full2010.pdf - page 396

358
Cramer Von Mises, Lilliefors, Chi-square and Kolmogorov – Smirnov tests, respectively. The power of all tests
increases as sample size and the level of significances increases. [In multiple regressions and when independent
variables have correlations, the regression coefficients are not multivariate-normally distributed. However, the
estimated regression coefficients (Under the assumption that errors are normally distributed.) are consistent with
the actual values in all the cases. ]
Keywords: Normal distribution, Logistic distribution, Linear regression
‡Î µœÎµ
Ĝ„µ¦ª· Á‡¦µ³®ro°¤¼¨šµŠ™· ˜·¤´„¤¸o°¤¤˜·ÁºÊ°Š˜o œªnµž¦³µ„¦¤¸„µ¦Â‹„‹ŠÂž„˜· Án œ Ĝ„µ¦
š—° t , „µ¦š—° F œ°„‹µ„œ´Êœ „µ¦ª· Á‡¦µ³®r„µ¦™—™°¥ ¤¸ ÁŠºÉ°œÅ š¸Éªnµ‡ªµ¤‡¨µ—Á‡¨ºÉ°œ¤¸„µ¦Â‹„‹ŠÂ
ž„˜· ×¥™oµ o°¤¼¨š¸ÉœÎµ¤µª· Á‡¦µ³®rŤnÁž}œÅž˜µ¤ÁŠºÉ°œÅ š¸É„ε®œ—Ūo „È°µ‹nŠŸ¨˜n °„µ¦ª·Á‡¦µ³®ro°¤¼¨ šÎµÄ®oŸ¨
„µ¦ª· Á‡¦µ³®rŤn™¼„˜o °Š ¨³°µ‹šÎµÄ®oÁ„·—‡ªµ¤Á¸¥®µ¥Á¤ºÉ°œÎµŸ¨‹µ„„µ¦ª·Á‡¦µ³®r Şčo —´Šœ´Êœ „µ¦˜¦ª‹° o°¤¼¨
ªnµ¤¸¨´„¬–³Á®¤µ³¤Â¨³°—‡¨o°Š„´ÁŠºÉ°œÅ ÁºÊ°Š˜oœ‹¹ ŠÁž}œÁ¦ºÉ°ŠÎµ‡´
„µ¦Â‹„‹ŠÂž„˜· ™º°Áž} œ„µ¦Â‹„‹Šš¸É ε‡´¤µ„ œ°„‹µ„Áž}œÁŠºÉ °œÅ Ĝ„µ¦ª·Á‡¦µ³®ršµŠ™·˜· nªœ
Ä®n¨oª Á®˜»„µ¦–rš¸ÉÁ„·— ¹ÊœÃ—¥›¦¦¤µ˜· nªœÄ®n„Ȥ¸¨´„¬–³Ä„¨oÁ‡¸¥Š„µ¦Â‹„‹ŠÂž„˜· ‡º° o°¤¼¨‹³¤¸¦¼ž¦nµŠ
¨´„¬–³Áž}œ¦¼ž¦³‰´ Š‡ªÉε o °¤¼¨nªœ¤µ„‹³°¥¼n˜¦Š„¨µŠ¦° Ç ‡n µÁŒ¨¸É¥ ¨³‡n °¥ Ç ¨—œo°¥¨ŠÅžÃ—¥¤¤µ˜¦š´ÊŠ
