Major accusations of data fabrication in ecology:  A Science paper last year reported that fish were eating microplastics, which increased predation rates (or changed response to predators; something like that). Looks highly likely the experiment was never run.

Science new article: Fishy business: Accusations of research fraud roil a tight-knit community of ecologists

Retraction Watch: High-profile Science paper on fish and plastics may earn notice of concern


Brian McGill of Dynamic Ecology on “Statistical Machismo”, in particular its relevance to the need for correction for detection probability in wildlife surveys.

2012: Statistical machismo?

2013: Is using detection probabilities a case of statistical machismo?



Related Blog Posts by Brian McGill

Why OLS estimator is an unbiased estimator for GLS


Generalized Linear Mixed Models: Modern Concepts, Methods and Applications (Stroup 2012; SAS)

 Growth Curve Analysis and Visualization Using R
Daniel Mirman
Chapman & Hall/CRC, Boca Raton, 2014.
ISBN 9781466584327. 188 pp. USD 79.95 (P).


Using R and lme/lmer to fit different two- and three-level longitudinal models (Pyschology; 2015; K. MAGNUSSON)



Nested by design: model fitting and interpretation in a mixed model era (Schielzeth & Nakagawa 2013, Methods in Ecology & Evoltuion)

The journal Methods in Ecology & Evolution has a blog with a number of interesting entries, including:


  1. Jarred Hadfield: There’s Madness in our Methods: Improving inference in ecology and evolution
  2. Flawed Analysis Casts Doubt on Years of Evolution Research
  3. Making Your Research Reproducible with R
  4. Peer Review Week: Should we use double blind peer review? The evidence

Climate Change:

  1. How Should Biologists Measure Climate Change?



  1. How to Synthesize 100 Articles in Under 10 Minutes: Reviewing Big Literature Using ACA (Automated Content Analysis)

Online R and Stats Forums: stackoverflow and beyond

There are several websites where you can post statistics and R questions and other users can post responses.  The two best places for R and stats info are the forum sites stackoverflow  and Cross Validated.  Questions on these sites are often very specific, as are responses.   A sister site where people post questions on academic conduct and reproducibility, among many other things, is Stack Exchange Academia.  Users at ResearchGate also frequently post stats and R related questions.  Questions on these latter two sites are usually broader and answers more opinion based.  There are also numerous videos on YouTube where people describe various R and stats topics.

ON ASKING QUESTIONS ON Stackoverflow & Cross validated.

In general a high quality Stack Overflow (SO) and Cross Validated (CV) question OR answer involves at least a little bit organization, for e thought, and research.  For a question, you should demonstrate that you thought about the problem and tried to solve it.  Ideally you provide code with sample data that allows anyone to replicate the problem you are having (they call this a “reproducible” answer).  People will often post comments asking for clarification (usually these are nice, sometimes they are terse; try not to take it personally; sometimes people are jerks, though!)

In contrast, Stackexchange Academia has a more open culture where more general questions with opinion-based answers are perfectly acceptable.  ResearchGate is the same, and people frequently ask questions as simple as “what is a good for studying ‘x’ or “What does the term ‘y’  mean”?  These types of questions will almost always get useful responses on Researchgate while on SO this type of question would not be well received.  Sometimes you will get less than helpful answers; this site also does not allow you to include code.  Overall, ResearchGate is more like a social media platform and the answer are more often “off the cuff” and so responders may or may not have read your question very carefully.

The official “how to ask a question” info for SO:

And another take:



Here are some of the questions I’ve asked over the last five years.  You can get a sense for what makes a good question, a not so good question, and the types of helpful (and terse) answers you can get.







Some useful information for using WordPress:

1)This link provide information for formatting source code in a WordPress blog.

For example, this bit of code
[sourcecode language="R"]
runif(1, 10)

Produces this result in the blog:


To drop the line numbers, put “gutter = FALSE” in the first set of square brackets, producing this:


To prevent wrapping and add a scroll box, add “wraplines = FALSE” to the header

summary(glm(y_variables ~ x_1 + x_2 + x_1*x_2, family = gaussian, data = my_new_data[data_subset , ]))) #This code isn't really real

WordPress Toolbar

By default there is a “tool bar” that allows you to download the code, print it, etc. To remove it, add “toolbar = “FALSE” to the header.  The result is this, which does not produce a the toolbar when you scroll over the code.

summary(glm(y_variables ~ x_1 + x_2 + x_1*x_2, family = gaussian, data = my_new_data[data_subset , ]))) #This code isn't really real

2)There are options for offline editing and publishing that can be found on the WordPress site.



hist(random_poisson, ylim = c(0, 30))
abline(v = mean(random_poisson), col = 3, lty = 1, lwd = 4)
abline(v = median(random_poisson), col = 2, lty = 2, lwd = 4)
legend = c("mean", "median"),
col = c(3,2), lty = c(1,2), lwd = 4,
bty = "n"
text(x = mean(random_poisson)-0.3, y = 30, labels = mean(random_poisson))
text(x = median(random_poisson)+0.3, y = 30, labels = median(random_poisson))

alt text test

Figure 1: test