Why is randomization necessary in ANOVA experiments?

[gtfitzpatrick]

Homework Statement

Tablets were made on six different tablet presses from an available 10 during the course of a run (batch). Two tablets from each press were collected during each hour of the 4-hour run. The tablets were then assayed in random order. The results are as follows:
Press
Hour 1 2 3 4 5 6

1 47,49 49,48 46,47 49,50 47,49 50,51
2 48,47 48,47 48,50 47,49 50,49 50,49
3 50,49 47,49 50,49 48,50 51,49 50,48
4 49,48 46,48 50,49 49,48 47,49 49,51

(i) Are the presses and hours fixed or random? What does this mean?
(ii) Explain why randomization is necessary in this experiment. How might randomization be achieved?
(iii) State all the null hypotheses that may be tested. Test each hypothesis (i.e. say whether you accept or reject it, giving reasons for your decision), working in an appropriate order. State your conclusions.
(iv) Outline two advantages of using two-way ANOVA, rather than two separate experiments, to investigate the effects of press and time on assay.

Homework Equations

The Attempt at a Solution

i)the presses are random as 6 were chosen from a possible 10. The hours are fixed as we are looking at the entire run.This means we should use the balanced anova?
ii)randomisation is a way of guarding against axtraneous variations regardless of wether or not we are aware of the source or even the presence of such variation.
randomisation of the run order of individual experiments?
tablet asseyed in rndom order?

iii)im not sure where to start here, i inputed the data into minitab in 3 columns presses tables and assay., I am not sure if this is right or where to go from here...

iv)1 andvantage is that doing multiple 2-sample t-tests would result in an increased chance of committing a type 1 error(ie wrongly rejecting a true null hypothesis)

 

[lippyka]

2nd advantage is that ANOVA allows for the testing of interaction effects, this would not be possible using multiple 2-sample t-tests