Boolean Values Best Practices Reference

Boolean Values Best Practices Reference

Title: Boolean Values Best Practices Reference

Date Modified: 2023-03-06

Version: 20230306.1

Part of TDWG Standard: Not part of any standard

Abstract: The following is a summary of best practices for the use of boolean values.

  • Use an enumerated domain containing two and only two values, true and false.

  • Domain values should be lowercase, non-localized, and not abbreviated

Contributors: Ben Norton (author, Head of Technology and Collections Data Curator, NCMNS, Contact:, Steve Baskauf (reviewer), Tim Robertson (reviewer), John Wieczorek (reviewer)

Creator: TDWG Technical Architecture Group (TAG)

Bibliographic citation: Technical Architecture Group. 2023. Boolean Values Best Practices Reference. Biodiversity Information Standards (TDWG).

1 Best Practices Recommendation

The recommendation of the TDWG Technical Architecture Group for providing boolean values is to use a value domain composed of the two lowercase and non-localized character strings, true and false. Alternate true and false versions such as the abbreviations T and F, uppercase strings TRUE and FALSE, or localized versions such as vrai and faux are not permitted. The [true, false] value domain is highly interoperable as the most widely used representation of boolean values. In addition, most major programming languages provide native support for the boolean data type with methods to cast values to the boolean data type readily. In contrast, support by major database vendors varies. However, those vendors without native support provide an alternate mechanism for handling the boolean data type.

1.1 Usage in Computing

In modern computing, the boolean data type is a primitive type (a type not derived from another) that represents the truth of an expression using two numerical values, 0 or 1. Handling boolean values, including the enumerated domain values and syntax, varies between programming languages and database technologies. Tables 1-3 in Appendix 1 summarize support for the boolean data type for many popular programming languages and database vendors. In contrast to programming languages, all database technologies do not provide native support for the boolean data type (e.g., MySQL, MSSQL, SQLite). In these cases, they offer an alternative mechanism or best practice, which is denoted accordingly in the tables.

To understand how boolean values are used in computing, we first must distinguish between the boolean data type and boolean expressions (operations and comparisons to evaluate truth). The latter uses a set of logical operators in a control structure (e.g., if/else, while loop, case, switch) to evaluate whether or not something is true. The value returned from a boolean expression is always a boolean value. The boolean data type refers to a primitive data type with values limited to a pair of valid values or enumerated domain (true,false, 0,1). Boolean data type variables represent whether or not something is true or false. The boolean data type is representative, whereas the boolean expression is evaluative. Support for booleans by the popular programming language Python provides a good example of this conceptual distinction. In Python, depending on purpose/context, boolean expressions utilize either the boolean operators and, or, not or comparison operators <, <=, >, >=, ==, !=, is, is not. Boolean values in Python are limited to a single enumerated domain that contains True and False.

1.2 Enumerated Domains

Throughout this document, pairs of valid values for the boolean data type are referred to as boolean enumerated domains. In statistics, informatics, and data science, enumerated domains are value domains comprised of a list of permissible values. For most technologies, values that belong to the boolean data type are restricted to the boolean enumerated domain. Values in boolean enumerated domains are not limited to specific character strings. Rather, they are restricted to one or more two-state values. Numeric boolean enumerated domains (or numeric boolean domains) are a special type that only contains the numbers 0 and 1. In Tables 1-3, boolean enumerated domains are denoted with square brackets.

1.3 Numerical Misinterpretation

When the boolean data type exists, 0 and 1 do not behave like traditional integers (where 1 + 1 = 2 ). Rather they are binary representations of a true/false concept. It is important to make clear the distinction so that columns with boolean data values are not misinterpreted as numerical integers with mathematical implications.

2 Recommendations

  1. When the boolean data type exists, 0 and 1 do not behave like traditional integers (where 1 + 1 = 2 ). Rather they are binary representations of a true/false concept. A clear distinction is crucial to mitigate the risk of misinterpreting boolean data values as integers with mathematical implications. Camel-cased terms expecting boolean values should be prefaced with is (e.g., isExtinct, isTypeSpecimen). Snake-cased terms should be appended with _yn (e.g., extinct_yn, type_specimen_yn).

  2. The boolean enumerated domain [true, false] is the most interoperable domain with the widest support by major programming languages and database technologies.

  3. The numeric boolean enumerated domain [0, 1] should be used for ETL (Extract, Transform, Load) processes with database technologies that lack native boolean support. Otherwise, the numeric domain should be avoided to mitigate risks associated with numerical misinterpretation.

  4. The values true and false should always be lowercase to minimize the interoperability issues with case-sensitive systems. The casing changes should be strictly need based in cases where capitalization is necessary for compatibility. For example, Python boolean values are proper-cased (True, False). Usage of the proper-cased versions of true and false should be strictly limited to the Python application codebase where it is needed for compatibility.

  5. Do not use abbreviations for boolean values (e.g., T, F, t). Abbreviations can be easily misinterpreted and are weak concept representations.

  6. Null is not a boolean value and therefore does not belong in a column with a boolean data type. A default value should be assigned to boolean data types to avoid null values entirely. If a null value is needed to express a property accurately, then the table and column decisions should be re-evaluated.

  7. Character string values that belong to a boolean enumerated domain should not be localized. For example, the French translations of true and false, vrai and faux, are not valid boolean enumerated domain values.

