# first: install dependent packages
install.packages(c("MASS", "akima", "robustbase"))
 
# second: install suggested packages
install.packages(c("cobs", "robust", "mgcv", "scatterplot3d", "quantreg", "rrcov", "lars", "pwr", "trimcluster", "parallel", "mc2d", "psych", "Rfit"))
 
# third: install WRS
install.packages("WRS", repos="http://R-Forge.R-project.org", type="source")

WRS cannot be hosted on CRAN, as CRAN demands help files for every user-visible function. This has not been done for WRS (yet). For the time being, this somewhat more complicated installation routine has to be used.

Other fields of science additionally established a pre-publication open peer review (also called the “pre-print culture”). Many researchers in mathematics or physics publish their preprints on arXiv and harvest open peer commentaries before submitting the manuscript to a peer-reviewed journal.

I believe devoutly that open PrePPR and PostPPR can significantly improve the quality of scientific output. But one crucial requirement indeed is etiquette, as Sanjay pointed out. I don’t want to see shitstorms coming over scientific articles, especially in the case of young scholars who worked hard to get their first paper published. Comments should be written in the spirit of a collaborative enhancement of research, and less in terms of “debunking”. We all are humans and mistakes can occur. Problems should be pointed out in order to strengthen scientific research, but in a friendly and constructive manner.

Researchers who conceive of science as a highly competitive business where claims have to be fortified and defended might have problems with open peer reviews (e.g., the escalation of the “Bargh rampage” [1][2][3]). But if we see science as a collaborative endeavour in search for knowledge, where no model is “right” but only “less wrong”, open peer reviews can be a very helpful tool.

————–

Some further readings:

• Neanderthal sex debate highlights benefits of pre-publication (news blog on nature.com)
• Interesting post by Jeremy Fox (and commenters), argueing for (standard) pre-publication review, from the perspective of marine biology and ecology
• The case for arXiv and a broader conception of peer-reviews (by Philippe Desjardins-Proulx)

We at IRET have made it to our mission to proliferate and foster creative ways of data analysis. Therefore, we proudly introduce an award in recognition of outstanding data creativity: the CREDAM Award. CREDAM is both an acronym (CREative DAta Management), and a statement: credam (lat.) means “I will believe”, or “I will trust”.

This years CREDAM Award goes to …….. the German government!

A new report on poverty in Germany is going to be published soon. What does the data say?

Seems like a pretty clear picture, and in a previous version of the report, the authors concluded (based on this and other data), that “income disparity increased” (see Süddeutsche Zeitung). But that is wrong!! But why is it wrong? Well, that interpretation “does not reflect the opinion of the German government”.

On the pressure of the leader of the minor coalition partner, Philipp Rösler (which currently would be elected by 4% of Germans), this conclusion was re-interpreted. Now, the report comes to the completely opposite conclusion: “income disparity decreases“!

As this is a great example of creative data analysis, which liberates us from restrictive and anally retentive “scientific” procedures, we are happy to award the first CREDAM trophy to the German government, especially Phillip Rösler. Congratulations!

(Maybe we should think about adopting this strategy for scientific reports as well. Given highly flexible approaches of data analysis, conclusions should rather be based on a majority vote of all (co-)authors and reviewers, not on empirical evidence.)

From the original post:

This is the evolution of a bivariate correlation between two questionnaire scales, “hope of power” and “fear of losing control”. Both scales were administered in an open online study. The video shows how the correlation evolves from r = .69*** (n=20) to r = .26*** (n=271). It does not stabilize until n = 150.

Data has not been rearranged – it is the random order how participants dropped into the study. This had been a rather extreme case of an unstable correlation – other scales in this study were stable right from the beginning. Maybe this video could help as an anecdotal caveat for a careful interpretation of correlations with small n’s (and with ‘small’ I mean n < 100) …

The right panel now displays the correlation in each step. The horizontal green line is the final correlation that is approached, the curved dotted line shows the marginal correlation that would be significant at that sample size. As the empirical curve always is above this dotted line, it is significantly different from zero in each step.

Evolution of correlations – improved from Felix Schönbrodt on Vimeo.

