Reputation
Next tag badge:
149/100 score
19/20 answers
Badges
14 244 483
Newest
 Good Answer
Impact
~4.4m people reached

16h
reviewed Close Algorithms implemented in Java to measure multi-core CPU performance
16h
reviewed Close Calculate expected returns
16h
reviewed Close Mixed models: Is there any way we could decompose the estimate variations by covariates?
16h
reviewed Leave Closed Will PCA also produce multicollinearity?
17h
comment Probability of descendants
This question seems closely related to stats.stackexchange.com/questions/17402 . The difficulty in the present case is not estimating the probability, but making appropriate assumptions and collecting relevant information. Without that, this question is indeed too broad to be answerable. Could you be more specific about what model you have for populations and for the process of creating descendants?
17h
reviewed Leave Closed Dependent/Conditional Probability over time
17h
comment Card deck shuffling, what is the probability that four lowest card of a suit are distributed to all four players?
This is strange way of dealing cards! Most of the time more than 13 of at least one suit appears in your sample... . The results have no evident connection to the situation described in the question. Doesn't it seem strange that they differ substantially from the two theoretical answers already offered?
17h
comment Linear model comparison - which does my data fit best?
I was not hinting that one should use the logarithm to reduce ranges of data. In this specific case, even your sample data provide good statistical indications that you should be using the logarithm. In addition, the fact that ratios matter is a strong indication to use logarithms. Taking logs is so simple and fast it's hard to construe it as an obstacle in any effort to understand or analyze data.
18h
revised Measure dependence/correlation of two variables in a multivariate setting
edited tags
18h
comment Linear model comparison - which does my data fit best?
Is it possible some of your observations could come from table salt water and others from seawater? If not, and you are sure the solution is one or the other, then could you assign probabilities to each of those possibilities? If none of the above, then please tell us more specifically about your circumstances and objectives.
19h
comment Determining the required sample size to meet a bound
Suppose you knew the average of three quantities happened to be $21.5$. Could you deduce the total of those quantities? If not, then please review what you know about averages and look at the formula.
19h
comment Dependent/Conditional Probability over time
Thank you for the additional information. It still appears to be incomplete, though: precisely how are the three buckets chosen for filling out of each six? And what exactly are you trying to compute at each time step? The chance that all buckets are full? The expected number of empty buckets? The probability distribution of the numbers of full buckets?
19h
comment Linear model comparison - which does my data fit best?
That's good information to know about the outlier--it will spare us from focusing unnecessarily on an incidental aspect of your example. It would be interesting to see a brief statistical summary of the values in log(Na) - log(Cl) (such as the output of summary) for your entire dataset.
19h
comment Linear model comparison - which does my data fit best?
It's a good question (+1). But in your particular example a severe problem exists: one observation (10% of the total) is far from the others. It will control every aspect of your estimates, but you hardly dare delete it outright because it represents a substantial fraction of the data! Since you are analyzing chemical concentrations, it is natural--and scientifically meaningful--to express them as logarithms. When you do that, your problematic value becomes much less of an issue, too. The answer is then pretty clear.
19h
reviewed Close Basic questions about a econometric analysis of paneldata
19h
reviewed No Action Needed Analyzing series of events, controlling lengths
19h
reviewed No Action Needed Question running poLCA in R
19h
comment Determining the required sample size to meet a bound
Once you identify the mathematical relationship between a total and an average, you will have answered your question.
19h
comment Predicted Probabilities Using Multi-level Logistic Regression
But what does that have to do with average predicted probabilities?
19h
comment How to understand degrees of freedom?
@Clarinetist The principal point of my answer is to suggest that what you have been taught is based on a confusion of two concepts of DF. Although that confusion causes no problems for standard least-squares Normal-theory models, it leads to errors even in simple, common circumstances like analyses of contingency tables. That matrix rank gives the functional DF. In a least-squares linear model it happens to give the correct DF for certain kinds of tests, such as F tests. For the chi-squared test, the special conditions are enumerated later in the answer as points (1) and (2).