The Web page, Recipe_Wiz.htm, contains a JavaScript program that allows the user to create Web pages containing a form to collect data in the Recipe design. It is similar to FactorWiz (Birnbaum, 2000). All of the open-source code is contained in this same page, so it can easily be modified or expanded for specialized purposes.
To use this program for the first time, note that default values have already been introduced in the fields for an example study of decisions whether or not to try a new vaccine to prevent COVID-19. The three variables in the default study are A = Price, B = Risk (of harmful side-effects), and C = Effectiveness of the vaccine.
You can select the text in the "Experiment Name" box and type in a name that will be displayed for the participant; this name can contain spaces. For the example study, this title might be, "Will you take the new COVID-19 vaccine?" Then use your computer's tab key to tab to the "short name" field, and type in a name such as you might use for the file name of the study; this name should contain no spaces and it will be saved as the first variable in the data file. For the example, it might be "vaccine_01".
Pushing the computer's tab key again will select the name of the name of Factor A. This name will appear in the created Web form and it can contain HTML. The default given is, PRICE: ; the tags are HTML for bold text. Pressing tab each time, or clicking and selecting the fields, you can type in the information for Factors B and C.
By pushing the tab key or by selecting the text in a field in the table, you can enter the levels of the factors. The program allows you to enter material that is to appear before each trial ("Preliminary Material"), after each trial ("After Material"), and as separators between the factors. This material can be HTML, including links to images or other Web-supported media. The default values are the HTML for line breaks.
Recipe_Wiz.htm is designed to provide a response scale consisting of a row of radio buttons; you can specify the number of buttons and the labels to appear at the endpoints of the scale. In the default for the example, there are 11 buttons labeled from "very very unlikely" to "very very likely" (to try the vaccine). If the participant clicks the button at the left, the coded response will be "1", and successive buttons are coded with increasing integers.
When all the fields in the table have been filled in, you can press the button labeled "Make the form" and the HTML will appear in the box. Pressing the button "Display Form" will open a new window and display the questionnaire that has been created. It will contain a four-trial warm-up and the trials for the study (in random order), as well as a list of standard demographic questions such as age, gender, education, etc. Pressing "Make the form" again will create the same form again, except the experimental trials will appear in a different random order.
Pressing the "Save" button displays a pop-up message reminding that the text in the box should be saved in a text editor (not a word processor), and it selects all the text. One can then use CTRL-C and CTRL-V to copy and paste this text into a text editor, where it can be saved and edited. The file name should have no spaces in the name and should have an extension of .htm or .html. For example, the filename might be "vaccine_01.htm", corresponding to the short name.
In the HTML appears "(put your instructions here)", where you can paste the instructions. You should also edit and add to the the warm-up trials to include the worst and best cases; I suggest increasing the number of warmups to 7 or more.
Once instructions have been added, the HTML file could then be uploaded to the WWW, where participants can be directed to complete the questionnaire.
When the participant presses the "submit" button, the data will be sent to a PERL script and saved to a text file at the following URL:\\ \url{http://ati-birnbaum.netfirms.com/data/data.txt}
One need only change the ACTION of the FORM tag to direct the data to another server. Instructions for running your own server and for installing a generic PERL script to save data to your own server are given in Reips and Birnbaum (2011).\footnote{It is advised to save data from any real experiment in a secure location, rather than on the open Web, as was done here. This exception has been set up to allow people to test the questionnaires and system, rather than for real data collection.}
Even though the trials are in random order, the data are organized by the generic.pl script in the proper order for Recipe analysis. The experimental responses start in the 12th column. The first columns list the short experiment name, date, time, IP address, and other background information.
Starting in the 12th column the responses are ordered by designs; the designs are ordered: A, B, C, AB, AC, BC, and ABC. Within designs, they are ordered in factorial order with the last-listed factor index moving the fastest. For example, for a 3 by 4, AB design the order is $AB_{11}, AB_{12}, AB_{13}, AB_{14}, AB_{21}, \dots, AB_{n_An_B}$, where $n_A$ and $n_B$ are the number of levels of A and B, respectively. The programs, Recipe_sim.htm and Recipe_fit.xlsx will help you understand how the data are organized and coded.