03 - bivariate analysis - ordinal
TRANSCRIPT
Senin, 21 Februari 2011
Variabel 1
Variabel 2
Nominal Ordinal Interval
Nominal Chi Quare χ2
Phi ϕ CoefficientCoefficient Contingency CCramer’s ν (nu)Lambda λ simetrikLambda λ asimetrik
Spearman rs t – test (hypothesis of difference)Z – test (hypothesis of difference)Eta η
Ordinal Kendall’s τGamma γSpearman rs
Sommer’s D asimetrik
Interval Pearson’s rRegression asimetrik
Nominal Bivariate Analysis - continued
...is a proportional reduction in error statistics
...it reflects the degree to which knowledge of the independent variable reduces errors in “predicting“ where cases will fall on the dependent variable
...it tells us how much, proportionally, knowledge of the independent variable improves our ability to guess the dependent category
Bidang Kerja (Y)
Gaya Hidup (X)
Achiever Anxious Pusher
Manufaktur 32 47 54 133
Jasa 32 60 39 131
Pemerintahan
23 41 95 159
87 148 188 423 N = Total number of cases Ly = The number of cases in the modal
Y category, ignoring X Lyx = the number of cases in the largest
cell within given X category Lx = The number of cases in the modal
X category, ignoring Y Lxy = the number of cases in the largest
cell within given Y category
Λxy = (Σ Lyx – Ly) / (N – Ly)
Λyx = (Σ Lxy – Lx) / (N – Lx)
Bidang Kerja (Y)
Gaya Hidup (X)
Achiever Anxious Pusher
Manufaktur 32 47 54 133
Jasa 32 60 39 131
Pemerintahan
23 41 95 159
87 148 188 423 Λxy = (Σ Lyx – Ly) / (N – Ly) = (32+60+95) - 159/ (423 – 159) = 0.11 Λyx = (Σ Lxy – Lx) / (N – Lx) = (54+60+95) - 188/ 423 – 188) = 0.09
Bidang Kerja (Y)
Gaya Hidup (X)
Achiever Anxious Pusher
Manufaktur 32 47 54 133
Jasa 32 60 39 131
Pemerintahan
23 41 95 159
87 148 188 423 Ly = The number of cases in the
modal Y category Lx = The number of cases in the
modal X category Lyx = the number of cases in the
largest cell within given X category Lxy = the number of cases in the
largest cell within given Y category
Λyx = (Σ Lyx + Σlxy – Ly - Lx)/ 2N - Ly - Lx
Λyx = (32+60+95)+(54+60+95) – 159 - 188)/ 2(423) – 159 – 188
Bandingkan korelasi symmetric dengan korelasi asymmetric yang diperhitungkan sebelumnya
Bila symmetric lebih besar (>) daripada asymmetric, berarti:• Mungkin ada variabel ketiga (intervening
variable)• Hubungan x – y tidak bersifat kausal
In practice
Measures Greek Symbo
l
Type of Data High Associatio
n
Lambda λ Nominal 1 - 0
Gamma γ Ordinal +1.0 , -1.0
Tau (Kendall’s)
τ Ordinal +1.0 , -1.0
Rho ρ Interval, Ratio +1.0 , -1.0
Chi - square
χ2 Nominal, Ordinal
Infinity
Bivariate Analysis
Statistical Significance
Strength
Direction
Form
Non Spuriousness
Direction of Influence
Theoretical Status of IV
Sufficient Necessary Contributory
Reciprocal
Asymmetric
Symmetric
Related to Differences
An unsophisticated forecaster uses statistics as a drunken man uses lamp-posts - for support rather than for illumination (Andrew Lang)
Facts are stubborn, but statistics are more pliable (Mark Twain)