# R Dataset / Package HistData / MacdonellDF

Webform
Category

Webform
Category

Webform
Category

Webform
Category

Webform
Category

Webform
Category

## Visual Summaries

Embed
<iframe src="https://embed.picostat.com/r-dataset-package-histdata-macdonelldf.html" frameBorder="0" width="100%" height="307px" />
Attachment Size
55.94 KB
Documentation

## Macdonell's Data on Height and Finger Length of Criminals, used by Gosset (1908)

### Description

In the second issue of Biometrika, W. R. Macdonell (1902) published an extensive paper, On Criminal Anthropometry and the Identification of Criminals in which he included numerous tables of physical characteristics 3000 non-habitual male criminals serving their sentences in England and Wales. His Table III (p. 216) recorded a bivariate frequency distribution of height by finger length. His main purpose was to show that Scotland Yard could have indexed their material more efficiently, and find a given profile more quickly.

W. S. Gosset (aka "Student") used these data in two classic papers in 1908, in which he derived various characteristics of the sampling distributions of the mean, standard deviation and Pearson's r. He said, "Before I had succeeded in solving my problem analytically, I had endeavoured to do so empirically." Among his experiments, he randomly shuffled the 3000 observations from Macdonell's table, and then grouped them into samples of size 4, 8, ..., calculating the sample means, standard deviations and correlations for each sample.

### Usage

data(Macdonell)
data(MacdonellDF)


### Format

Macdonell: A frequency data frame with 924 observations on the following 3 variables giving the bivariate frequency distribution of height and finger.

height

lower class boundaries of height, in decimal ft.

finger

length of the left middle finger, in mm.

frequency

frequency of this combination of height and finger

MacdonellDF: A data frame with 3000 observations on the following 2 variables.

height

a numeric vector

finger

a numeric vector

### Details

Class intervals for height in Macdonell's table were given in 1 in. ranges, from (4' 7" 9/16 - 4' 8" 9/16), to (6' 4" 9/16 - 6' 5" 9/16). The values of height are taken as the lower class boundaries.

For convenience, the data frame MacdonellDF presents the same data, in expanded form, with each combination of height and finger replicated frequency times.

### Source

Macdonell, W. R. (1902). On Criminal Anthropometry and the Identification of Criminals. Biometrika, 1(2), 177-227. doi:10.1093/biomet/1.2.177 http://www.jstor.org/stable/2331487

The data used here were obtained from:

Hanley, J. (2008). Macdonell data used by Student. http://www.medicine.mcgill.ca/epidemiology/hanley/Student/

### References

Hanley, J. and Julien, M. and Moodie, E. (2008). Student's z, t, and s: What if Gosset had R? The American Statistican, 62(1), 64-69.

Gosett, W. S. [Student] (1908). Probable error of a mean. Biometrika, 6(1), 1-25. http://www.york.ac.uk/depts/maths/histstat/student.pdf

Gosett, W. S. [Student] (1908). Probable error of a correlation coefficient. Biometrika, 6, 302-310.

### Examples

data(Macdonell)# display the frequency table
xtabs(frequency ~ finger+round(height,3), data=Macdonell)## Some examples by james.hanley@mcgill.ca    October 16, 2011
## http://www.biostat.mcgill.ca/hanley/
## See:  http://www.biostat.mcgill.ca/hanley/Student/###############################################
##  naive contour plots of height and finger ##
###############################################

