Druss2k

Germany

Oct
8
accepted The meaning of translation completeness w.r.t a random variable
Sep
29
awarded Popular Question
Aug
22
revised In practice, how is the penalisation matrix for splines, created by smoothCon (package mgcv), specified?
edited title
Aug
22
asked In practice, how is the penalisation matrix for splines, created by smoothCon (package mgcv), specified?
Aug
21
awarded Enthusiast
Aug
19
revised Problems with calculation of numerical identification w.r.t. ANOVA smooth for large scale matrices
deleted 1 character in body; edited title
Aug
19
asked Problems with calculation of numerical identification w.r.t. ANOVA smooth for large scale matrices
Aug
14
revised How does the cubic spline basis exactly look like
deleted 77 characters in body; edited tags
Aug
14
revised How exactly is the sum (or mean) centering constraint for splines (also w.r.t. gam from mgcv) done?
deleted 4475 characters in body; edited tags; edited title
Aug
11
revised How exactly is the sum (or mean) centering constraint for splines (also w.r.t. gam from mgcv) done?
added 3660 characters in body
Aug
11
revised How exactly is the sum (or mean) centering constraint for splines (also w.r.t. gam from mgcv) done?
added 3660 characters in body
Aug
11
awarded Inquisitive
Aug
8
revised How exactly is the sum (or mean) centering constraint for splines (also w.r.t. gam from mgcv) done?
added 4 characters in body
Aug
7
comment Model regression of means different size and variance
Do you also count zeros, i.e., e-commerce for which no reply is posted? If so you could do some kind of poisson regression where the depended variable count the number of replys on e-commerce. Here the, though this can be relaxed, the variance changes if the mean will change.
Aug
7
comment Robust estimation in SPSS generalized mixed models
If you've observed quite a few people and the difference is mainly due to one person I would not care. Maybe youve just randomly drawn someone who eats fish all day long...
Aug
7
comment Robust estimation in SPSS generalized mixed models
If your results are based upon a type of regression-analysis it does not matter that your data is not normal. The assumption about normality are w.r.t the residuals and are used if you start doing reference such as using p-values. Usually you compare two groups to analyse if they are somewhat different when some policy was applied to one group. Here we usually draw people randomly such that the attributes of the people in each group can seen to appear with the same probability in both groups. Such an attribute is the preference to eat fish.
Aug
7
comment Robust estimation in SPSS generalized mixed models
If you have intakes measured in grams on the LHS and food group, meal type on the RHS (maybe some further grouping variables), a random effect w.r.t food group should show you that, I would guess from my personal behaviour, "fish-people" are likely to eat fish again and "not-fish-people" will stay "not-fish-people". But to answer the question how many intakes of a certain food is going to appear at a certain meal I doubt a random effect will really matter. I would guess that the interaction between food group and meal is, as you already said, very informative.
Aug
7
comment Robust estimation in SPSS generalized mixed models
But if the result shows that the intake of fish is, depending on meal and group, unlikely if compared with other intakes, this information is useful. Something you should consider is, or you already did, that grams cannot have a negative value though it is a common praxis to ignore this problem and just do something similar to OLS. Depending on the question it could be interesting why only a few persons eat a lot of fish but to answer the general question whether a certain food is consumed, the outcome that fish is unlikely to be a part of the meal is somewhat useful. In the framework
Aug
7
comment Robust estimation in SPSS generalized mixed models
So basically on the LHS of the equation you've a (4 x 2) - table of (food x #yes/#no). On the RHS is some measurement of dietary? Are you interested in the question if people who do a diet will eat differently if compared to the ones who are not doing a diet?
Aug
7
comment Robust estimation in SPSS generalized mixed models
How many different Foods are going to be tested? Do you only consider categories like fruits, vegetables, meat,..?
1 2 3 4 5