9.6.3 SPSS (optional) SPSS jpn2003_jpn.sav CSV GRETL pisa2003stu_jpn.csv, pisa2003sch_jpn.csv gretl label SAS SPSS script OECD PISA SAS SPSS csv SPSS

Size: px

Start display at page:

Download "9.6.3 SPSS (optional) SPSS jpn2003_jpn.sav CSV GRETL pisa2003stu_jpn.csv, pisa2003sch_jpn.csv gretl label SAS SPSS script OECD PISA SAS SPSS csv SPSS"

あきみかたづ
5 years ago
Views:

1 9.4 PISA Research Questions: 2003 PISA PISA, (170) ( ( ISEI PISA PISA2003 Web site pandas read_csv website >>> import pandas as pd >>> x=pd.read_csv(" >>> print(x.columns) >>> x.shape # OECD (optional) PISA OECD text csv Excel download StQ_CodeBook_2003.pdf 33

2 9.6.3 SPSS (optional) SPSS jpn2003_jpn.sav CSV GRETL pisa2003stu_jpn.csv, pisa2003sch_jpn.csv gretl label SAS SPSS script OECD PISA SAS SPSS csv SPSS SPSS PSPP PISA PISA (PISA2012_SPSS_school.txt ) txt sps SPSS SPS SPSS sps SPSS xxxx d: (CNT= JPN ) csv sav STRATUM PISA STRATUM ISCED SISCED ( A,B,C ) STRATUM Code book STRATUM "390201" "JPN - stratum 01 : 01: Public and Academic" "390202" "JPN - stratum 02 : 02: Public and Practical" "390203" "JPN - stratum 03 : 03: Private and Academic" "390204" "JPN - stratum 04 : 04: Private and Practica" SISCED SISCED (246) Expected educational level of student (ISCED) Format: F1 Columns: ISCED 1 1 ISCED 2 2 ISCED 3B, C 3 ISCED 3A, ISCED 4 4 ISCED 5B 5 ISCED 5A, 4 3A 5A 3B 5B 3C 4 ISCED 34

3 3A - 3A - 3A - 3A - 3A - 3C - () 3C - 3C - 3C - / / 4 / - / / PISA2003 1,2,3 Level-3B 4,5 Level-5B ISCED-3C ISCED-5B - ISCED-5B - ISCED-5A - ISCED-6 - MISCED,FISCED SISCED MISCED,FISCED ST03Q01 PV1MATH-PV5MATH math ESCS White Collar, 6-9 Blue Collar) high skilled white collar (ISCO codes 1,2 and 3) includes legislators, senior officials and managers, professionals and technicians and associate professionals; low skilled white collar (ISCO codes 4 and 5) includes clerks and service workers and shop and market sales workers; high skilled blue collar (ISCO codes 6 and 7) includes skilled agricultural and fishery workers and craft and related trades workers; low skilled blue collar (ISCO codes 8 and 9) includes plant and machine operators and assemblers and elementary occupations. FSECATEG (233) Father White collar/blue collar classification Format: F1 Columns: White Collar high skilled 2 White Collar low skilled 3 Blue Collar high skilled 4 Blue Collar low skilled 9 Missing ESCS SC01Q01 SC01Q01 (5) School location Q1 Format: F1 Columns: Village (less 3 000) 2 Small town (3 000 to ) 3 Town ( to ) 35

4 4 City ( to ) 5 Large city (more ) 7 N/A 8 Invalid 9 Miss 9.7 pivot table: : (ESCS) (ESCS) ( ESCS, : ESCS) : ESCS Oaxaca decomposition: : y = a + b x, ŷ = a + ˆb ˆx, ŷ = y V (y) = y = a + ˆb x V (y) V (x) ˆb ˆb = b V (x) V (y) 9.8 pivot table pandas 36

5 9.8.1 pivot table pivot table DataFrame ( ) 5 x["daigaku"]=x.kitai==16 aa1=pd.pivot_table(data=x,values=[ daigaku, math,"escs"],\ index=["male"], aggfunc="mean") print(aa1) ESCS daigaku math male aa1=pd.pivot_table(data=x,values=[ daigaku, math, kitai,"escs"],\ index=["area"], aggfunc="mean") ESCS daigaku kitai math area # # x["danshiko"]=1*(x.mmale==1) x["joshiko"]=1*(x.mmale==0) x["betsugaku"]=2*(x.mmale==1)+1*(x.mmale==0) x["betsugaku"]=x["betsugaku"].replace(2,"danshiko")# x["betsugaku"]=x["betsugaku"].replace(1,"joshiko") x["betsugaku"]=x["betsugaku"].replace(0,"kyogaku") ESCS daigaku kitai math toshi print(aa1) d=aa1.ix[1]-aa1.ix[0];print(d) #ESCS #daigaku #kitai #math

