy. Ø What is intended to be done with these and a bit of background on its development. Ø Frequently used quantitative techniques Ø Univariate, Bivariate and ANOVA techniques Ø Quant's and the great recession

Ø Regression Analysis: Probably the most widely used technique in analysis of policy. One of the main reason for this is the fact that predicting an outcome of a policy is very much in demand, and regression analysis is considered the best technique for it. Also, it is considered a good test for predicting quality determinants of a policy. Regression analysis can either be based on simple regression or multiple regression. Ø An example: it is used to predict or forecast the economic growth rates. The economic growth rate (measured of changes in real GDP) is a tremendously important number for policymakers and other groups of people (like investors). It has substantial repercussions for a well being of a country. Ø A very important step before going ahead with regression is to satisfy some of the assumptions for carrying out this test, otherwise the results will be biased or spurious. Ø Time Series Analysis: As the name can tell, this statistical analysis is related to patterns that emerge over time. It normally is used in long term studies, but within these, it has the power to predict the causation between variables

(dependent and independent) over much shorter time spans. For example, one of the areas in which the government is most interested in is the relation between unemployment and its effects on growth. Time series analysis lets patterns emerge over time, and within that time period one can also get a statistical picture of relation between variables over a quarter or half-year. Ø Based on the patterns gauged from long term time series and analysis, policy makers can get a better handle of historical causation variables and lessons for future program implementation. It can also be used forecasting based on past trends (weather people do that a lot). Ø For use of time series analysis, it is critically important that the historical data is of good quality and thorough. Otherwise, results can provide misleading trends and thus prove to be a problematic for purposes of present day policy making. Ø Six steps in the time series test (page 358).

Ø Event History Analysis: A sub-series of time series analysis that is more geared towards rare events in a time series, and why some individuals/groups/organizations tend to be more affected by these events. Ø What is different about this test from time series analysis is that certain unique terms/sets are used to test the hypothesis. These include a risk set (who is likely at risk), survivor set (measures the decline of risk over time) and a hazard set (the frequency of occurrence of an event at a certain time). Ø Began to be used in the 1970’s for statistical inquiry into matters related to international affairs (like occurrence of diplomatic events or rare events like war). Ø Factor Analysis: This one is related to the intangible variables like trust, deception, satisfaction, envy, etc. It is based on the assumption that there are some specific, unobservable factors that underlie the relation (as measured by statistical tests) that is not visible to the observer by just looking at the numbers. Once these factors or variables are taken into consideration, properly adjusted in the model and tested, then we get a

much more vibrant result. Ø It is of two types; Exploratory and Confirmatory. The difference between the two basically boils down to the assumption of effectiveness or strength of unobserved variables. Confirmatory analysis is more towards unobserved variables, and vice versa. Ø Path Analysis: To put it in simple words, path analysis is the test for gauging the effect of an intermediary (or a third factor between the independent and dependent variable). Normally what we have seen with the above tests is that they are all concentrated on the link between dependent and independent variable. This test differs in the sense that it takes into account the fact that the independent variable may affect an unknown (path) variable that in turn significantly affects an independent variable. Path variable measures this ‘indirect’ relation between the two variables. Ø The earlier example of vote polling in Baluchistan. There is the path variable (security) that affects both citizens’ participation due to government’s public policies.

Game Theory: Ø A few helpful concepts from Economics