Final Exam Study Guide
Final Exam Study Guide 400
Popular in Research in Political Science
Popular in Public Relations
This 8 page Study Guide was uploaded by Dora Notetaker on Friday December 4, 2015. The Study Guide belongs to 400 at University of Alabama at Birmingham taught by John E McNulty in Summer 2015. Since its upload, it has received 34 views. For similar materials see Research in Political Science in Public Relations at University of Alabama at Birmingham.
Reviews for Final Exam Study Guide
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 12/04/15
PSC 4002D Research in Political Science Dr McNulty Study Guide for Final Exam SPSS 0 Data from a survey that represents everyone in the United States where people were asked a bunch of questions 0 The ID column does not mean anything so it is in essence nominal data Because a person s ID number does not tell you anything about the person 0 Variable data tells you what variable type you can use for viewing the data 0 So as long as Year is numeric it is 2008 but if you change it to money it will say 2008 0 If you go to the Values category under Variable View and you click on the extended view of any value a window will pop up that will explain what the valuesnumbers in that category stand for o R stands for Respondent How to find the Mean Mode Median Range Min Max Variance Standard Deviation Quartiles of a data set in SPAA o Analyze Descriptive Statistics Descriptives Choose a category Click Options choose what you need to see Continue OK or o Analyze Descriptive Statistics Frequencies choose a category Statistics choose values you want calculated Continue OK We re going to be using bivariate or multivariate variables 0 For bivariate and multivariate variables if the value of X explanatory variable changes the value of Y dependent variablevariable of interest changes 0 For example the candidates in the presidential debates want to affect poll numbers and thereby donations o The question is how to do that Attacking the other candidates or attacking the moderators There an infinite amount of possibilities but they have to figure out what works for them If you have 2 numeric variables you have to come up with a correlation coefficient The correlation coefficient tells you on a scale how much the 2 variables correlate If you have 0 correlation it means you have no correlation If you have 1 correlation it is a perfect positive correlation If you have 1 correlation it is a perfect negative correlation The answer depends on the statistical significance of the correlation What does statistical significance mean 0 Represented with r or R o It is the likelihood that you ll see a correlation of that size that happens despite the fact that there is not relationship o It measure the likelihood of the null hypothesis So the smaller the statistical significance the more likely a true relationship Also stat sign never actually reaches zero Because if you do a measurement millions of times at one point you are going to get the answer 0 So if you flip a coin a million times at one point you re going to get 7 heads out of 10 0 But 5 is when you say you are 95 sure It depends the strength of the correlation and the size of the sample you are using p You come up with a measure of how likely it is that we see a correlation of that size If vou ve got 1 numeric 1 categorical variable 0 You figure out the mean for the numeric category and the standard deviation 0 For example if you are trying to figure out height and you have men and women you figure out the mean amp the standard deviation for both men and women 0 Then you subtract the means from each other and the stdeviations from each other Mmen Mwomen This is called the mean difference Mmen Mwomen ZSD mSDm Z is just a value we put there could be any number If on a test you re asked for 95 confidence you put 95 for Z 0 You re standard deviations could be small for height so this would work out to be 95 confidence 0 Which means you are 95 confident that the height difference between men and women is between 2 inches and 8 inches 0 Which tells you that the height difference is not 0 so you can reject with 95 confidence the null hypothesis 0 Null hypothesis in this case is that there is no difference between men and women s height 0 Null hypothesis is the hypothesis of nonrelationship o The goal of your demonstration is to reject the null hypothesis 0000 For 2 categorical variables you have to use the chisquare statistic Chisquare statistic o It is used to compare 2 categorical variables International Relations researchers use dyads to compare countries 0 What makes countries go to war with each other 0 For each dyads they ask 2 questions 0 Do they share a border 0 Have they been to war with each other 0 They put these in a table 0 This is our Observed table War No War Border 500 3000 3500 No border 1000 30000 31000 1500 33000 34500 0 made up numbers 0 sums are on the outside of the table c We create an Expected table with the same numbers on the outside War No War Border Row1Grand total Row1Grand total x 35 x Column1 Column2 No Row2Grand total Row2Grand total x 310 border x Column1 Column2 15 330 345 0 And this equals War No War Border 35345 x 15 15 335 35 No 135 2965 310 border 15 330 345 The sums on the outsides of the graph should be the same as before So you could subtract one of the rows of cell you have to work before you don t have to use multiplication or division is known as degrees of freedom or df 0 df of rows 1 x of columns 1 For the Chisquare statistic 3 2 be amt1 Biep tad 39 expote d War No War Border a b No c d border 0 a 5152 15 82 o b BO3352 335 04 o c 101352 135 09 o d 30029652 2965 00 0 Sum 82040900 95 x2 95 We looked at our sample SPSS table to see how to get the ChiSquare Statistic In SPSS use the following buttons tabs to get the Chi square statistic 1 Analyze tab Descriptive Statistics Crosstabs Choose 2 variables Statistics check the Chi square Continue OK 0 When you have the crosstabs window open be sure to go to Cells and hit both the Expected and the Observed tabs The middle table is the Expected table The Chisquare Tests table is your results When looking at the Chi square table on the bottom in the Pearson Chi Square row the number under Value is your chi square the df is your degrees of freedom and the last value in the row is your percentage 0 So if it says 019 it means you are 19 likely that the relationship could have happened by random change so there