生态学试验设计与解释中常见问题.pdf

29 7 V ol. 29, N o. 7 2009 7 ACTA ECOLOG ICA SIN ICA Jul. , 2009 1, 2 1, 2 1, * 2 牛海山 , 崔骁勇 , 汪诗平 , 王艳芬 ( 1. , 59 810008 2. , 19 100049 ) : , , , , [ 2] 3: ( 1) , H ur lbert , ( 2) , , ( 3) , ( ) ( ) , 3 , : : 1000-0933 ( 2009) 07- 390 1- 10 : Q 143 : A Three common errors in Chinese ecological experimental design and interpretation 1, 2 1, 2 1, * 2 N IU H a-i Sh an , CU I X iao-Y ong , WANG Shi-P ing , WANG Y an-Fen 1N orthw est Institu te of P la teau B iology, CAS, X in ing 810008, C hina 2 G raduate University of CA S, B eij ing 100049, China A cta E co log ica S in ica, 2009, 29( 7) : 3901~ 3910. Abstract: The log ic of expermi en ts is d iscussed, and three comm on m istak es in design ing and in terp ret ing exp ermi en ts am ong ecolog ists are ana lyzed. R esearch ers som et mi es overlook log ic p rob lem s, w h ile paying too much attent ion to statistical techn iques. A statistically sign ifican t d ifferen ce can dem onstrate treatm en t effect for a w ell-des ign ed expermi ent, bu t it m igh t not for a bad expermi en t. Any extran eous variab les th at change sy stem atically along w ith th e treatm en t th reaten the in terna l va lid ity of an exp ermi en t. Th e b ias in som e cases arises from in terd epend ence betw een expermi en tal un its, w h ich is the case w ith pseudo-rep licat ion, a te