Taking it to an obviously absurd extreme to illustrate the point

Ah, the reducio ad absurdum debating technique. It's one of my favorites. However, my all time favorite debating technique is "proof by vehement assertion".   

The points you bring up are all reasonable. However, I would suggest that the attack 8 defense 4 scenario for which Polarpawn spent the time and effort in getting the data is not too unreasonable. I don't think anyone would confuse this discussion with a proof of Fermat's last theorem. I'm actually interested in getting some result even if it is somewhat flawed. I think it's sufficient if we get a reasonable indication of what's going on here. I for one do not require proof beyond a reasonable doubt and frankly I begin to tire of the subject.

I agree entirely.

As I said, my suggestion was a bit late!

It's nice to see somebody aware of the reductio ad absurdum method. Painfully often, the response is "but that could never happen so your argument is stupid". *sigh*.

I am also (overly) fond of the "proof by vehement assertion" technique. In my (younger) youth it earned me the birthday gift of a mug with "everybody is entitled to my opinion" written on it. I often realise that time has tempered things only slightly.

This is a graph I made with Matlab of the averages and their standard deviations.

We wanted to know if there's a difference between experienced ships and rookie ships regarding damage inflicted (first two bars) and the difference between experienced ships and rookie ships ragarding damage received (last two bars).

Even though the variables were not really normally distributed (bell shaped), I believe the number of samples justifies using the t-test for analysis. From the graph, it's clear that there's little difference comparing bar 1 and 2 and comparing bar 3 and 4 (Rookie damage inflicted is even slightly higher!)
(for clarity: it doesn't make sense comparing bar 1 and 2 with 3 and 4, because they're different variables, can't compare apples with pears).

I performed two t-tests for independent samples:
1. Comparing veteran and rookie on damage inflicted:

EXPLEVEL N Mean Std. Deviation
Veteran 65 3.1538 2.29967
Rookie 65 3.1846 2.06074

p = .94, meaning It's 94% certain that the found difference between the averages is coincidence. I conlude that there is no significant difference between veteran and rookie ships regarding damage inflicting capabilities.

2. Comparing veteran and rookie on damage received:

EXPLEVEL N Mean Std. Deviation
Veteran 65 1.6308 1.05430
Rookie 65 1.6462 1.20456

p = .94, meaning It's 94% certain that the found difference between the averages is coincidence. I conlude that there is no significant difference between veteran and rookie ships regarding damage receiving capabilities.

Final conclusion:
Veteran ships do not shoot better or have better protection than Rookie ships of the same type.

Disclaimer: other methods could arguably be better/more advanced/more appropriate, but I used this one for its relative simplicity.

Ps. someone mentioned confidence intervals; the SPPS output generates 95% confidence intervals of the difference, but the information given here should be enough I believe. If anyone wants more details, just let me know.

Thanks Polarpawn, for gathering the data, and every one else in this thread for the discussion and suggestions!

Thanks to SideMancer for performing the analysis. I don't think anyone can doubt the outcome. Just like I've been saying all along.    Oh wait, no I didn't, darn.    Anyway, this goes to show how poor a subjective opinion can be when it comes to statistics.

Ya, I was disappointed to, but the proof is in the graph- I'm glad I we finaly have an answer. Thank you Mumblefratz, Sidemancer (For all your work and the education you gave me on Statistics) and everyone else!