Here the code that created the movie. It’s not fully self-contained – the function plotReg plots the dual-panel display, dat0, A, and B are parameters passed to this function. You can insert any other function here. The function loops through the rows of a data frame and saves a plot at every step into a subfolder. Finally, the function needs the command line version of ffmpeg, which connects the pictures to a movie.

makeMovie <- function(fname, dat0, A, B, fps=15) {
 
	# create a new directory for the pictures
	dir.create(fname)
 
	# create the picture sequence
	picName <- paste(fname, "/", fname, "_%03d.jpg", sep="")
	jpeg(picName, width=800, height=450, quality=95)
	for (i in 15:nrow(dat0)) {
	  print(i)
	  plotReg(A, B, i, keep=15)
	}
	dev.off()
 
	# delete any existing movie file
	unlink(paste(fname,".mpg",sep=""))
 
	# point system to R's working directory
	system(paste("cd ", gsub(" ", "\\ ", getwd(), fixed=TRUE)))
 
	# show & execute the command line expression for ffmpeg to glue the pictures together
	print(paste(paste0("ffmpeg -r ", fps, " -i ", fname, "/", fname, "_%03d.jpg -sameq -r 25 ",  paste0(fname,".avi"))))
	system(paste(paste0("ffmpeg -r ", fps, " -i ", fname, "/", fname, "_%03d.jpg -sameq -r 25 ",  paste0(fname,".avi"))))
}

Evolution of parameters in optimization process from Felix Schönbrodt on Vimeo.

More on estimating parameters of differential equations is coming later on this blog!

Things I’ve learned:

• ffmpeg does not like pngs. They are internally converted to jpg in a very low quality and I could not find a way to improve this quality. Lesson learned: Export high quality jpgs from your R function
• Use a standard frame rate for the output file (i.e., 24, 25, or 30 fps)
• My final ffmpeg command: ffmpeg -r 10 -i modFit%03d.jpg -r 25 -b:v 5000K modelFit.avi
  • -r 10: Use 10 pictures / second as input
  • -i modFit%03d.jpg: defines the names of the input files, modFit001.jpg, modFit002.jpg, …
  • -r 25: Set framerate of output file to 25 fps
  • -b:v 5000K: set bitrate of video to a high value
  • modelFit.mp4: video name and encoding type (mp4)

In version 0.20, a number of functions dealing with ANCOVA have been added and some others improved. Unfortunately, only very few help files exist for the functions. I would recommend to check out the source code, as most functions have a comment section roughly explaining the parameters. Alternatively, consult Wilcox’ books for descriptions of the functions.

References:
Wilcox, R. R. (2012). Introduction to Robust Estimation and Hypothesis Testing, 3rd Ed. Academic Press.
Wilcox, R. (2012). Modern Statistics for the Social and Behavioral Sciences: A Practical Introduction. CRC Press.
Wilcox, R. R. (2010). Fundamentals of Modern Statistical Methods: Substantially Improving Power and Accuracy. Springer, 2nd Ed.
Wilcox, R. R. (2009). Basics Statistics: Understanding Conventional Methods and Modern Insights. New York: Oxford.

First, you have to install the command line tool pdftotext (a binary can be found on Carsten Blüm’s website). Then, run following script within a directory with pdfs:

# helper function: get number of words in a string, separated by tab, space, return, or point.
nwords <- function(x){
	res <- strsplit(as.character(x), "[ \t\n,\\.]+")
	res <- lapply(res, length)
	unlist(res)
}
 
# sanitize file name for terminal usage (i.e., escape spaces)
sanitize <- function(str) {
	gsub('([#$%&~_\\^\\\\{}\\s\\(\\)])', '\\\\\\1', str, perl = TRUE)
}
 
# get a list of all files in the current directory
fi <- list.files()
fi2 <- fi[grepl(".pdf", fi)]
 
 
## Parse files and do something with it ...
res <- data.frame() # keeps records of the calculations
for (f in fi2) {
	print(paste("Parsing", f))
 
	f2 <- sanitize(f)
	system(paste0("pdftotext ", f2), wait = TRUE)
 
	# read content of converted txt file
	filetxt <- sub(".pdf", ".txt", f)
	text <- readLines(filetxt, warn=FALSE)
 
	# adjust encoding of text - you have to know it
	Encoding(text) <- "latin1"
 
	# Do something with the content - here: get word and character count of all pdfs in the current directory
	text2 <- paste(text, collapse="\n")	# collapse lines into one long string
 
	res <- rbind(res, data.frame(filename=f, wc=nwords(text2), cs=nchar(text2), cs.nospace=nchar(gsub("\\s", "", text2)))) 
 
	# remove converted text file
	file.remove(filetxt)
}
 
print(res)