# make a 22 x 42 table
attach(Macdonell)
ht <- unique(height)
fi <- unique(finger)
fr <- t(matrix(frequency, nrow=42))
detach(Macdonell)
dev.new(width=10, height=5)  # make plot double wide
op <- par(mfrow=c(1,2),mar=c(0.5,0.5,0.5,0.5),oma=c(2,2,0,0))dx <- 0.5/12
dy <- 0.5/12plot(ht,ht,xlim=c(min(ht)-dx,max(ht)+dx),
ylim=c(min(fi)-dy,max(fi)+dy), xlab="", ylab="", type="n" )# unpack  3000 heights while looping though the frequencies
heights <- c()
for(i in 1:22) {
for (j in 1:42) {
f  <-  fr[i,j]
if(f>0) heights <- c(heights,rep(ht[i],f))
if(f>0) text(ht[i], fi[j], toString(f), cex=0.4, col="grey40" )
}
}
text(4.65,13.5, "Finger length (cm)",adj=c(0,1), col="black") ;
text(5.75,9.5, "Height (feet)", adj=c(0,1), col="black") ;
text(6.1,11, "Observed bin\nfrequencies", adj=c(0.5,1), col="grey40",cex=0.85) ;
# crude countour plot
contour(ht, fi, fr, add=TRUE, drawlabels=FALSE, col="grey60")
# smoother contour plot (Galton smoothed 2-D frequencies this way)
# [Galton had experience with plotting isobars for meteorological data]
# it was the smoothed plot that made him remember his 'conic sections'
# and ask a mathematician to work out for him the iso-density
# contours of a bivariate Gaussian distribution... dx <- 0.5/12; dy <- 0.05  ; # shifts caused by averagingplot(ht,ht,xlim=c(min(ht),max(ht)),ylim=c(min(fi),max(fi)), xlab="", ylab="", type="n"  )

sm.fr <- matrix(rep(0,21*41),nrow <- 21)
for(i in 1:21) {
for (j in 1:41) {
smooth.freq  <-  (1/4) * sum( fr[i:(i+1), j:(j+1)] )
sm.fr[i,j]  <-  smooth.freq
if(smooth.freq > 0 )
text(ht[i]+dx, fi[j]+dy, sub("^0.", ".",toString(smooth.freq)), cex=0.4, col="grey40" )
}
}

contour(ht[1:21]+dx, fi[1:41]+dy, sm.fr, add=TRUE, drawlabels=FALSE, col="grey60")
text(6.05,11, "Smoothed bin\nfrequencies", adj=c(0.5,1), col="grey40", cex=0.85) ;
par(op)
dev.new()    # new default device#######################################
## bivariate kernel density estimate
#######################################if(require(KernSmooth)) {
MDest <- bkde2D(MacdonellDF, bandwidth=c(1/8, 1/8))
contour(x=MDest$x1, y=MDest$x2, z=MDest\$fhat,
xlab="Height (feet)", ylab="Finger length (cm)", col="red", lwd=2)
with(MacdonellDF, points(jitter(height), jitter(finger), cex=0.5))
}#############################################################
## sunflower plot of height and finger with data ellipses  ##
#############################################################with(MacdonellDF,
{
sunflowerplot(height, finger, size=1/12, seg.col="green3",
xlab="Height (feet)", ylab="Finger length (cm)")
reg <- lm(finger ~ height)
abline(reg, lwd=2)