s no true relationship 0 So you re 981 confident that there is a relationship between the variables thereby disproving the Null hypothesis which says there is no relationship To select cases you need Data Select Cases Choose the variable you want Under if choose what you need OK 0 You will see the cases you don t need in your actual data table with a line through them TO compare means Analyze Compare Means Indep Samples T Test Choose 2 variables Define Groups put 0 and 1 hit OK We looked at a new data set stats about all of the states and the District of Columbia 0 You can find it as STATES1O in Canvas under Files If the confidence interval does not equal 0 then you can reject the null hypothesis Most of the data on this SPSS data set are categorical data even if they are given values that are numbers which make them easier to identify 0 But you must treat them as categorical because that s what they are You can use scatter plots to show your data Because it gives you a lot of information in a small amount of time It is a bivariate table with one variable on the xaxis and one of the yaxis If there is no relationship between the variables you ll get a data cloud a bunch of plots If there is a positive relationship between them a line will go up towards the upper right hand corner o If there is a negative relationship between them the line will go toward the upper left hand corner To make a graph in SPSS 0 Go to Graphs Legacy Dialogs Choose your graph To make a scatter plot graph 0 Go to Graphs Legacy Dialogs ScatterDot Simple Scatter Define Choose your 2 variables OK We used Percentage population White and Percentage populations over age of 65 If you double click on your chart the Chart Editor comes up 0 When you re in the Chart Editor hit a small button on the top line that looks like a line through a graph 0 When you click on it a line will appear through your graph which gives you where most of the dots are therefore where the correlations is o On the top right corner it ll give you the R square statistic Rsquare statistic 0 Measures how much of the variation in the data is explained by the model o If none of the dots were on the line your R square would be close to zero 0 R stands for the correlations coefficient Conceptualizing and going over the chisquare statistic Coin flip 0 You flip it and it is 5050 that you ll get one or the other 0 But you can flip it 10 times and get heads 10 times There s an expectancy and a result You may expect to get 5 heads and that s pretty likely But the more heads you expect the less likely you ll get heads But if you flip a coin a 1000 times the less likely it is that you ll get 600 and 400 of each 0 Because the bigger numbers you use the more the results get pushed toward the mean But if you have 2 coins 0 You flip them 6 times you expect 3 heads 3 tails for one and 3 heads and 3 tails for the other H T H 3 3 T 3 3 This table is a chi square distribution The chi square distribution is just taking probability of 2 different distributions and using a formula to calculate how unlikely it is o The further you get from your expected probability the less likely it is that it is random chance 0 The chi square distribution is a way of measuring how far you are from your expected Degrees of freedom 0 rows1 columns1 o A 2 by 2 table will have 1 degree of freedom 0 The amount of cases you have 1 Way to calculate chi square in SPSS o Analyze Descriptive Statistics Crosstabs Click on Statistics Check Chi square Click on Cells Check Expected Put in 2 categories that you want Hit OK Review of the Chisquare statistic 0 Used when looking at 2 categorical variables o It is the formal measure of the difference btw what a proportionally normal table and the table you see c The bigger the chisquare is the more likely that you can get a random distribution by random chance 0 The size of the chisquare is conditioned by the degrees of freedom Difference in means test Used when you have 1 categorical 1 numeric variable It tells you if that difference is big enough that you can reject that it s just random vana on It will calculate 2 means for 2 different groups separated by a category There s going to be some random variation your 2 means are never going to be the same SPSS is going to square them add them and take the square root Equa on Diff in means or Z39 summed standard deviations 0 Z 95 So if you look at the difference between the average height of men and women and the difference is say 6 o 6or22 o 6or4 GPA s GPA for men 31 GPA for women 32 Difference 01 01 or 09 So the difference in means between 10 and 08 So we can t reject that there is no difference between men and women 0 The difference is so small 0 We accept the Null hypothesis Regression 2 numerics Example higher income students vs lower income students Y Cl BX 8 GPA 0L 3P arem39s Income 8 How to do a difference in means test in SPSS Analyze tab Compare means Independent samples test Choose 1 numeric and 1 categorical variable 0 r s age when 1st child born 0 grouping variable Respondent s sex Define groups 0 1 and 2 0 Do this by looking at what Difference in means Find the 2 means on the chart Std Error value x 196 confidence interval 213 Mean difference is an estimate for the whole us population based on 2000 ppl in survey Your confidence in that estimate is defined by the standard deviation and the size of your sample which is how you get the standard error What is your confidence interval We always use 95 So we always have to multiply by 196 if 95 but SPSS does this for us so we don t have to multiply lnterval not including 0 means you can reject the null hypothesis TTest Mean difference i Confidence interval STD error difference 213 i 196276 2672 1588 Can you reject the null hypothesis that there is no correlation btw your variables Yes because 0 in not in between the confidence intervals How to do regression in SPSS Open STATES10 data set Analyze tab Regression Linear Choose 2 variables 0 Dep Median age 2008 o lndep of population foreign born Analyzing the results 0 t is a ratio BSTD error t Stg 263 No there is not a relationship P value in much higher than 005 New variables Age at 2008 Birth rate Std error is lower than the B Ratio is 19 t value over 2 is 95 significant What will be on our final exam In SPSS 0 Do couple of Chisquares 0 Do couple of Difference in means tests 0 Do couple of Regressions o lnterpret results from each
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'