Page 188 - สถิติทางวิทยาศาสตร์สุขภาพ เพื่อการวิจัย

10.9.2 °“√§”π«≥¢π“¥µ—«Õ¬à“ß ”À√—∫°“√‡ª√’¬∫‡∑’¬∫
√–À«à“ߪ√–™“°√ 2 °≈ÿà¡„π°√≥’ ÿࡵ—«Õ¬à“ß·∫∫°≈ÿà¡

°“√ÕÕ°·∫∫ß“π«‘®—¬¡’°“√ ÿà¡‚√ß欓∫“≈À√◊ÕÀ¡Ÿà∫â“π‡ªìπÀπ૬ ÿà¡ Ÿµ√∑’Ë„™â„π
°“√§”π«≥¢π“¥µ—«Õ¬à“ß¡’°“√ª√—∫§à“§«“¡ —¡æ—π∏å√–À«à“ߧ«“¡·ª√ª√«π¿“¬„π°≈ÿà¡·≈–
√–À«à“ß°≈ÿà¡ (intra-cluster correlation (¯)) ‡æ‘Ë¡‡¢â“‰ª„π Ÿµ√°“√§”π«≥¢π“¥µ—«Õ¬à“ß

„π°√≥’°“√‡ª√’¬∫‡∑’¬∫§à“‡©≈’ˬ

n = n(1 + (n - 1)¯)
CLUS
‚¥¬∑’Ë n= ¢π“¥µ—«Õ¬à“ß∑’˧”π«≥®“°°“√ ÿà¡·∫∫ßà“¬
¯=§à“§«“¡ —¡æ—π∏å√–À«à“ߧ«“¡·ª√ª√«π¿“¬„π°≈ÿà¡·≈–√–À«à“ß°≈ÿà¡

(0 ∆ ¯ ∆ 1)
§à“ ¯ À“¡“®“°ß“π«‘®—¬∑’˺à“π¡“

„π°√≥’°“√‡ª√’¬∫‡∑’¬∫§à“ —¥ à«π

n = n(1 + (n - 1)K)
CLUS
k
1 - Â np ( 1 - p )
ii
i
K= = i 1
k(n - 1) ( p 1 - ) p

‚¥¬∑’Ë n= ¢π“¥µ—«Õ¬à“ß∑’˧”π«≥®“°°“√ ÿà¡·∫∫ßà“¬
K= degree of within-cluster dependency (0 ∆ k ∆ 1)

k k
 n i  np
ii
n = i=1 p = i=1
k i  n
i

k= ®”π«π cluster
n = ®”π«πµ—«Õ¬à“ß∑’Ë cluster i
i
p = —¥ à«π¢Õß cluster i
i
¢âÕ¡Ÿ≈∑’Ë„™â§”π«≥§à“ K ‰¥â®“°√“¬ß“π«‘®—¬∑’˺à“π¡“

10.9 °“√§”π«≥¢π“¥µ—«Õ¬à“ß„π°√≥’°“√ ÿࡵ—«Õ¬à“ß·∫∫°≈ÿà¡ 187

∫∑∑’Ë10 187 30/7/09, 9:21 AM