Skip to main content

Welcome

 In our role as statisticians, we see many identical questions from researchers on how to perform common statistical methods.  Rather than repeatedly give the same answers, we plan to post the current recommended practice for each type of problem, and update as needed.  Our focus will be on R and SAS software, widely used and what we have the most experience with.  Like most statistical consultants, we are educators, hoping to teach you how to think about and do statistics.
   Statistics is a broad topic, reaching into almost every area of science.  Since the different sciences have different types of data, different sources of variation, statistical methodology can be area specific.  The questions we see mainly come from agriculture, so there will be a bias towards methods most widely used there. We will strive to place sufficient key words in the posts to enable searches to uncover appropriate material.
   Time permitting, we will post on topics suggested by you, do reviews of the literature, summarize internet resources, etc.  Nothing like free statistics training, but this blog is clearly not intended for quick responses.  For that try
https://www.talkstats.com/
https://communities.sas.com/
https://stats.stackexchange.com/
to name popular statistics, SAS and R forums, respectively.

Comments

Post a Comment

Popular posts from this blog

Estimation of the Peak in Quadratic Regression

 Problem:   You are running a standard quadratic (polynomial) regression analysis, and are specifically interested in the X and Y values at the peak.  If you use standard regression software, typically there will be no option that allows the peak to be estimated, with standard errors. Example:   You are studying Growth as a function of Age.  Of particular interest is when maximum Growth occurs, and at what Age. SAS code to generate artificial data, and run the analysis is: data one; do Age=1 to 20; Growth=95 + 2.7*Age - .3*Age*Age + 5*rannor(22); end; proc nlin plots=fit; parms int=2 lin=1 quad=1; model Growth = int + lin*Age + quad*Age*Age; estimate 'Age at peak' -lin/(2*quad); estimate 'Growth at peak' int + lin*(-lin/(2*quad)) + quad*(-lin/(2*quad))*(-lin/(2*quad)); run; The standard quadratic regression model with intercept, linear and quadratic slopes, is coded into Proc NLIN which has the ability to estimate any fun...
Post a Comment
Read more

Factorial ANOVA with control treatment not integrated into the factorial

Factorial treatment designs are popular, due to advantages of research on multiple treatment factors and how they interact.  But if the design includes a control treatment that is not part of the factorial, problems occur in estimation of least squares means.  A typical example is shown here, with 2 fertilizer and 3 irrigation treatments, giving 6 factorial treatment combinations, plus a control that is defined by a 3rd level of fertilizer, and a 4th level of irrigation: Fert1:Irrig2               Fert2:Irrig1             Fert1:Irrig1 Fert2:Irrig3               Fert1:Irrig3             Fert2:Irrig2           Control Other situations might have the control sharing a level of one of the factors, for example the control might be defined as Fert2:Irrig4.  But this still causes problems with estim...
Post a Comment
Read more

Can I look at reported standard errors (SE) and decide if means differ?

No guarantees, but roughly if means differ by 3*SE then they are statistically significant.   This is based on the Least Significant Difference, which is 2*sqrt(2)*SE.  Often people use non-overlapping confidence intervals as a decision rule, but this is equivalent to 4*SE, which is a bit conservative. Things that make 3*SE fail: 1)   Actually statistical differences depend on the standard error of difference, SED, not SE.   Anything in the model that makes these differ will make the rule fail, such as covariates and blocking factors. 2)   In general, mixed models with random effects will make the rule fail, because random variance is included in SE, but not in SED.   But this will make 3*SE rule conservative, 3*SED will be even smaller.   If 3*SE suggests a statistical difference, difference most likely exists. Also take a look at Error Bars paper .
Post a Comment
Read more