# So get out and publish

Document Sample

```					10b       COMBINED EXPERIMENT OF MULTIPLE-HARVEST TYPE

1.    AN EXAMPLE; PERENNIAL CROP WITH MULTIPLE HARVESTS

The data are yields of Alfalfa from 3 reps, 4 lime treatments, 4 successive harvests

Harvest Lime Trt.    Rep 1     Rep 2    Rep 3 SUM
No. 1        A           97       100      128       325
B           89       115      123       327
C           77       123      142       342
D          100       120      122       342
No. 2        A          166       158      155       479
B          161       169      161       491
C          184       178      168       530
D          197       169      169       535
No. 3        A          134       132      125       391
B          128       142      135       405
C          147       121      138       406
D          146       140      143       429
No. 4        A           90        88      100       278           n=         48
B           92        88       89       269          AVG X=     129
C          109       103      109       321          SS Tot = 41994
D          118        98      106       322          SS Rep =227.63
SUM                    2035      2044     2113      6192

2.    SUMMARY TABLES

SUMS for treatments and reps (combining the 4 harvests)
Trt.    Rep 1     Rep 2     Rep 3    SUM        AVG
A           487      478      508    1473      122.8
B           470      514      508    1492      124.3
C           517      525      557    1599      133.3
D           561      527      540    1628      135.7

SUMS for harvests and reps (combining the 4 treatments)
Harv.  Rep 1      Rep 2    Rep 3    SUM         AVG
No. 1     363       458      515    1336       111.3
No. 2     708       674      653    2035       169.6
No. 3     555       535      541    1631       135.9
No. 4     409       377      404    1190        99.2

3.    CALCULATION OF REQUISITE SS VALUES
SS TRTS=         1480.2 SS Harvests =            34764    SS Harv*Rep =        3306.5
SS Trt*Rep =     546.71 SS Trt*Harv =           415.33

4.    ANOV FOR COMBINED HARVESTS AS A SPLIT-PLOT

First, we will treat the 4 harvests as a random sample of values "nested" within the plots, i.e., with
lime treatments as MAIN PLOTS and harvests as Sub-Plots in a classical split-plot arrangement.

Problem 10b (continued)

Source                    df        SS        MS            F    F05,01
TOTAL                     47     41994.0
Lime Treatment             3      1480.2     493.4     5.41 4.76,9.78
Rep                        2       227.6     113.8     1.25
Error (a)                  6       546.7      91.1
Harvest                    3     34763.8   11587.9    60.98 3.01,4.72
Lime x Harvest             9       415.3      46.1     0.24 2.30,3.26
Error (b)                 24      4560.3     190.0
CV =       10.69%

5.     ANOTHER APPROACH;             ANOV AS A STRIP-BLOCK

Yields of alfalfa in temperate climates change SYSTEMATICALLY, rising in the summer and
dropping in autumn. Thus, the split-plot design (which assumes Randomness of these harvest
values) is not appropriate. Much better is the strip-block design, with 3 error terms. This tests
H (3 df) against H*R (6 df), T (3 df) against T*R (6 df), and T*H (9 df) against T*H*R (18 df).
"Blocks" now is applied to the 12 totals of harvests and reps (i.e., 11 df = 2 for reps, 3 for
harvests and 6 for Harv*Rep).

Source                    df        SS        MS            F F05,01
TOTAL                     47     41994.0
Harvest                    3     34763.8    11588     21.03 4.76,9.78
Rep                        2       227.6     113.8
Harvest * Rep              6      3306.5     551.1
Trt                        3      1480.2     493.4     5.41 4.76,9.78
Treatment * Rep            6       546.7      91.1
Treatment * Harv.          9       415.3      46.1     0.66 2.46,3.60
T*H*R                     18      1253.8      69.7
CV =      6.47%

NOTE: This ANOV provided a better test of T*H interaction, since in the Split-Plot ANOV the
error (b) was very large vs. the T*H*R here. The ratio of these Error Variances is:
Residual Error Variance ratio =       2.73
6.    COMBINATION OF ANOVS FOR SINGLE HARVESTS

ANOVs are wisely completed on a template for each harvest, copying the template (including
all formulae) to successive pages of your notebook. Combining these pages by 'addition' and
adding a few new SS calculations will complete the ANOV table. As an example, here is the
ANOV for Harvest No. 1. SS values from this can be added to those of Harvest No. 2, etc.

Source           df       SS         MS         F
Trt               3       86.0       28.7    0.22
Rep               2     2948.2     1474.1   11.08
Error             6      798.5      133.1
Total            11     3832.7

Combining Trt SS over harvests = SS for Trt*Harv (and so forth).

10b    Off hard-copy

Hi. Please learn to enjoy publishing your research!! Nothing is more frustrating than to see
talented researchers "sit on their data". It is very easy to do with perennial species like
forest and fruit trees.

Here in Hawaii, old data sheets get yellowish and ugh-smelling, with rusty paperclips or staples.
It is even worse with disks of data, for they end up in strange locations, with mysterious lists
of files that you've never heard of and strange analytic approaches. And the disks probably

"Oh, if I just get one more year's data"....then all will be right? No, in all likelihood, more data
will not necessarily make better data.

reviewed ones won't accept it. And don't think for a minute that 'negative data' are
unimportant. We're all testing departure from the null hypothesis in most cases anyway!
he plots, i.e., with
ot arrangement.

ummer and
these harvest
s. This tests
H*R (18 df).
te (including
here is the
Hello. This is a reminder that this file
was created by:

Prof. James L. Brewbaker
Department of Horticulture
College of Tropical Agriculture and Human Resources
University of Hawaii, Honolulu, Hawaii 96822

Please respect its authorship. And enjoy!

```
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
 views: 2 posted: 11/5/2012 language: English pages: 141
How are you planning on using Docstoc?