Research Methods Week 10?
Research Methods Week 10? Pols 201
Popular in Research Methods in Political Science
Popular in Political Science
This 5 page Class Notes was uploaded by Lauren Jones on Monday April 4, 2016. The Class Notes belongs to Pols 201 at University of Tennessee - Knoxville taught by Adam Eckerd in Winter 2016. Since its upload, it has received 32 views. For similar materials see Research Methods in Political Science in Political Science at University of Tennessee - Knoxville.
Reviews for Research Methods Week 10?
Report this Material
What is Karma?
Karma is the currency of StudySoup.
Date Created: 04/04/16
31 March 2016 Descriptive Statistics Regression o All based off same premise Explain why variables vary Test theories o Mostly we have been encouraging, variable x and variable y Prediction Measuring effect sizes Perform Valid comparisons o Get probability look at the nature of relationships, see what predicted medical expenses would be, if they vary, why? Y=mx+b X is the intendent, b is slope, y is the dependent o B is the regression coefficient Represents the amount of change expected E o Reason for y hat If the relationship is completely deterministic, then every point lies exactly on the line But this is never case o Points always fall above or below, even though strong correlations Y hat is the predictive value of Y hat does not equal Y o Y hat would be that value for every given x Ei o Half the time it will be positive, and half will be negative Minimize the distance, w Minimize the Ordinary Least Squares o Constant Sometimes meaningless, determine the meaningful need of things Regression o Estimates what happens in the population Each point is a data point from the sample Regression and Correlation o Coefficient of determination Means that are squared that is the proportion of variance in Y determined or explained by x Tells us about variation, we’ve got lots of variation o The squared tells us the proportion, the closest to one, the more of the variation we have explained You can interpret using the x line. o X as b Hypothesis test o A slope of 0 is a horizontal line As x increases, y does not change You can measure relationships better by using regression o Type one and Type two errors Occur when choosing a level of significance Specify the hypothesis Determine alpha value o Usually .05% Calculate sample statistic Compare statistic to hypothesized parameter Calculate the t value Find a p value Make a decision of H0 o Then state! The unlikeliness We can infer that more education leads to an increase in income Confidence Interval Method o 95% confident that a sample was derived from a population where there is positive significant relationship Specify hypothesis Calculate b Calculate confidence interval for b Determine if 0 is within the interval You get the result as before o Don’t care about specific nature if we take any two communities We expect higher per up Practical versus statistical significance o Ten extra dollars 835 Actually look at the value of the slope Only meaningful when it is. o Modifiable areal unit problem Depending of area, could change if you change the area o Simpson’s Paradox You can miss relationships vs certain groups when aggregating o Regression Fallacy Making decisions based on extreme observations Results will tend to regress to the mean Example of one used for everything o Ecological Fallacy Based on an aggregated trend At the individual level, does not hold o At state level, it looks like one Creating good charts General Principle of Graphic Display o Efficient display of meaningful and unambigious data Define what the numbers represent o Deceptive choices in design can distort numbers and relationships that are trying to be represented Life factor statistic Numerical measure of the data distortion A chart is to simplify numerical comparisions Common errors o Include elements in the display that have nothing to do with the comparisons o Components of a chart Chart titles May state the conclusion o Subtitles contain what the x and y mean Axis Titles and Labels Repeat what is clear from the main title or labels Axis Scale and Data Labels Te value or magnitude is defined by these labels Legends Used when chart has more than one data series Gridlines Little ink as possible not ot overwhelm Sources Specifying source of data is important for citation Other Chart Elements There can be unconventional things When Graphic Design Goes Badly o The most general standards of charting data are thus the following Present meaningful data Define the data unambigoulsy Do not distort the data Present data efficiently Types of Charts o Pie Charts Represent distribution of categorical components of a single variable Poor representation of data Don’t use them, especially three d ones o Bar Charts Typically display relationship between one or more categorical variables Minimize ink to data ratio o Do not use three d o Legends inside or below plot area o Time Series Line Charts Most efficient means of displaying large data amounts o Boxplots Plots the median and four quartiles o data for an interval level variable Best used for comparing the distribution of the same variable for two or more groups or points at a time Displays how a single case compares to a large number of other cases o Sparklines Eliminate all nondate graphical elements of the chart, providing a simply display of variation in trend and numerical information enough to make sense of the information presented Voting o Voter turnout is an indicator of either the strength of a osciety’s culture or the quality ofinstitutions Low turn out in America is linked to individualism and distrust of the government o Political gerrymandering Process by which elected officials choose voters rather than the other way around o Dsicrepency between votes counted and people saying if they voted Sample mortality effect People may not want to fully participate People lie o Barriers You have to register a month previously