6,078 reputation
1229
bio website integrativestatistics.com
location Boston, MA
age
visits member for 3 years, 8 months
seen 16 hours ago

Aug
25
awarded  Good Question
Aug
23
revised Combining categorical and continuous variables to calculate a factor
added 597 characters in body
Aug
23
revised Combining categorical and continuous variables to calculate a factor
avoid ambiguity.
Aug
23
answered Combining categorical and continuous variables to calculate a factor
Aug
23
comment Seemingly unrelated regressions and difference in coefficients across groups
Hello - could you define/explain 'seemingly unrelated'? Typically differences in coeff. by group would be tested using interaction terms.
Aug
20
comment Call Centre Models
The revisions look helpful and hopefully someone will have more to say to answer you.
Aug
19
comment Use of factor analysis + regression
Not a good one. Any such guide would have to simplify so much that it would use a lot of methodologically suspect shortcuts such as "Tom Swift's Electric Factor Analysis Machine" or the notoriously risky stepwise regression algorithms. I'm afraid you've entered an area that requires long hours of study if you are looking for sound results and/or an understanding of causal relationships. Welcome!
Aug
19
comment Call Centre Models
It would help if you would be clearer about what needs to be predicted. Otherwise some fool will answer, "I don't need methods. I hereby predict calls at a call center" and he/she will be right.
Aug
19
comment High Standard Deviation for Leave one out cross-validation?
The 0.5 doesn't sound wrong, but I don't see why standard deviation is informative in this case. Also, are you sure the "high variance" concern doesn't apply to prediction accuracy as measured by something other than this binary variable?
Aug
17
comment piloting a multinomial logistic regression model
Several points will need clarification. "Multinational" = multinomial? Your output doesn't seem to fit a multinomial model: if it did you would have obtained three lists of coefficients since your dependent variable has 4 levels. Perhaps you could show your code (the commands you used)...If "farmers mark" was continuous, what did it mean and why did you categorize it into groups?
Aug
17
comment How to interpret if my sample statistic is way out in the tail of the bootstrap distribution
Thank you for pointing that out.
Aug
16
comment small sample size, large number of variables (most categorical) - how to proceed?
I upvoted the question as well-considered and well-written. Still, I think ultimately you'll want to have an in-depth conversation with a mentor or consultant rather than rely on Q&A here. That way the person could consider your goals, the audience for your analysis, the level of rigor you want to maintain, the amount of time you have to devote, and your level of expertise with different methods.
Aug
16
comment How to interpret if my sample statistic is way out in the tail of the bootstrap distribution
I think your main point is valuable, by I don't see why you assume that each discards 1/e of the data.
Aug
10
comment Is the population of blue-eyed Martians decreasing?
"decreasing throughout the 20th century" is too vague a phrase to allow for an hypothesis test. One could interpret it in various ways. 1) as "each proportion must be less than the preceeding one," a condition that is clearly not met here. 2) as "the 2000 value must be less than the 1910 value," which is met. 3) as "there will be a negative coefficient for a time variable in a logistic regression (where there are 250 blue-eyed and 150 others in 1910, 1000 and 1000 in 1920, etc.)"
Aug
9
comment SPSS: Pearson's r not significant but confidence intervals do not include 0
Doesn't it concern you that your initial bootstrapped confidence interval doesn't contain the observed value of r?
Aug
7
comment What kind of analysis gives you the statement "If you DONT reach X amount by time T, then your chances go down by P percentage?
You've used the tag "survival": survival analysis (time-to-event or duration analysis) should be fruitful here.
Aug
7
comment How to combine 2 variables each be strongly correlated with a 3rd variable?
Yes: regression (ordinary least squares regression) will enable you to create, using a weighted combination of a and b, a predicted version of c that correlates more highly with c than either a or b does.
Aug
7
comment How to build a prediction model for exam score based on previous scores
I have edited the question substantially to make the wording fit the meaning I believed you intended. If any of my changes have distorted your meaning, please undo them.
Aug
7
revised How to build a prediction model for exam score based on previous scores
added 42 characters in body
Aug
6
awarded  Custodian