Efficiency of Change Over Designs in Animal Experiments

Abstract

The role of statistical planning in ensuring the validity and exchanging the efficiency of experimentation is beginning to be recognized in dairy cattle experiments switch over design is a design particularly suited to animal experimentation. In the present work, a study has been made of the relative efficiency of the change over design in its variance forms including extra period design with the conventional designs vis., the completely randomized design and randomized block design in the case of dairy experiments. Daily records of milk yield of a herd of cows and another of buffaloes under uniform condition of feeding and management have been utilized for the purpose. The utility of allowing a part of each period under uniform treatment and taking the yield in pre-treatment period as a consistent variable has also been studied.

The change over design without blocks compared with completely randomized design was found 3 to 12 time more efficient as judge with the data on Rurrah buffaloes herd at Ambala military dairy fare and 11 to 47 time from the data on Konkrej herd as Anand. The switch over design with blocks was found to be 9 to 185 times as efficient as randomized block design with Anand data and 8 to 28 times as efficient with table data. In this investigation three criteria were used for the formation of blocks. 1. Season of solving. 2. Previous lactation yield. (3) Persistency of milk yield based on the previous lactation yield.

It was concluded that formation of blocks based on persistency of milk yield provided greater efficiency that other two. Surprisingly the design with blocks based on season of claving proved to be more efficient than the one based on previous lactation yield. Three duration of period under each treatment of 50, 60 and 80 days were tried.

The extra period design was more efficient than the ordinary switch over design. If the total experimental period was kept constant, the relative efficiency of extra period design compared with switch over design was more than when the treatment period was kept constant and the total duration of experiment varies. The modified extra period design without blocks omitting the first period was found slightly more efficient than the extra period design without blocks. The modified design with blocks was more efficient than the corresponding extra period design in by 608 cases. The relative efficiency of modified extra period design vis-à-vis the ordinary switch over design was more than the total experimental period was kept constant than when it was varied.

The adoption of covariance technique eliminated considerable error variation whether the design was with or without blocks. The use of blocks in switch over design but without utilization of consistent variate was found almost equally efficient or the design without blocks but with the utilization of consistent variate. It was observed that as the number of treatments was increased for constant duration of treatment period, the relative efficiency decreased in almost all the cases. It was observed that as the duration of treatment period increased for constant treatments, the efficiency decreased.