Supplementary Exercise 3.14 of IPS7e
------------------------------------

Study to assess any impact of offering an on-site childcare facility as
a job benefit. The ability to attract applicants to two similar fictitious 
companies is compared. The brochure for one company (B) is printed in two versions,
one of which mentions a childcare facility whereas the other one does not. 
A total of 40 unmarried women seeking employment are available to participate
in the study. They will receive brochures for both companies, but the treatment
group determines which version of the brochure for company B they will
receive. The treatment groups will be of equal size.


(a) The design is a completely randomized design with 2 treatment groups, both 
comprising 20 women. The PSLS and IPS textbooks use a simple display for the design,
roughly as follows:

Random assignment  -> Group 1 (20 women)  -> Brochure with childcare
                   -> Group 2 (20 women)  -> Brochure without childcare

In my view, this display is not crucial for understanding the design.

The response measured for all women is which company their prefer; i.e., a 
dichotomous, or binary, response.


(b) From Table A (PSLS) or Table B (IPS), lines 121-124, we get the following list 
of two-digit numbers between 1 and 40:

29 07 34 22 10 25
13 38 15 05 (29) 09 08 27 (13)
(08) (15) (07) 27 (10) (25) (27) 23 30 28 18
03 01 36 ...

The numbers in parenthesis are repeats, but the other numbers form one
of the two groups in the trial. The numbers are allocated to the names
on the list by numbering these from 1 to 40, e.g. as in the data file.


Randomization using the Random numbers applet:
----------------------------------------------
Enter 40 as the maximum and 20 as the sample size, and click Sample. 
One trial with the applet produced a sample of:

34 20 32 37 10 11 8 19 28 25 21 2 38 29 40 6 18 22 9 30

- these numbers define the first treatment group.


Randomization using Minitab:
----------------------------
 WOpen "R:\Chapter 3\ex03_014.mtw".
 Base 140906.
 Name c2 "randorder"
 Sample 40 'name' 'randorder'.
 Print 'name' 'randorder'.

Data Display 

Row  name       randorder
  1  Abrams     Brown
  2  Adamson    Morse
  3  Afifi      Roberts
  4  Brown      Gerson
  5  Cansico    Kaplan
  6  Chen       Cansico
  7  Cortez     Green
  8  Curzakis   Afifi
  9  Danielson  Williams
 10  Durr       Rosen
 11  Edwards    Cortez
 12  Fluharty   McNeill
 13  Garcia     Garcia
 14  Gerson     Lattimore
 15  Green      Chen
 16  Gupta      Lippman
 17  Gutierrez  Janle
 18  Howard     Edwards
 19  Hwang      Quinones
 20  Iselin     Iselin
 21  Janle      Travers
 22  Kaplan     Gutierrez
 23  Kim        Turing
 24  Lattimore  Ng
 25  Lippman    Rivera
 26  Martinez   Kim
 27  McNeill    Wong
 28  Morse      Abrams
 29  Ng         Sugiwara
 30  Quinones   Gupta
 31  Rivera     Ullmann
 32  Roberts    Fluharty
 33  Rosen      Thompson
 34  Sugiwara   Curzakis
 35  Thompson   Martinez
 36  Travers    Danielson
 37  Turing     Adamson
 38  Ullmann    Durr
 39  Williams   Howard
 40  Wong       Hwang

The first 20 names of the second column (randorder) are used for the first
treatment group, and the last 20 names for the second treatment group.