6 aa1=pd.pivot_table(data=x,values=[ daigaku, math, kitai,"escs"],\ index=["type"], aggfunc="mean") ESCS daigaku kitai math type priv_f priv_s pub_f pub_s aa1=pd.pivot_table(data=x,values=[ daigaku, math, kitai,"escs"],\ index=["betsugaku"], aggfunc="mean" ESCS daigaku kitai math betsugaku danshiko joshiko kyogaku : Series DataFrame β 0 b 0, b 1,, b k ˆβ = b 1 + k b i c 1,i i=2 b 1 x 1 y b i a 1,i x i y = ˆα + ˆβx 1, y = a + b 1 x 1 + b 2 x b k x k x i = a i + c i x 1 x 1 ˆβ y aa1=pd.pivot_table(data=x,values=[ daigaku, math,"escs","male"],\ index=["toshi"], aggfunc="mean") ESCS daigaku male math toshi d=aa1.ix[1]-aa1.ix[0]# ix 38

7 y_data=x[ daigaku ] x_data=x[["toshi","male","escs","math"]] model=pd.ols(y=y_data, x=x_data, intercept=true)# intercept print(model) Summary of Estimated Coefficients toshi male ESCS math intercept End of Summary b=model.beta# indirct=b*d# index #ESCS #male #math ESCS daigaku math male Summary of Estimated Coefficients male < ESCS math toshi itercept End of Summary [ESCS daigaku NaN intercept NaN male NaN math toshi ( =

8 xv03=v03[(v03.type==" ")]# x_data=xv03[["male"]] y_data=xv03[ daigaku ] model03=pd.ols(y=y_data, x=x_data, intercept=true) print(model03) xv15=v15[(v15.type==" ")]# x_data=xv15[["male"]] y_data=xv15[ daigaku ] model15=pd.ols(y=y_data, x=x_data, intercept=true) print(model15) Summary of Estimated Coefficients male < intercept Summary of Estimated Coefficients male intercept #2003 y_data=v03[ daigaku ] v03["male_math"]=v03.male*v03.math v03["male_escs"]=v03.male*v03.escs x_data=v03["male","escs","math","male_math","male_escs"]] model=pd.ols(y=y_data, x=x_data, intercept=true)# intercept print(model) Summary of Estimated Coefficients male ESCS math male_math male_escs intercept male male * math male * ESCS

9 Summary of Estimated Coefficients male ESCS math male_math male_escs intercept Oaxaca Oaxaca X city,i = a city + b city X city,i + error X village,i = a village + b village X village,i + error X city X village = a city a village + b city Xcity b village Xvillage = a city a village + b city ( X city X village ) + (b city b village ) X village city village ols model object model.beta property dir list (Oaxaca) xxv03=xv03[(xv03.male==1)]# x_data=xxv03[["math","escs"]] y_data=xxv03[ daigaku ] model_m=pd.ols(y=y_data, x=x_data, intercept=true) print(model_m) x_m=x_data.mean() xxv03=xv03[(xv03.male==0)]# x_data=xxv03[["math","escs"]] y_data=xxv03[ daigaku ] model_f=pd.ols(y=y_data, x=x_data, intercept=true) print(model_f) Summary of Estimated Coefficients math ESCS intercept

10 Summary of Estimated Coefficients math ESCS intercept x_f=x_data.mean() dx=x_m-x_f print(dx) math ESCS dtype: float64 # ESCS dbeta=model_m.beta-model_f.beta print(dbeta) math ESCS intercept x_f["intercept"]=1# x_m["intercept"]=1 keisu=dbeta*x_f # hensu=model_m.beta*dx # print(keisu.sum()) print(hensu.sum()) # keisu=dbeta*x_m hensu=model_f.beta*dx print(keisu.sum()) print(hensu.sum() ESCS , :2003 MSECATEG (232) Mother White collar/blue collar classification Format: F1 Columns: White Collar high skilled 2 White Collar low skilled 3 Blue Collar high skilled 4 Blue Collar low skilled 42

11 9 Missing FSECATEG (233) Father White collar/blue collar classification Format: F1 Columns: White Collar high skilled 2 White Collar low skilled 3 Blue Collar high skilled 4 Blue Collar low skilled 9 Missing 2. IMMIG (243) Country of birth Format: F1 Columns: Native students 2 First-Generation students 3 Non-native students 9 Missing 43