šµŠŽo µ¥Â¨³ ªµ ‹¹ŠšÎµÄ®o„µ¦Â‹„‹ŠÂž„˜·Áž}œ„µ¦Â‹„‹Šš¸Éœ·¥¤Äo „´œ°¥n µŠÂ¡¦n®¨µ¥ °¸„®œ¹ÉŠ„µ¦Â‹„‹Š
˜n °ÁœºÉ°Šš¸Éœn µœÄ‹ ‡º° „µ¦Â‹„‹ŠÂÃ¨‹· ˜·„ Áž}œ„µ¦Â‹„‹Šš¸É¤¸„µ¦œÎµÅžž¦³¥» „˜rĜ®¨µ¥—oµœ Ánœ —o µœ
„µ¦˜¨µ— —oµœ¸ ªª·š¥µ —oµœÁš‡ÃœÃ¨¥¸ —oµœ¡¨´ŠŠµœ š¸É ε‡´‡º°Äo 宦´˜´ªÂ„µ¦Á‹¦·Á˜·Ã˜ ( Growth Models)
„µ¦Â‹„‹ŠÂÃ¨‹· ˜·„ ¤¸¦¼ž¦nµŠ¨´„¬–³Áž}œ¦¼ ž¦³‰´Š‡ªÉε š¸É¤¸‡ªµ¤¤¤µ˜¦ °ŠÁ ˜Áž}œÅ—o˜´ÊŠÂ˜n f ™¹Š f Ž¹ÉŠ
¤¸‡ªµ¤Ä„¨o Á‡¸¥Š„´„µ¦Â‹„‹ŠÂž„˜·¤µ„ ×¥˜n µŠ„´œÁ¡¸¥Šnªœ®µŠš¸É®œµ„ªn µ ¨³¤¸‡ªµ¤Ã—nŠ„ªn µÁ¨È„œo °¥
Ášnµœ´Êœ
ÁœºÉ°Š‹µ„„µ¦Â‹„‹ŠÂž„˜· ¨³„µ¦Â‹„‹ŠÂÃ¨‹· ˜·„¤¸‡ªµ¤Ä„¨oÁ‡¸¥Š„´œ¤µ„ ×¥ÁŒ¡µ³Äœ„¦–¸š¸É
o°¤¼¨™¼ „Á„ȝ¤µÁ¡¸¥ŠµŠnªœ ®¦º° o °¤¼¨¤¸‹Îµœªœœo°¥ šÎµÄ®o Ťn µ¤µ¦™Â¥„Å—o ªnµ o°¤¼¨¤¸„µ¦Â‹„‹ŠÂž„˜·
®¦º°¤¸„µ¦Â‹„‹ŠÂÃ¨‹· ˜·„ —´Šœ´Êœ‹¹ŠœÄ‹š¸É ‹³«¹„¬µªnµ ˜´ª™·˜·Ä—š¸É ¤¸ž¦³·š›·£µ¡Äœ„µ¦‡´—„¦°Š o °¤¼¨š¸É ¤¸
„µ¦Â‹„‹ŠÂÃ¨‹· ˜·„‹µ„„µ¦Â‹„‹ŠÂž„˜· ™oµœÎµ o°¤¼¨š¸É¤¸„µ¦Â‹„‹ŠÂÃ¨‹· ˜·„¤µšÎ µ„µ¦ª·Á‡¦µ³®r
o°¤¼¨šµŠ™·˜·š¸É ¤¸ÁŠºÉ°œÅ ÁºÊ°Š˜oœªnµž¦³µ„¦¤¸„µ¦Â‹„‹ŠÂž„˜· ‹³nŠŸ¨„¦³š°¥nµŠÅ¦ ¨³‹³nŠŸ¨Äœ
„¦–¸Ä—oµŠ
°»ž„¦–r¨³ª·›¸„µ¦
Šµœª·‹´¥‡¦´ÊŠœ¸Êŗo„ε®œ— °Á ˜Â¨³ª·›¸—εÁœ·œ„µ¦ª·‹´¥ —´Šœ¸Ê
1. „µ¦š—°‡ªµ¤µ¤µ¦™ °Š˜´ ª™·˜· 宦´ „µ¦‡´—„¦°Š„µ¦Â‹„‹ŠÂÃ¨‹· ˜·„‹µ„„µ¦Â‹„‹Š
ž„˜· čo ˜´ª™·˜·š—° 6 ˜´ª™·˜· ŗo „n ˜´ª™·˜· Kolmogorov - Smirnov, ˜´ª™·˜· Shapiro Wilk, ˜´ª™·˜·
Anderson Darling , ˜´ª™·˜· Lilliefors, ˜´ª™·˜· Cramer Von Mises ¨³˜´ ª™·˜· Chi-square