I have a set of healthy control individuals and several different patients (with different diagnosis, so I am not trying to compare them as a group. From now I will only refer to a single patient compared to all controls). I make a measurement of a certain biochemical parameter for each control and the patient with several repetitions. This means I have for example 12 measurements for 4 different control individuals, 3 for each (in reality, to be mor complicated, the number is not always the same) and 3 measurements for my patients. I am trying to compare the patient to the controls as a single group. I am using Statistica software (that's what we have in my lab and we are all statistical analphabets). I found this simple explanation http://www.biostathandbook.com/nestedanova.html?fbclid=IwAR0BGo1XFXzd-MVTLZ5_4ivp1WUOrP51as7KU64Y29OnjqcDNabR3zU7b7A and it seemed to me like exactly what I want. Tried to use it in Statistica, in the Nested design ANOVA dialog, the syntax is on picture with a row saying "DESIGN: skupina + id(skupina). "Skupina" means group, that is what I named the first independent variable, the values for it are either "control" or "patient" . ID is an id for each control and patient, so 1-4 for controls and 5 for the patient. I put ID and "skupina" as independent and ID as random effect. So far so good even though I have no idea whether it's correct. However, right before he left for his holiday, I was given an advice from a professional statistician (I guess the easiest "not wrong" way, beacause of my incompetence), and the syntax looks as is on the other picture, with DESIGN: skupina;. This was made from a general linear model in the General linear models dialog by making "skupina" and id independent (I guess it's called categorical factors) and the measurements dependent variables, then making the id a random factor and then erasing it from the syntax, changing it from "DESIGN: skupina + id;" to "DESIGN: skupina". I guess I believe that he is right, however I also find the ANOVA explanation convincing for my use. I don't know whether I used it right, only basis for that is this instruction from makers of the software: http://documentation.statsoft.com/STATISTICAHelp.aspx?path=Gxx%2FGlm%2FExamples%2FExample5MixedModelNestedANOVADesign&fbclid=IwAR1UfelnDjtnOgVh3CqlJ8i9GQfqYGgNf_226hxOSih_lfe4O1bEDl8Fa3Q I have 2 questions: 1) understanding what does the nested ANOVA do and take into account (very, VERY superficially, based just on the article) and comparing it to the syntax, I absolutely can't wrap my head around what his model does, takes into account, in human language, and how is it different and why. 2) If I were to use it (I have to say a find the p values from it much more useful for my thesis :-D :-D, and beacause I feel like he knew what he was saying, he saw the raw data, or at least big part of them), what would I call it and how would I describe it in the "Statistical methodes" section of my thesis. Thank you very muchenter image description hereenter image description here


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.