  8. To accommodate the not null restriction, columns defined using the boolean data type in database schemas should always include a default boolean enumerated domain value.

Appendix 1. Boolean Support and Compatibility

Table 1A. Databases

Database Domains Data Type Remarks
MySQL [1,0] TINYINT(1) MySQL does not provide native support for the boolean data type. Instead, boolean values as commonly stored as 0 or 1 in the single byte data type, TINYINT(1)
PostgreSQL [on,off], [true/false], [1,0] BOOL In PostgreSQL, the boolean enumerated domain consists of 3 pairs of two-state values (six values in total). Boolean values must be consistent. Mixing values from two or more pairs within a single column is not allowed.
MSSQL [1,0] BIT MSSQL does not provide native support for the boolean data type. The common practice is to store boolean values in a BIT data type column as 0 or 1.
MS Access [Yes,No] Yes/No MS Access does not directly support boolean data types. Instead, MS Access provides a Yes/No data type that may be formatted in three ways for front-end presentation (Yes/No, True/False, On/Off). In Access, Yes/No data type is equivalent to the SQL Data Type BIT with one minor adjustment. In Access, Yes (true) is stored as -1, and No (false) is stored as 0. If no value is provided, the Yes/No data type defaults to No.
Oracle [YES/NO, TRUE/FALSE, ON/OFF] BOOLEAN In Oracle, the boolean enumerated domain consists of three case-insensitive, two-state value pairs. Using values from two or more two-state pairs within a single table column is not allowed.
SQLite [0,1] INTEGER SQLite does not provide native support for the boolean data type. Instead, Boolean values are stored using the integer data type with values belonging to the numeric boolean enumerated domain (0,1).


Table 1B. Database References

Table 2A. Programming Languages

Language Values Remarks
Python [0/1, True/ False] In Python, boolean data type (bool) is a subclass of integer represented using proper casing. The alphabetic character strings act programmatically as numerical values 0 and 1. Boolean values are returned by a truth evaluation using standard boolean operators (and, or, not) or comparison operations (<, <=, >, >=, ==, !=, is, is not).
Java [true, false] In Java, the boolean data type is one of the eight primitive data types. Boolean values are used in two ways: 1) A data type assigned to variables where the values must belong to the enumerated domain (true/false); 2) Value returned from an if/else control structure used to evaluate whether or not something is true. Boolean values are annotated using the keyword boolean. Example: boolean x = true The variable x belongs to the boolean data type and is set to true.
C# [true, false] In C#, the boolean data type is denoted as bool. C# does not provide explicit data type conversions to or from bool.
PHP [true, false] Boolean values express a truth value. Domain values are case-insensitive. Values/variables may be cast to the boolean data type using (bool) or (boolean) cast functions or automatically when returned from a conditional control structure (if/else).
JavaScript [true, false] In JavaScript, the concept of boolean values is used in two ways: 1) boolean - a primitive data type limited to the boolean enumerated domain; 2) Boolean - an object that is an object wrapper for a boolean value. The syntax of the Boolean object is a function using an uppercase B (Boolean()) that casts a character string to a boolean value. See the table below JavaScript Boolean.
Perl See remarks Perl doesn't have a boolean data type. Instead, 'boolean' values are returned when any scalar value* is evaluated in an if/else control structure. Boolean values are only defined within the context of a value returned from a specific control structure.
Ruby See remarks Ruby doesn’t have a boolean data type. Instead, it has boolean objects. It has true and false, which are singleton objects of TrueClass & FalseClass. You receive a boolean value with methods like empty?, all?, or match?
Visual Basic for Applications [True,False] VBA (a version of Visual Basic) is used to extend the functionality of Microsoft Office applications. Outside of MS Office, VBA cannot function. Given its usage with Excel and MS Access, VBA is included in this document. Native support for the boolean data type in MS Excel and MS Access can only be achieved through a VBA module.

*In Perl, a scalar value is a single numeric or string value. Perl defines several specific types of scalar values that are either numeric or strings; these include integer, floating-point, character string 

Table 2B. References

Table 2C. JavaScript Boolean

Operator boolean Boolean
typeof boolean (primitive data type) object
instanceof Boolean false true


Table 3A. File Formats

Format Values Remarks
JSON [true, false] JSON boolean values are represented as the character strings, true and false. In JSON, alphanumeric character strings are qualified with double quotation marks. Boolean value strings are not (see example below).
XML [true, false] As defined in the XML Schema Official Documentation (see references), the boolean data type (xs:boolean) belongs to the group of primitive data types. It supports the mathematical concept of binary logic (true/false). Boolean data type values are represented using two enumerated domains [true, false] or [0, 1]


Table 3B. References

JSON Example

JSON example

Appendix 2. Brief History of Boolean

The mathematical concept of boolean logic, referred to as Boolean Algebra, was first proposed by George Boole in a mid-nineteenth century publication entitled The Mathematical Analysis of Logic. In 1854, Boole expanded upon his initial work with a second publication, An Investigation of the Laws of Thought. A century later, Boole's original works would lay the foundation for modern binary computing, where information is stored in 1s and 0s or bits.

Boole implemented algebraic expressions that return truth values, a binary representation using a boolean value that indicates whether or not something is true or false. Here, all variables have a value of 0 or 1, referred to as boolean values. Boolean values (also known as truth values) are commonly represented using enumerated domains containing value pairs (true/false, 0,1, on/off, yes/no).