2019
Samenvatting Methodologie en
Toegepaste Biostatistiek 2
HOORCOLLEGES + TWISK
LOES KIEVITSBOSCH
, Samenvatting Methodologie en Toegepaste Biostatistiek 2 | Loes Kievitsbosch
Inhoudsopgave
Hoorcollege 1 ANOVA en Lineaire Regressie ......................................................................................... 4
Lineaire regressie ................................................................................................................................ 4
Bivariate regressie ............................................................................................................................... 5
Interpretatie constante ................................................................................................................... 7
Interpretatie richtingscoëfficiënt .................................................................................................... 7
Total Sum of Squares ........................................................................................................................... 9
Sum of Squared Errors......................................................................................................................... 9
Standaarddeviatie van errors ........................................................................................................ 10
T-toets voor onafhankelijke steekproeven ....................................................................................... 11
T-toets en regressie ........................................................................................................................... 12
Pearson correlatie (r)......................................................................................................................... 12
Correlatie ....................................................................................................................................... 12
Correlatie en regressie ...................................................................................................................... 13
ANOVA: Analysis of Viariance............................................................................................................ 14
ANOVA: Post-hoc toets ................................................................................................................. 14
Hoorcollege 2 Lineaire Regressie Associatie-/Predictiemodellen ...................................................... 16
Bivariate (enkelvoudige regressie) .................................................................................................... 16
Centreren, ......................................................................................................................................... 16
Waarom centreren? ...................................................................................................................... 17
Multiple lineaire regressie................................................................................................................. 17
Stratificeren ................................................................................................................................... 20
Productterm .................................................................................................................................. 20
Dummyvariabelen ............................................................................................................................. 24
Associatiemodel ................................................................................................................................ 26
Predictiemodel .................................................................................................................................. 26
Forward selectie ............................................................................................................................ 26
Backward selectie (voorkeur) ........................................................................................................ 26
Modellen vergelijken: F-toets ........................................................................................................... 27
Multicollineariteit .............................................................................................................................. 27
Hoorcollege 3 Logistische regressie-analyse ......................................................................................... 30
Lineaire regressie-analyse ................................................................................................................. 30
Logistische regressie-analyse ............................................................................................................ 31
Chi kwadraat toets (Chi²/x²/χ²) ......................................................................................................... 32
Vrijheidsgraden ................................................................................................................................. 32
Chi-kwadraat verdeling ..................................................................................................................... 32
1
, Samenvatting Methodologie en Toegepaste Biostatistiek 2 | Loes Kievitsbosch
Meer dan 2 groepen vergelijken ....................................................................................................... 33
Van lineair naar logistisch.................................................................................................................. 34
Logistische regressie-analyse met een dichotome determinant ...................................................... 35
Logistische regressie-analyse: statistische significantie .................................................................... 36
Wald toets ......................................................................................................................................... 36
BI rond de Odds ratio ........................................................................................................................ 36
Likelihood-ratiotoets ......................................................................................................................... 37
Logistische regressie-analyse met een categoriale determinant ...................................................... 39
Logistische regressie-analyse met een continue determinant ......................................................... 39
Confounding en Effectmodificatie..................................................................................................... 40
Hoorcollege 4 Dichotome uitkomstvariabelen, multivariabel .............................................................. 41
Logistische regressie-analyse ............................................................................................................ 41
Aannames logistische regressie ........................................................................................................ 41
Confounding ...................................................................................................................................... 41
Effectmodificatie ............................................................................................................................... 41
Associatiemodellen ........................................................................................................................... 42
Multipele regressie ........................................................................................................................ 43
Predictiemodellen ............................................................................................................................. 43
Fasen predictiemodel ........................................................................................................................ 44
Assumptiecheck............................................................................................................................. 44
Ontwikkeling predictor selectie..................................................................................................... 45
Kwaliteit, Validiteit ........................................................................................................................ 45
Implementatie ............................................................................................................................... 48
Collineariteit ...................................................................................................................................... 49
Samenvatting..................................................................................................................................... 50
Hoorcollege 5 Survivaldata.................................................................................................................... 51
Survival modellen .............................................................................................................................. 51
Terminologie ..................................................................................................................................... 51
Kaplan Meier curves: berekening ...................................................................................................... 52
Cox regressie ..................................................................................................................................... 53
Samengevat: regressie analyses ........................................................................................................ 53
Hazard ratio ....................................................................................................................................... 54
Assumpties ........................................................................................................................................ 54
Overview............................................................................................................................................ 56
Hoofdstuk 4 Twisk – De analyse van continue uitkomstvariabelen ...................................................... 57
Hoofdstuk 5 Twisk de analyse van dichotome uitkomstvariabelen ...................................................... 63
2
, Samenvatting Methodologie en Toegepaste Biostatistiek 2 | Loes Kievitsbosch
Hoofdstuk 6 Twisk – De analyse van overlevingsdata (survivaldata) .................................................... 72
Hoofdstuk 7 Twisk: Multiple regressieanalyse: associatie- en predictiemodellen ............................... 80
Hoofdstuk 9 Twisk: Een stap verder ...................................................................................................... 85
Responsiecollege ................................................................................................................................... 88
Lineaire regressieanalyse .................................................................................................................. 88
Lineaire regressie interpretatie (afhankelijk van de determinant) ............................................... 88
Logistische regressieanalyse ............................................................................................................. 89
Cox regressie ..................................................................................................................................... 89
Interpretatie Cox-regressieanalyse ............................................................................................... 89
Confounding en effectmodificatie..................................................................................................... 90
Valkuilen in de statistiek.................................................................................................................... 90
Assumpties ........................................................................................................................................ 90
