User blog:Fujihita/Design of Experiment: 20.3cm (no.2) night battle accuracy bonus test

Preface
Design of Experiment outline for 20.3cm (no.2) gun night battle accuracy bonus test.

Results and further details will be added during and after the study.
 * Work in progress.

Data collection
The goal of this experiment is to reject (or accept) the following null hypothesis:
 * Ho = "20.3cm gives the same night battle accuracy bonus as 14cm does"


 * For this experiment:
 * The measured response aka. data in this experiment is of categorical type: Miss or Hit.
 * The number of factors involved is 1 (Equipment) with 2 levels (14cm and 20.3cm no.2)

Take note that the data type being collected is categorical, therefore, Chi-squared test statistic should be used instead of t-test. Chi-squared test requires a predefined "expected" mean for all responses.

Unfortunately, we don't have a hypothetical value for the bonus. However, given the scope of this test is to prove (or disprove) the existence of a hidden bonus and not to measure the (unconfirmed) bonus, we'll use a control group with 14cm guns to set a baseline.

Considering 20.3cm (no.2) has a visible +1 acc, the control group will be equipped with 14cm guns. This should negate the visible accuracy bonus as well as possible (unconfirmed) "no gun penalty". The 14cm gun is assumed to give no special bonuses whatsoever to CA although a simple Type 0 Recon Seaplane vs. 14cm experiment could be conducted to verify the assumption.

Finally, we have the data collection sheet as follow:


 * Criteria
 * The ships used in this study must be level 1 unmodernized CA.
 * The sortie map must be World 5-3, node C
 * The formation must be Line Ahead
 * The target ship must be Ri-class flagship
 * Equipment must not have any upgrade

Analysis
Perform Chi-square analysis on 20.3cm data using 14cm's observed result as baseline. Hit and Miss are the new sample groups within 20.3cm no.2 test group.