… gives following result (wc = word count, cs = characgter count, cs.nospace = character count without spaces):

> print(res)
                                                    filename    wc     cs cs.nospace
1                              Applied_Linear_Regression.pdf 33697 186280     154404
2                                           Baron-rpsych.pdf 22665 128440     105024
3                              bootstrapping regressions.pdf  6309  34042      27694
4                            Ch_multidimensional_scaling.pdf   718   4632       3908
5                                               corrgram.pdf  6645  40726      33965
6                   eRm - Extended Rach Modeling (Paper).pdf 11354  65273      53578
7                                           eRm (Folien).pdf   371   1407        886
8  Faraway 2002 - Practical Regression and ANOVA using R.pdf 68777 380902     310037
9                             Farnsworth-EconometricsInR.pdf 20482 125207     101157
10                                           ggplot_book.pdf 10681  65388      53551
11                                       ggplot2-lattice.pdf 18067 118591      93737
12                               lavaan_usersguide_0.3-1.pdf 12608  64232      52962
13                                  lme4 - Bootstrapping.pdf  2065  11739       9515
14                                                Mclust.pdf 18191  92180      70848
15                                              multcomp.pdf  5852  38769      32344
16                                       OpenMxUserGuide.pdf 37320 233817     197571

Please also fix other issues that may be apparent in checks with a current R-devel.

Now, how can this be done? Here’s my workflow on Mac OS (might be slightly different on Win or Linux):

1.   Install R-develparallel to your existing (stable) R version
   1.  Before installation, run sudo pkgutil --forget org.r-project.R.Leopard.fw.pkg in the Terminal, otherwise the installer will overwrite your existing version
   2. Rename your R.app and R64.app or move them temporarily into another folder, as the installer of R-devel probably will replace them by new version that are not compatible with your existing stable R version.
2.   Use RSwitch to change the active R version
3.   Install packages which your own packages depends on; you have to do it from source, as the binaries for the R-devel do not exist: install.packages("lme4", type="source")
4.  Check your own package using following flag: R CMD check pkg --as-cran
5. Check if your package also works on Windows using winbuilder

Furthermore, check whether your package follows the CRAN Repository Policies.

PS: Finally, I managed to get rid of the annoying R CMD check warnings like “no visible binding for global variable ‘x’”. These occured due to the ggplot2 syntax. Here‘s a solution.

As a follow-up on my R implementation of Solomon’s watercolor plots, I made some improvements to the function. I fine-tuned the graphical parameters (the median smoother line now diminishes faster with increasing CIs, and the shaded watercolors look more pretty). Furthermore, the function is faster and has more features:

• You can define any standard regression function for the bootstrap procedure.
  • vwReg(y ~ x, df, method=lm)
  • vwReg(y ~ x + I(x^2), df, method=lm)
• Provide parameters for the fitting function.
  • You can make the smoother’s span larger. Then it takes more points into account when doing the local fitting. Per default, the smoother fits a polynomial of degree two – that means as you increase span you will approach the overall quadratic fit: vwReg(y ~ x, df, span=2)
  • You can also make the smoother’s span smaller, then it takes less points for local fitting. If it is too small, it will overfit and approach each single data point. The default span (.75) seemed to be the best choice for me for a variety of data sets: vwReg(y ~ x, df, span=0.5)
  • Use a robust M-estimator for the smoother; see ?loess for details: vwReg(y ~ x, df, family=”symmetric”)
• Provide your own color scheme (or, for example, a black-and-white scheme). Examples see pictures below.
• Quantize the color ramp, so that regions for 1, 2, and 3 SD have the same color (an idea proposed by John Mashey).

At the end of this post is the source code for the R function.

Some picture – please vote!

Here are some variants of the watercolor plots – at the end, you can vote for your favorite (or write something into the comments). I am still fine-tuning the default parameters, and I am interested in your opinions what would be the best default.

Plot 1: The current default

Plot 2: Using an M-estimator for bootstrap smoothers. Usually you get wider confidence intervalls.