if(require(car)) {
dataEllipse(height, finger, plot.points=FALSE, levels=c(.40, .68, .95))
}
})
############################################################
## Sampling distributions of sample sd (s) and z=(ybar-mu)/s
############################################################# note that Gosset used a divisor of n (not n-1) to get the sd.
# He also used Sheppard's correction for the 'binning' or grouping.
# with concatenated height measurements...mu <- mean(heights) ; sigma <- sqrt( 3000 * var(heights)/2999 )
c(mu,sigma)# 750 samples of size n=4 (as Gosset did)# see Student's z, t, and s: What if Gosset had R?
# [Hanley J, Julien M, and Moodie E. The American Statistician, February 2008] # see also the photographs from Student's notebook ('Original small sample data and notes")
# under the link "Gosset' 750 samples of size n=4"
# on website http://www.biostat.mcgill.ca/hanley/Student/
# and while there, look at the cover of the Notebook containing his yeast-cell counts
# http://www.medicine.mcgill.ca/epidemiology/hanley/Student/750samplesOf4/Covers.JPG
# (Biometrika 1907) and decide for yourself why Gosset, when forced to write under a
# pen-name, might have taken the name he did!# PS: Can you figure out what the 750 pairs of numbers signify?
# hint: look again at the numbers of rows and columns in Macdonell's (frequency) Table III.
n <- 4
Nsamples <- 750y.bar.values <- s.over.sigma.values <- z.values <- c()
for (samp in 1:Nsamples) {
y <- sample(heights,n)
y.bar <- mean(y)
s  <-  sqrt( (n/(n-1))*var(y) )
z <- (y.bar-mu)/s
y.bar.values <- c(y.bar.values,y.bar)
s.over.sigma.values <- c(s.over.sigma.values,s/sigma)
z.values <- c(z.values,z)
}
op <- par(mfrow=c(2,2),mar=c(2.5,2.5,2.5,2.5),oma=c(2,2,0,0))
# sampling distributions
hist(heights,breaks=seq(4.5,6.5,1/12), main="Histogram of heights (N=3000)")
hist(y.bar.values, main=paste("Histogram of y.bar (n=",n,")",sep=""))hist(s.over.sigma.values,breaks=seq(0,4,0.1),
main=paste("Histogram of s/sigma (n=",n,")",sep=""));
z=seq(-5,5,0.25)+0.125
hist(z.values,breaks=z-0.125, main="Histogram of z=(ybar-mu)/s")
# theoretical
lines(z, 750*0.25*sqrt(n-1)*dt(sqrt(n-1)*z,3), col="red", lwd=1)
par(op)#####################################################
## Chisquare probability plot for bivariate normality
#####################################################mu <- colMeans(MacdonellDF)
sigma <- var(MacdonellDF)
Dsq <- mahalanobis(MacdonellDF, mu, sigma)Q <- qchisq(1:3000/3000, 2)
plot(Q, sort(Dsq), xlab="Chisquare (2) quantile", ylab="Squared distance")
abline(a=0, b=1, col="red", lwd=2)
--

