Scribd
Upload
128 views58 pages

Analysis of Variance (Anova) : Single Factor No Restriction On Randomization

The document discusses analysis of variance (ANOVA) for a single factor experiment without restrictions on randomization. It provides an example of conducting a randomized experiment with 25 trials where the order of trials is determined randomly by drawing numbers. The document contrasts this with a non-randomized experiment where all trials for one treatment level are conducted together before moving to the next level. It then discusses the objectives and aspects of experiments that should be randomized.

Uploaded by

Nasikhuddin
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
128 views58 pages

Analysis of Variance (Anova) : Single Factor No Restriction On Randomization

The document discusses analysis of variance (ANOVA) for a single factor experiment without restrictions on randomization. It provides an example of conducting a randomized experiment with 25 trials where the order of trials is determined randomly by drawing numbers. The document contrasts this with a non-randomized experiment where all trials for one treatment level are conducted together before moving to the next level. It then discusses the objectives and aspects of experiments that should be randomized.

Uploaded by

Nasikhuddin
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 58

Analysis of Variance ( ANOVA )

Single Factor No Restriction On Randomization

Contoh :

Untuk melakukan eksperimen secara random, maka ke-25 percobaan tsb bisa dilakukan dengan cara : Misal, diambil secara random (acak) angka antara 1-25, kemudian diperoleh angka 8. Maka pada angka 8 tersebut akan dilakukan eksperimen yg pertama kali. Proses seperti ini terus dilakukan hingga ke 25 eksperimen selesei dilakukan. Misal hasilnya sebagai berikut :

Bentuk eksperimen Non-Randomized Order


( Menggunakan contoh yg sama )

Misal, bentuk eksperimen pertama : All five 15% cotton speciments are tested first Eksperimen berikutnya : All five 20% cotton speciments are tested next, and so on

Hasilnya sbg berikut :

Informasi :
Berdasarkan ilustrasi contoh diatas :

Yang dikatakan sebagai faktor adalah Cotton Weigth Percent Treatment [ hanya ada 1 faktor ]

Level-nya : ada 5 Level (15, 20, 25, 30, 35) Persen. Replicates (Replikasi/Perulangan, n) = 5 kali

Jumlah pengamatan = 25

Randomized

Tujuan : To prevent the effects of unknown nuisance variable, perhaps varying out of control during the experiment from contaminating result.
Possible Nuisance Variable :

Operator differences Raw material changes Batch/lot differences Equipment wear, etc

What should be randomized ?


Assignment of experimental units to treatments Order of running experiments Order of evaluating experimental results

The Analysis of Variance

Misal, terdapat a treatments atau different levels of single factor yang ingin kita bandingkan, Dan, pengamatan setiap a treatment berupa random variable,
Maka bentuk datanya bisa ditampilkan sbg berikut :

Models for the Data

Where : Yij is the ijth observation ij is the mean of the ith factor level or treatment ij is a random error component that incorporates all other sources of variability in the experiment including measurement, variability arising from uncontrolled factors, differences between the experimental units (such as test material, etc) to which the treatments are applied, and the background noise in the process.

is a parameter common to all treatment called the Overall Mean is a parameter unique to the ith treatment, called ith treatment effect

Tabel Anova Single Factor, Fixed Effects Model

Contoh :
The development engineer is interested in determining if the cotton weigth percentage in a synthetic fiber affects the tensile strenght, and he has run a completely randomized experiment with five levels of cotton weight percentage and five replicate.

Perhitungan :

Penyederhanaan data
Bentuk awal :

Bentuk penyederhanaan : (Tiap data dikurangi dengan 15)

Estimation of the Model Parameters

Contoh

Tabel distribusi t

Unbalanced Data
Banyaknya data observasi pd tiap treatment tidak sama (unbalanced)

Contoh : ANOVA Unbalanced Data

Bagaimana Penyeleseiannya ??

Uji Setelah ANOVA

Perbandingan antar treatments

CONTRASTS

Uji Hipotesa Contrast

Uji t dan Uji F ( Pd kondisi Sample size in each treatment are equal )

Uji t contrast
untuk Sample size in each treatment are unequal

Contrast
Untuk menguji Mean Jumlah contrasts TIDAK BOLEH melebihi jumlah degree of freedom between treatment means.

Orthogonal Contrasts

Contoh :

TUGAS Individu
1.

2.

Buat resume (sejelas mungkin) tentang Scheffes Method Buat 1 soal + Jawabannya untuk Scheffes Method

Tugas ditulis tangan, dikumpulkan awal pertemuan selanjutnya.

Scheffes Method ??

Metode Scheffe

Scheffes (1953) has proposed a method for comparing any and possible contrasts between treatment means. Menguji rataan treatment yang telah dibentuk dalam kontras. Pd uji Scheffe, Kontras yang akan diuji TIDAK HARUS bersifat ORTHOGONAL. Kontras, digunakan untuk membandingkan rataan antar/beberapa treatment. Untuk melihat treatment mana yang memberikan pengaruh signifikan.

Menguji Hipotesa Dgn Scheffes Method

differs significantly from zero (persamaan) kontras yang dibandingkan tidak sama dengan nol mengarah kepada hipotesa H1. Jika terjadi kondisi , maka (persamaan) kontras sama dengan nol (H0) DITOLAK

Contoh :

Dengan menggunakan data pd contoh persoalan Cotton Weight Percentage. Misal, kontras yg ingin diuji :

(1) Apakah terdapat perbedaan rataan pengaruh antara gabungan/perpaduan treatment 1 dan 3 dengan treatment 4 dan 5 ? (2) Apakah terdapat perbedaan rataan pengaruh antara treatment 1 dengan treatment 4 ?

Solusi :

Kesimpulan : TERIMA Ho untuk Kontras yang pertama Tidak cukup bukti untuk menyatakan bahwa perpaduan treatment 1 dan 3, akan memberikan hasil yang berbeda dengan perpaduan treatment 4 dan 5.

TOLAK Ho untuk kontras yang kedua Dapat dikatakan bahwa rataan pengaruh treatment 1 dan 4 akan memberikan hasil yang berbeda.

Uji Rataan Treatment


Untuk menguji rataan seluruh treatment, Ho : i = j untuk semua i j, digunakan uji :
Tukeys test The Fisher Least Significant Difference (LSD) Method Duncans Multiple Range Test The Newman-Keuls Test

Tukeys Test

Tukey (1953) proposed a procedure for which the overall significance level is when the sample size are equal and at most when the sample size are unequal.
His procedure can also be used to construct confidence intervals on the differences in all pairs means. For these intervals, the simultaneous confidence level is 100(1- ) percent when the sample sizes are equal, and at least 100(1- ) percent when sample sizes are unequal.

Ingin diuji, dengan dugaan awal apakah :

Maka :

Metode mana yg akan digunakan ?

You might also like