Plot 3:Increasing the span of the smoothers

Plot 4: Decreasing the span of the smoothers

Plot 5: Changing the color scheme, using a predefined ColorBrewer palette. You can see all available palettes by using this command: library(RColorBrewer); display.brewer.all()

Plot 6: Using a custom-made palette

Plot 7: Using a custom-made palette; with the parameter bias you can shift the color ramp to the “higher” colors:

Plot 8: A black and white version of the plot

Plot 9: The anti-Tufte-plot: Using as much ink as possible by reversing black and white (a.k.a. “the Milky-Way Plot“)

Plot 10: The Northern Light Plot/ fMRI plot. This plotting technique already has been used by a suspicious company (called IRET – never heard of that). I hurried to publish the R code under a FreeBSD license before they can patent it! Feel free to use, share, or change the code for whatever purpose you need. Isn’t that beautiful?

Plot 11: The 1-2-3-SD plot. You can use your own color schemes as well, e.g.: vwReg(y~x, df, bw=TRUE, quantize=”SD”)

Any comments or ideas? Or just a vote? If you produce some nice plots with your data, you can send it to me, and I will post a gallery of the most impressive “data art”!

Cheers,

Felix

# Copyright 2012 Felix Schönbrodt
# All rights reserved.
# 
# FreeBSD License
# 
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met:
# 
# 1. Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
#       
# 2. Redistributions in binary form must reproduce the above copyright
# notice, this list of conditions and the following disclaimer in the
# documentation and/or other materials provided with the distribution.
#       
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER `AS IS'' AND ANY
# EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
# PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
# LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
# NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
# SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
# 
# The views and conclusions contained in the software and documentation
# are those of the authors and should not be interpreted as representing
# official policies, either expressed or implied, of the copyright
# holder.
 
# Version history:
# 0.1: original code
# 0.1.1: changed license to FreeBSD; re-established compability to ggplot2 (new version 0.9.2)
 
## Visually weighted regression / Watercolor plots
## Idea: Solomon Hsiang, with additional ideas from many blog commenters
 
 
# B = number bootstrapped smoothers
# shade: plot the shaded confidence region?
# shade.alpha: should the CI shading fade out at the edges? (by reducing alpha; 0 = no alpha decrease, 0.1 = medium alpha decrease, 0.5 = strong alpha decrease)
# spag: plot spaghetti lines?
# spag.color: color of spaghetti lines
# mweight: should the median smoother be visually weighted?
# show.lm: should the linear regresison line be plotted?
# show.CI: should the 95% CI limits be plotted?
# show.median: should the median smoother be plotted?
# median.col: color of the median smoother
# shape: shape of points
# method: the fitting function for the spaghettis; default: loess
# bw = TRUE: define a default b&w-palette
# slices: number of slices in x and y direction for the shaded region. Higher numbers make a smoother plot, but takes longer to draw. I wouldn'T go beyond 500
# palette: provide a custom color palette for the watercolors
# ylim: restrict range of the watercoloring
# quantize: either "continuous", or "SD". In the latter case, we get three color regions for 1, 2, and 3 SD (an idea of John Mashey)
# add: if add == FALSE, a new ggplot is returned. If add == TRUE, only the elements are returned, which can be added to an existing ggplot (with the '+' operator)
# ...: further parameters passed to the fitting function, in the case of loess, for example, "span = .9", or "family = 'symmetric'"
vwReg <- function(formula, data, title="", B=1000, shade=TRUE, shade.alpha=.1, spag=FALSE, spag.color="darkblue", mweight=TRUE, show.lm=FALSE, show.median = TRUE, median.col = "white", shape = 21, show.CI=FALSE, method=loess, bw=FALSE, slices=200, palette=colorRampPalette(c("#FFEDA0", "#DD0000"), bias=2)(20), ylim=NULL, quantize = "continuous",  add=FALSE, ...) {
	IV <- all.vars(formula)[2]
	DV <- all.vars(formula)[1]
	data <- na.omit(data[order(data[, IV]), c(IV, DV)])
 
	if (bw == TRUE) {
		palette <- colorRampPalette(c("#EEEEEE", "#999999", "#333333"), bias=2)(20)
	}
 
	print("Computing boostrapped smoothers ...")
	newx <- data.frame(seq(min(data[, IV]), max(data[, IV]), length=slices))
	colnames(newx) <- IV
	l0.boot <- matrix(NA, nrow=nrow(newx), ncol=B)
 
	l0 <- method(formula, data)
	for (i in 1:B) {
		data2 <- data[sample(nrow(data), replace=TRUE), ]
		data2 <- data2[order(data2[, IV]), ]
		if (class(l0)=="loess") {
			m1 <- method(formula, data2, control = loess.control(surface = "i", statistics="a", trace.hat="a"), ...)
		} else {
			m1 <- method(formula, data2, ...)
		}
		l0.boot[, i] <- predict(m1, newdata=newx)
	}
 
