Write a codebook to an html file
cb2html [varlist] using filename [, replace title(string) pprecision(#) fprecision(#) sprecision(#) maxvalues(#) summarize(#) stringmax(#) notes type pagebreak(#) combine(var1 var2) upper vallabel ]
cb2html writes a codebook in html format for the variables specified or, if no variables are specified, all the variables in the data in memory, into filename.html. It writes the dataset label, the variable name, label, and the data type of each variable. It writes the frequency, percent, value, and label of each value by default, or it writes the n, mean, std dev, min, and max. Optionally, it writes other information that has been stored in variable and dataset notes.
Survey research organizations that collect and disseminate survey data are most likely to find this command of use. The codebook it produces is intended to be an end-user product suitable for publishing or printing. It provides a method of combining the frequencies and/or means of the data with the full text of the questionnaire so that researchers have all information in a single document.
After creating an html file, you can edit it in Microsoft Word. This allows you to change fonts, insert text, etc., as you would with any MS Word document without editing the html code. You can then print from MS Word or convert the file to PDF or some other format. (Warning: MS Word inserts a lot of extra html code. This may cause problems if you use the codebook as an html document after editing it in MS Word.)
using filename must be specified. The suffix .html is added automatically.
replace writes over filename.html if it already exists. If filename.html already exists, and replace is not specified, the file will not be written.
title(string) allows you to specify a title that is displayed in the browser title bar at the top. If title is not specified, the dataset label is displayed instead (assuming it exists).
pprecision(#) allows you to change the number of decimal values displayed in the percents column. The default value is 0 (e.g., 12%). A value of 1 will preserve one decimal of precision (e.g., 12.3%), a value of 2 will preserve two decimals of precision (e.g., 12.34%), etc. Any positive integer may be given.
fprecision(#) allows you to change the number of decimal values displayed in the values column for floating point variables (stored in Stata as data type float or double). The default value is 0 (e.g., 12). A value of 1 will preserve one decimal of precision (e.g., 12.3), a value of 2 will preserve two decimals of precision (e.g., 12.34), etc. Any integer in the range 0 - 22 may be given.
sprecision(#) allows you to change the number of decimal values displayed for all statistics of summarized variables. The default value is 0 (e.g., 12). A value of 1 will preserve one decimal of precision (e.g., 12.3), a value of 2 will preserve two decimals of precision (e.g., 12.34), etc. Any integer in the range 0 - 22 may be given.
maxvalues(#) allows you to limit the number of discrete values that will be written for a variable. If a variable, e.g., each respondent's ID, has more than # discrete values, only the 5 smallest and 5 largest values will be written. A row will also be written that gives the range of values omitted from the display along with the frequency and percent of the omitted values.
The default value of # is 10. Any integer may be given, including numbers smaller than 10. However, at minimum, the 5 smallest and 5 largest values will always be printed, or all the values will be printed if the variable has fewer than 10 values.
You can override maxvalues for one or more variables by defining a variable characteristic. See Notes below.
summarize(#) allows you to select a lower limit above which the frequency, mean, sd, min, and max are printed instead of the discrete values. This option overrides maxvalues.
You can override summarize for one or more variables by defining a variable characteristic. See Notes below.
stringmax(#) In Stata/SE a string value can be up to 244 characters long. By default, the entire string value will be written to the codebook. You can shorten long strings to # characters using this option. Leading and trailing blanks will be removed, then the first # characters will be written to the html file.
notes lists note1, note2, and note3 attached to the variables and note1 attached to the dataset (see help notes).
By default, only the variable and dataset labels are written, and labels are limited to 80 characters in Stata/SE. Optionally, you may create notes for variables, e.g., containing the full text of the question, skip instructions, or supplemental instructions to interviewers. You may also create a note for the dataset, e.g., describing the survey or outlining the contents of a section.
Notes can be up to 67,784 characters; however, you may want to limit your note length for readability. Notes should not contain double quotes, but single quotes can be used. Notes may contain standard html code, such as <br> to start a new line.
You may define up to three notes for each variable and one note for the dataset. Stata numbers notes by the order that the note command appears in your do-file. The first note for a given variable becomes "note1", the second "note2", and the third "note3". These notes are handled differently as described below.
If note1 exists for a variable, it will be written in place of the variable label. If note1 exists for the dataset, it will be written after the dataset label. Note1 is written without bold or italics.
You may also create note2 for variables, e.g., containing the skip conditions for the variable. If note2 exists, it will be written in a separate row above note1. Note2 is written in bold and italics.
You may also create note3 for variables, e.g., containing instructions to interviewers. If note3 exists, it will be written in a separate row above note2. Note3 is written in bold.
type writes the Stata data storage type (byte, int, float, long, double, or str#). By default, whether the variable is numeric or character (string) is written.
pagebreak(#) Use this option if you intend to print the codebook and don't want the table for any variable split across pages. It inserts a page break into the html file after # lines (or fewer) have been written. First, it checks whether the entire table for the next variable will fit on the page within the line limit of #. If not, it inserts a page break, moving the next variable's table to the following page. The default is no page breaks, which minimizes paper use, but is likely to result in pages breaking in the middle of a table. A page break is automatically inserted after the data label and data note in order to treat that information as a separate title page.
The number of lines per page that you should choose depends on many factors, including your printer type, browser version, and the font size and type you have set in your browser. This command uses Times New Roman, which is a variable-spaced font. It does not specify a size for the font. (In Mozilla Firefox you can adjust the font size in Tools, Options, Content. In Internet Explorer, use View, Text Size.)
The number of lines also is affected by the lengths of your variable notes, variable labels, and value labels. This command assumes 100 characters per line for each variable note and 60 for each variable and value label, and it adjusts the estimated lines per page accordingly.
There's no guarantee that this option will prevent every table from splitting across pages. With a little experimenting, you should find a # that works for almost all variables and still keeps the number of printed pages reasonably low.
combine(var1 var2) allows you to combine two variables temporarily into a single variable and obtain frequencies of the combination. An example is year and month of interview.
upper will print variable names in upper case. The default is lower case.
vallabel will print the value in the space used for the value label if no label has been defined. The default is a blank space.
You can define a variable characteristic called freq for one or more variables with the value Yes. This tells cb2html to print all values of that variable regardless of the settings for maxvalues() or summarize(). You can define the characteristic like this:
char def var1[freq] Yes
This command requires Stata 10. It produces W3C valid HTML 4.01 Transitional code, which displays properly in Mozilla Firefox, and Internet Explorer.
. cb2html using mycodebook
. cb2html using mycodebook, title(Codebook for My Project)
. cb2html vj89-vm92a v334_* using mycodebook, replace
. cb2html using mycodebook, maxvalues(20) pprecision(2) fprecision(3)
. cb2html using mycodebook, summarize(20) sprecision(3)
. cb2html using "c:\My Project\mycodebook.html" pagebreak(54) upper
. cb2html using mycodebook, notes type stringmax(20) replace
. cb2html using mycodebook, combine(int_year int_month)
I incorporated many coding suggestions from Kit Baum, Nick Cox, and Dan Blanchette, I benefitted greatly from Ed Van Duinen's advice on web programming, and I added many options suggested by programmers too numerous to mention at The Carolina Population Center and elsewhere. Any problems you find in this program are mine alone.
Phil Bardsley, Carolina Population Center, University of North Carolina - Chapel Hill, USA. Contact firstname.lastname@example.org if you observe any problems.
Manual: [R] codebook, [R] notes, [R] label, [R] inspect, [R] labelbook, [R] describe