0

I set up an experiment design (CCD) with

6 factors: H2, CO, C2H2, temperature and humidity

5 levels: -alpha, -1, 0, 1, alpha

Test units are gas sensors (2 manufacturers, each 20 sensors)

How do I calculate any effect sizes? Basically I want to know: Are there effects and how strong are they? E.g. how strong is the influence of C2H2 or humidity?

And in addition: How are the manufacturers performing against each other?

Ben
  • 3,443
  • 1
    Could you provide some additional information such as the number of factors tested and the number of levels per factor? The purpose of design of experiments is to determine which factors have a significant effect on the outcome. If the experiment is balanced then one can gain insight on the effect size by comparing the outcome from the low level to the high level for each factor. – Dave2e Apr 23 '20 at 16:13
  • @Dave2e Thanks a lot! I added the numbers but I think you already gave the answer, or? I just need a bit clearification. Let's say a factor is e.g. relative humidity. To figure out the effect size, I would compare those low levels against those high levels? – Ben Apr 29 '20 at 12:59
  • I wonder if I would rather use the star or the fractional points for the low/high levels? – Ben Apr 29 '20 at 13:47
  • When you’ve read about $n>30$, what has been the context? I’ve typically heard that discussed when we don’t know the distribution of our data and want to invoke the central limit theorem to assert approximate normality of $\bar{X}$ for a non-normal distribution. (A joke in statistics is that $30=\infty$. This can fail, but you will see $30$ used this way.) – Dave Apr 29 '20 at 13:48
  • @Dave Meanwhile it can be connected to two aspects: I read that below n<30 you would use the t-distribution instead of normal distribution and in relationship with repeatability I asked this already https://stats.stackexchange.com/questions/446882/number-of-observations-to-study-the-reproducibility – Ben Apr 29 '20 at 13:51
  • @Ben, I am completely confused on what you actually tested and what you current question is. I would suggest starting over with this question and clearly stating what your experimental objective was and clearly explaining what was measured and how where the factors varied. – Dave2e Apr 29 '20 at 14:50
  • I updated resp. rephrased the question. Better to understand now? – Ben Apr 30 '20 at 08:52

0 Answers0