	# compute median and CI limits of bootstrap
	library(plyr)
	library(reshape2)
	CI.boot <- adply(l0.boot, 1, function(x) quantile(x, prob=c(.025, .5, .975, pnorm(c(-3, -2, -1, 0, 1, 2, 3))), na.rm=TRUE))[, -1]
	colnames(CI.boot)[1:10] <- c("LL", "M", "UL", paste0("SD", 1:7))
	CI.boot$x <- newx[, 1]
	CI.boot$width <- CI.boot$UL - CI.boot$LL
 
	# scale the CI width to the range 0 to 1 and flip it (bigger numbers = narrower CI)
	CI.boot$w2 <- (CI.boot$width - min(CI.boot$width))
	CI.boot$w3 <- 1-(CI.boot$w2/max(CI.boot$w2))
 
 
	# convert bootstrapped spaghettis to long format
	b2 <- melt(l0.boot)
	b2$x <- newx[,1]
	colnames(b2) <- c("index", "B", "value", "x")
 
	library(ggplot2)
	library(RColorBrewer)
 
	# Construct ggplot
	# All plot elements are constructed as a list, so they can be added to an existing ggplot
 
	# if add == FALSE: provide the basic ggplot object
	p0 <- ggplot(data, aes_string(x=IV, y=DV)) + theme_bw()
 
	# initialize elements with NULL (if they are defined, they are overwritten with something meaningful)
	gg.tiles <- gg.poly <- gg.spag <- gg.median <- gg.CI1 <- gg.CI2 <- gg.lm <- gg.points <- gg.title <- NULL
 
	if (shade == TRUE) {
		quantize <- match.arg(quantize, c("continuous", "SD"))
		if (quantize == "continuous") {
			print("Computing density estimates for each vertical cut ...")
			flush.console()
 
			if (is.null(ylim)) {
				min_value <- min(min(l0.boot, na.rm=TRUE), min(data[, DV], na.rm=TRUE))
				max_value <- max(max(l0.boot, na.rm=TRUE), max(data[, DV], na.rm=TRUE))
				ylim <- c(min_value, max_value)
			}
 
			# vertical cross-sectional density estimate
			d2 <- ddply(b2[, c("x", "value")], .(x), function(df) {
				res <- data.frame(density(df$value, na.rm=TRUE, n=slices, from=ylim[1], to=ylim[2])[c("x", "y")])
				#res <- data.frame(density(df$value, na.rm=TRUE, n=slices)[c("x", "y")])
				colnames(res) <- c("y", "dens")
				return(res)
			}, .progress="text")
 
			maxdens <- max(d2$dens)
			mindens <- min(d2$dens)
			d2$dens.scaled <- (d2$dens - mindens)/maxdens	
 
			## Tile approach
			d2$alpha.factor <- d2$dens.scaled^shade.alpha
			gg.tiles <-  list(geom_tile(data=d2, aes(x=x, y=y, fill=dens.scaled, alpha=alpha.factor)), scale_fill_gradientn("dens.scaled", colours=palette), scale_alpha_continuous(range=c(0.001, 1)))
		}
		if (quantize == "SD") {
			## Polygon approach
 
			SDs <- melt(CI.boot[, c("x", paste0("SD", 1:7))], id.vars="x")
			count <- 0
			d3 <- data.frame()
			col <- c(1,2,3,3,2,1)
			for (i in 1:6) {
				seg1 <- SDs[SDs$variable == paste0("SD", i), ]
				seg2 <- SDs[SDs$variable == paste0("SD", i+1), ]
				seg <- rbind(seg1, seg2[nrow(seg2):1, ])
				seg$group <- count
				seg$col <- col[i]
				count <- count + 1
				d3 <- rbind(d3, seg)
			}
 
			gg.poly <-  list(geom_polygon(data=d3, aes(x=x, y=value, color=NULL, fill=col, group=group)), scale_fill_gradientn("dens.scaled", colours=palette, values=seq(-1, 3, 1)))
		}
	}
 
	print("Build ggplot figure ...")
	flush.console()
 
 
	if (spag==TRUE) {
		gg.spag <-  geom_path(data=b2, aes(x=x, y=value, group=B), size=0.7, alpha=10/B, color=spag.color)
	}
 
	if (show.median == TRUE) {
		if (mweight == TRUE) {
			gg.median <-  geom_path(data=CI.boot, aes(x=x, y=M, alpha=w3^3), size=.6, linejoin="mitre", color=median.col)
		} else {
			gg.median <-  geom_path(data=CI.boot, aes(x=x, y=M), size = 0.6, linejoin="mitre", color=median.col)
		}
	}
 