Dataset imported from https://www.r-project.org.

Picostat Manual
###### How To Register With a Username
1. Go to the user registration page.
4. Click Submit.
5. Click the link that was sent to the email address you registered with.
6. Clicking the link will open another page on Picostat where you can select a password.
7. Click Save and enter any profile details you wish to enter.
###### How To Register With Google Single Sign On (SSO)
1. Go to the user login page.
5. Google will redirect you back to Picostat with your new account created and you will be logged in.
6. Enter any profile details you wish to share.
1. Go to the user login page.
3. Click "Login". You will be redirected to your user homepage authenticated.
1. Go to the user login page.
3. If you already registered with Picostat via Google SSO, you will be redirected to your user homepage authenticated.
###### How To Import a Dataset
1. Create a Picostat account or login with your existing picostat account (see above).
2. Go to the dataset import page.
3. Select a license for the dataset. The default is "No License" but allows Picostat to host a copy of the dataset as per the privacy policy. You may wish to uncheck the "Public" option if you do not wish to share your dataset with others. R Datasets that come by downloading R have a GNU General Public License v3.0 which may also be selected from the Picostat dropdown.
4. Enter a title for the dataset
5. Choose a dataset input methods. Available options include:
• Random data - this populates your dataset with random numbers between 0 and 100. You can specify the number of rows and columns for the random dataset.
• CSV, TSV or TXT file - you will have the option upload a file within the current file size limit and also specify the header and whether or not the dataset is a contingency table. With contingency tables, the first column becomes a label for the rows. Currently with Picostat, there is limited support for contingency tables. Choose "Yes" to the Header option if the first line of the data contains titles for the rows. Also choose the Separator for the dataset. A separator is what breaks the data up. In some cases, a comma would separate data values in a row. You will also have the option to add documentation in the form keyboarded text and also uploaded documentation attachments. You can also specify a license for the documentation.
• Copy and Paste. This selection contains many of the same fields as importing a file with an additional textarea to copy and paste data to.
• Empty dataset. Start with a blank dataset and manually add data with the Picostat dataset editor.
• Excel file - Choose this option if you would like to convert your Excel spreadsheet to a Picostat dataset. With this selection, you will have the option to specify whether to use the first row in the Excel file as column names. If you would like to choose a specific sheet to use, you can also specify with entering its name in the text input.
• sas7bdat file - SAS is a powerful statistical software package that has its own proprietary file format. Choose this option if you are importing a SAS file.
• SPSS sav file - SPSS is a statistical package owned by IBM. You can import SPSS files by choosing this option.
6. Choose whether or not the dataset contains a header. Some of the dataset input methods allow you to specify whether or not a Header exists on the file. Sometimes dataset files contain a Header as the first row which names the columns. If you choose "Yes" to this, the first row in the dataset will become column headers.
7. You can also add documentation and specify a documentation license. This can be used to help explain your dataset to those unfamiliar with it.
8. Choose whether or not to upload an supporting attachments.
9. Pass the captcha. To prevent spam submissions, Picostat has a captcha which is used to prevent automated submissions by bots.
10. Choose a privacy setting for the dataset. You can also specify whether or not the dataset is Public. If you uncheck this setting, only you and the Picostat administrator will be able to view the dataset.
11. Submit the form. Once the form is validated, you will be redirected to the dataset homepage where you can choose to edit or perform statistical operations on the dataset.
###### How To Perform Statistical Analysis with Picostat
1. Go to any dataset homepage. You can get a full list at the dashboard.
2. Near the top of the page there will be two drop downs. One for analysis and one for education. Here we will choose Analyis. Choose from one of the following:
• Numerical Summaries - Here you can get the:
1. Arithmetic mean
2. Median
3. Quartiles
4. Minimum and Maximum
5. Stem-and-leaf plot
6. Standard deviation and Variance
7. IQR
8. Cumulative frequencies
• Plot - a plot of two columns on the cartesian coordinate system
• Boxplot - a Boxplot (box-and-whisker plot) of a column.
• Correlation Coefficient - Compute the correlation coefficient between two columns.
• Cumulative Frequency Histogram - Display a cumulative frequency histogram
• Dotplot
• Hollow Histogram - Plot two columns on the same histogram with a different color for each column.
• Pie Chart
• Regression - Perform a simple linear regression and compute the p-value and regression line. Also plots the data with the regression line.
• Stem and Leaf Plots - Plot a one or two-sided stem-and-leaf plot from one or two columns respectively.
• Visual Summaries - plots the following:
1. Frequency Histogram
2. Relative Frequency Histogram
3. Cumulative Frequency Histogram
4. Boxplot (Box-and-whisker plot)
5. Dotplot
3. PDF - Check this box if you want to download a PDF of the output from the statistical analysis application.
4. HTML - Check this box if you want a link to the HTML page created with the application.
5. R File - Check this box if you want a link to download the R commands used to generate output from this statistical application. You should be able to copy the commands verbatim into R Studio or R to recreate the analysis.
###### How To Use Educational Applications with Picostat
1. Go to any dataset homepage. You can get a full list at the dashboard.
2. Near the top of the page there will be two drop downs. One for analysis and one for education. Here we will choose Education. Choose from one of the following:
• How To Create a Barplot - This will show you how to create a bar chart after selecting a column with the mouse.
• How To Create a Stacked Barplot - This application will show you how to create a stacked bar plot from a column vector.
• How To Create a Pie Chart - This application will show you how to create a pie chart from a column of data
• How To Compute the Mean - This application will show you how to compute the mean from a column vector
• How To Create a Plot - This app will show you how to plot two columns in the cartesian coordinate system.
• How To Compute the Media - This statisistical app will show you how to compute the median from a column vector.
3. PDF - Check this box if you want to download a PDF of the output from the education application.
4. HTML - Check this box if you want a link to the HTML page created with the application.
Recent Queries For This Dataset

No queries made on this dataset yet.

Title Authored on Content type
OpenIntro Statistics Dataset - sat_improve August 9, 2020 - 2:38 PM Dataset
R Dataset / Package Stat2Data / CAFE March 9, 2018 - 1:06 PM Dataset
ChickWeight February 26, 2017 - 11:28 AM Dataset
R Dataset / Package boot / downs.bc March 9, 2018 - 1:06 PM Dataset
R Dataset / Package wooldridge / beauty March 9, 2018 - 1:06 PM Dataset