Owen và DeRouen, 1980). Có nghĩa là ước tính là một trọngCác phương trình chuyển đổi để chuyển đổi này vào dự toán giá trị trung bình và phương sai của y untransformedThay các tham số ước tính η x và σ x cho | TABLE Dioxin and Furan Data from a Designed Factorial Experiment Sample Period Sampler 1 2 3 4 A B A B A B A B Dioxins Sum TetraCDD Sum PentaCDD 28 11 Sum HexaCDD 24 Sum HeptaCDD 17 155 16 62 29 45 30 40 OctoCDD 55 28 14 21 12 17 Furans Sum TetraCDF 26 18 13 13 Sum PentaCDF 31 11 22 12 17 24 Sum HexaCDF 31 11 28 14 18 19 Sum HeptaCDF 103 32 80 32 41 47 62 OctoCDF 19 18 Note Values shown are concentrations in ng m3 normal dry gas at actual CO2 percentage. The data in Table were collected at a municipal incinerator by the Danish Environmental Agency Pallesen 1987 . Two different kinds of samplers were used to take simultaneous samples during four sampling periods spread over a three-day period. Operating load temperature pressure etc. were variable. Each sample was analyzed for five dioxin groups TetraCDD PentaCDD HexaCDD HeptaCDD and OctoCDD and five furan groups TetraCDF PentaCDF HexaCDF HeptaCDF and OctoCDF . The species within each group are chlorinated to different degrees 4 5 6 7 and 8 chlorine atoms per molecule . All analyses were done in one laboratory. There are four factors being evaluated in this experiment two kinds of samplers S four sampling periods P two dioxin and furan groups DF five levels of chlorination within each group CL . This gives a total of n 2 X 4 X 2 X 5 80 measurements. The data set is completely balanced all conditions were measured once with no repeats. If there are any missing values in an experiment of this kind or if some conditions are measured more often than others the analysis becomes more difficult Milliken and Johnson 1992 . When the experiment was designed the two samplers were expected to perform similarly but that variation over sampling periods would be large. It was also expected that the levels of dioxins and furans and the amounts of each chlorinated