	# Confidence limits
	if (show.CI == TRUE) {
		gg.CI1 <- geom_path(data=CI.boot, aes(x=x, y=UL), size=1, color="red")
		gg.CI2 <- geom_path(data=CI.boot, aes(x=x, y=LL), size=1, color="red")
	}
 
	# plain linear regression line
	if (show.lm==TRUE) {gg.lm <- geom_smooth(method="lm", color="darkgreen", se=FALSE)}
 
	gg.points <- geom_point(data=data, aes_string(x=IV, y=DV), size=1, shape=shape, fill="white", color="black")		
 
	if (title != "") {
		gg.title <- theme(title=title)
	}
 
 
	gg.elements <- list(gg.tiles, gg.poly, gg.spag, gg.median, gg.CI1, gg.CI2, gg.lm, gg.points, gg.title, theme(legend.position="none"))
 
	if (add == FALSE) {
		return(p0 + gg.elements)
	} else {
		return(gg.elements)
	}
}

it is a well-known scientific truth that research results which are accompanied by a fancy, colorful fMRI scan, are perceived as more believable and more persuasive than simple bar graphs or text results (McCabe & Castel, 2007; Weisberg, Keil, Goodstein, Rawson, & Gray, 2008). Readers even agree more with fictitious and unsubstantiated claims, as long as you provide a colorful brain image, and it works even when the subject is a dead salmon.

The power of brain images for everybody

What are the consequence of these troubling findings? The answer is clear. Everybody should be equipped with these powerful tools of research communication! We at IRET made it to our mission to provide the latest, cutting-edge tools for your research analysis. In this case we adopted a new technology called “visually weighted regression” or “watercolor plots” (see here, here, or here), and simply applied a new color scheme.

But now, let’s get some hands on it!

The example

Imagine you invested a lot of effort in collecting the data of 41 participants. Now you find following pattern in 2 of your 87 variables:

You could show that plain scatterplot. But should you do it? Nay. Of course everybody would spot the outliers on the top right. But which is much more important: it is b-o-r-i-n-g!

What is the alternative? Reporting the correlation as text? “We found a correlation of r = .38 (p = .014)”. Yawn.

Or maybe: “We chose to use a correlation technique that is robust against outliers and violations of normality, the Spearman rank coefficient. It turned out that the correlation broke down and was not significant any more (r = .06, p = .708).”.

Don’t be silly! With that style of scientific reporting, there would be nothing to write home about. But you can be sure: we have the right tools for you. Finally, the power of pictures is not limited to brain research – now you can turn any data into a magical fMRI plot like that:

Isn’t that beautiful? We recommend to accompany the figure with an elaborated description: “For local fitting, we used spline smoothers from 10`000 bootstrap replications. For a robust estimation of vertical confidence densities, a re-descending M-estimator with Tukey’s biweight function was employed. As one can clearly see in the plot, there is  significant confidence in the prediction of the x=0, y=0 region, as well as a minor hot spot in the x=15, y=60 region (also known as the supra-dextral data region).”

Magical Data Enhancer Tool

With the Magical Data Enhancer Tool (MDET) you can …

• … turn boring, marginally significant, or just crappy results into a stunning research experience
• … publish in scientific journal with higher impact factors
• … receive the media coverage that you and your research deserve
• … achieve higher acceptance rates from funding agencies
• … impress young women at the bar (you wouldn’t show a plain scatterplot, dude?!)

FAQ

Q: But – isn’t that approach unethical?
A: No, it’s not at all. In contrast, we at IRES think that it is unethical that only some researchers are allowed to exploit the cognitive biases of their readers. We design our products with a great respect for humanity and we believe that every researcher who can afford our products should have the same powerful tools at hand.

Q: How much does you product cost?
A: The standard version of the Magical Data Enhancer ships for 12’998 $. We are aware that this is a significant investment. But, come on: You deserve it! Furthermore, we will soon publish a free trial version, including the full R code on this blog. So stay tuned!

Best regards,

Lexis “Lex” Brycenet (CEO & CTO Research Communication)
International Research Enhancement Technology (IRET)