Ryan Warshavsky Variable aspect that differs from subject to subject Explanatory predictor cause Response predicted effect Data the value of the variable Quantitative numerical numbers measurements Discrete natural gap Continuous arbitrarily close together Qualitative categorical classification Bias Sample Doesn t represent population Selection Bias diff between pop of interest pop you want and effective pop pop you can reach Non response don t answer Response lie interviewer effect Ordinal natural ordering Nominal no order Interval data no meaningful zero Ratio data meaningful zero Time Series one variable over time Cross Sectional multiple variable at a specific point in time Part Sample Whole Population population Sampling Techniques Nonstatistical Convenience whoever is around bias Voluntary choose to be involved bias Statistical SRS entire pop has equal chance of being picked Sampling error sample to sample variation statistic describes sample parameter describes Stratified divide pop into strata according to characteristic take SRS of each strata Cluster divide pop into clusters take SRS of clusters use all in each selected cluster Categorical Data use pie carts and bar graphs Marginal Distributions distribution of 1 variable irrespective of the other given by column row totals Conditional Distribution use column row percentages distribution of one variable responding to other Simpson s Paradox association of several groups can be reversed when groups are combined Quantitative Data use histograms steam leaf boxplots 5 summary spread Skewed right few outliers on right side skewed left few outliers on left side Measures of centrality mean median Mean not robust Median robust 5 summary Min Q1 median of lows Median Q3 median of highs Max Histogram is more informative than boxplots tells shape Fences Upper Q3 1 5 Q3 Q1 Lower Q1 1 5 Q3 Q1 Varience s2 x1 x 2 xn x 2 n 1 Shows variation about the mean Standard Deviation varience z value mean standard deviation avg squared distance from mean how many st dev from mean Lurking Variable not included in study but effects variables studied Confounding Variable effect on response variable cannot be distinguished Regression Line y bo b 1 x b 1 r bo y b1 x z y r zx s y sx Se of the squared residuals Spread about the regression line n 2 y y s y r x x s x Residual observed expected P A event A sample space of outcomes A Total of outcomes Mutually Exclusive events contain no common outcomes can t both happen Addition or rule P A B P A P B Compliment Rule P A c 1 P A General Addition Rule P A B P A P B P A B Independence if one does not influence the other knowing A doesn t give info about B Multiplication And rule P A and B P A P B Check for independence P B A P B P A B P A P B P B A P B A P A General Multiplication Rule P A B P A P B A
View Full Document
Unlocking...