Using Descriptive Statistics and Graphs

Abstract

This chapter discusses the use of descriptive statistics and graphs to present to policymakers and conduct exploratory data analysis (EDA) . Descriptive statistics, that include measures of central tendency and dispersion, are discussed and demonstrated using real data. The utilization of graphs, which includes histograms, box charts, and scatter plots , are also presented.

6.5 Appendix

*Chapter 6 Syntax cd "C:\Users\Marvin\Dropbox\Manuscripts\Book\Chapter 4\Stata" use "US high school graduates in 2012 enrolled in PSE, by state.dta" *compute the arithmetic means ameans public private mean public private *measures of central tendency, we can use summarize, detail or (sum, detail) sum, detail *Measures of dispersion *employing the tabstat command: cd "C:\Users\Marvin\Dropbox\Manuscripts\Book\Chapter 5\Stata files" use "Example 5.2.dta" tabstat NetTuition FTEStudents, stat(cv) *Distributions cd "C:\Users\Marvin\Dropbox\Manuscripts\Book\Chapter 6\Stata files" use "Example 6.1.dta", clear *frequencies and cross tabulations (crosstabs) tab X1RACE tab X1RACE, sort *frequencies and cross tabulations (crosstabs) with descriptive statistics tab X1RACE, summarize(EarnHr) tab X1RACE X1SEX, sum(EarnHr) means *create a new categorical variable (e.g., RaceEthnic) reflecting /// race/ethnicity with value labels based on a fewer categories, /// linking the new variable to the value labels *look at how the variable is coded codebook X1RACE *create new categorical variable gen RaceEthnic = 0 replace RaceEthnic = 1 if X1RACE==2 replace RaceEthnic = 2 if X1RACE==3 replace RaceEthnic = 3 if X1RACE==4 | X1RACE==5 replace RaceEthnic = 4 if X1RACE==6 replace RaceEthnic = 5 if X1RACE==1 | X1RACE==7 replace RaceEthnic = 6 if X1RACE==8 *create variable and value labels lab var RaceEthnic "Race/Ethnicity" label define RaceEthnic1 1 Asian 2 Black 3 Hispanic 4 Multiracial 5 Other 6 White *link the new variable to the value labels label values RaceEthnic RaceEthnic1 *frequencies and cross tabulations (crosstabs) with descriptive statistics tabulate RaceEthnic X1SEX, sum(EarnHr) means *with labels table X1RACE X1SEX, contents(mean EarnHr) clear *descriptive statistics for panel data *open panel dataset use "Example 6.3.dta" xtdes *descriptive statistics for time-invariant variables xttab region_compact *descriptive statistics for time-variant variables xttrans ugradmerit *Graphs - EDA *histogram , with a superimposed normal curve *create a new variable gen stapr_fte = stapr/fte *create histogram with a superimposed normal curve histogram stapr_fte, normal *box chart of the same data *histogram with the frequency of categories with percent histogram region_compact, discrete addlabels /// ylabel(,grid) xlabel(0 1 2 3 4, valuelabel) percent *histogram with the frequency of categories with percents and /// superimposed normal curve *box charts to examine the distribution of data by categories graph box stapr_fte, by(region_compact) *scatter plots to show the simple relationship between two continuous variables graph twoway scatter stapr_fte netuit_fte if year==2016 *scatter plots to show the simple relationship between two continuous /// variables with fitted regression line twoway (scatter stapr_fte netuit_fte) (lfit stapr_fte netuit_fte) if year==2016 *adding the option mlabel(state) twoway scatter stapr_fte netuit_fte, /// mlabel(state) || lfit stapr_fte netuit_fte || if year==2016 *use of a Stata user-written program, aaplot (Cox 2015) *install aaplot ssc install aaplot *run aaplot for two different time periods (1990 & 2016) aaplot netuit_fte stapr_fte if year==1990 aaplot netuit_fte stapr_fte if year==2016 *close dataset Clear *end