3
Samenvatting Methodologie en
Toegepaste Biostatistiek 2
HOORCOLLEGES + TWISK
LOES KIEVITSBOSCH
, Samenvatting Methodologie en Toegepaste Biostatistiek 2 | Loes Kievitsbosch
Inhoudsopgave
Hoorcollege 1 ANOVA en Lineaire Regressie ......................................................................................... 4
Lineaire regressie ................................................................................................................................ 4
Bivariate regressie ............................................................................................................................... 5
Interpretatie constante ................................................................................................................... 7
Interpretatie richtingscoëfficiënt .................................................................................................... 7
Total Sum of Squares ........................................................................................................................... 9
Sum of Squared Errors......................................................................................................................... 9
Standaarddeviatie van errors ........................................................................................................ 10
T-toets voor onafhankelijke steekproeven ....................................................................................... 11
T-toets en regressie ........................................................................................................................... 12
Pearson correlatie (r)......................................................................................................................... 12
Correlatie ....................................................................................................................................... 12
Correlatie en regressie ...................................................................................................................... 13
ANOVA: Analysis of Viariance............................................................................................................ 14
ANOVA: Post-hoc toets ................................................................................................................. 14
Hoorcollege 2 Lineaire Regressie Associatie-/Predictiemodellen ...................................................... 16
Bivariate (enkelvoudige regressie) .................................................................................................... 16
Centreren, ......................................................................................................................................... 16
Waarom centreren? ...................................................................................................................... 17
Multiple lineaire regressie................................................................................................................. 17
Stratificeren ................................................................................................................................... 20
Productterm .................................................................................................................................. 20
Dummyvariabelen ............................................................................................................................. 24
Associatiemodel ................................................................................................................................ 26
Predictiemodel .................................................................................................................................. 26
Forward selectie ............................................................................................................................ 26
Backward selectie (voorkeur) ........................................................................................................ 26
Modellen vergelijken: F-toets ........................................................................................................... 27
Multicollineariteit .............................................................................................................................. 27
Hoorcollege 3 Logistische regressie-analyse ......................................................................................... 30
Lineaire regressie-analyse ................................................................................................................. 30
Logistische regressie-analyse ............................................................................................................ 31
Chi kwadraat toets (Chi²/x²/χ²) ......................................................................................................... 32
Vrijheidsgraden ................................................................................................................................. 32
Chi-kwadraat verdeling ..................................................................................................................... 32
1
, Samenvatting Methodologie en Toegepaste Biostatistiek 2 | Loes Kievitsbosch
Meer dan 2 groepen vergelijken ....................................................................................................... 33
Van lineair naar logistisch.................................................................................................................. 34
Logistische regressie-analyse met een dichotome determinant ...................................................... 35
Logistische regressie-analyse: statistische significantie .................................................................... 36
Wald toets ......................................................................................................................................... 36
BI rond de Odds ratio ........................................................................................................................ 36
Likelihood-ratiotoets ......................................................................................................................... 37
Logistische regressie-analyse met een categoriale determinant ...................................................... 39
Logistische regressie-analyse met een continue determinant ......................................................... 39
Confounding en Effectmodificatie..................................................................................................... 40
Hoorcollege 4 Dichotome uitkomstvariabelen, multivariabel .............................................................. 41
Logistische regressie-analyse ............................................................................................................ 41
Aannames logistische regressie ........................................................................................................ 41
Confounding ...................................................................................................................................... 41
Effectmodificatie ............................................................................................................................... 41
Associatiemodellen ........................................................................................................................... 42
Multipele regressie ........................................................................................................................ 43
Predictiemodellen ............................................................................................................................. 43
Fasen predictiemodel ........................................................................................................................ 44
Assumptiecheck............................................................................................................................. 44
Ontwikkeling predictor selectie..................................................................................................... 45
Kwaliteit, Validiteit ........................................................................................................................ 45
Implementatie ............................................................................................................................... 48
Collineariteit ...................................................................................................................................... 49
Samenvatting..................................................................................................................................... 50
Hoorcollege 5 Survivaldata.................................................................................................................... 51
Survival modellen .............................................................................................................................. 51
Terminologie ..................................................................................................................................... 51
Kaplan Meier curves: berekening ...................................................................................................... 52
Cox regressie ..................................................................................................................................... 53
Samengevat: regressie analyses ........................................................................................................ 53
Hazard ratio ....................................................................................................................................... 54
Assumpties ........................................................................................................................................ 54
Overview............................................................................................................................................ 56
Hoofdstuk 4 Twisk – De analyse van continue uitkomstvariabelen ...................................................... 57
Hoofdstuk 5 Twisk de analyse van dichotome uitkomstvariabelen ...................................................... 63
2
, Samenvatting Methodologie en Toegepaste Biostatistiek 2 | Loes Kievitsbosch
Hoofdstuk 6 Twisk – De analyse van overlevingsdata (survivaldata) .................................................... 72
Hoofdstuk 7 Twisk: Multiple regressieanalyse: associatie- en predictiemodellen ............................... 80
Hoofdstuk 9 Twisk: Een stap verder ...................................................................................................... 85
Responsiecollege ................................................................................................................................... 88
Lineaire regressieanalyse .................................................................................................................. 88
Lineaire regressie interpretatie (afhankelijk van de determinant) ............................................... 88
Logistische regressieanalyse ............................................................................................................. 89
Cox regressie ..................................................................................................................................... 89
Interpretatie Cox-regressieanalyse ............................................................................................... 89
Confounding en effectmodificatie..................................................................................................... 90
Valkuilen in de statistiek.................................................................................................................... 90
Assumpties ........................................................................................................................................ 90
3
