In formal organizational communications, it is common for individuals, departments and organization to report primarily positive information about their activities, while systematically downplaying or covering up negative information. While this simple phenomenon has been noted by researchers, it has never been adequately explained and its profound implications for a wide range of organizational processes have gone largely unexplored. Instead, most organizational theories are based on implicit assumptions about the accuracy, quality, representativeness, and reliability of information that are rarely met in practice. Where biased information processing in organizations has been considered by researchers, it has been typically attributed to individual motives or political maneuvering. This research considers the phenomenon in terms of structural factors inherent in the design of modern organizations, by examining how language is used within organizational structures to represent events and actions in a biased manner for communication purposes.

A theoretical framework for explaining the phenomenon is developed, which considers the interplay of several elements:​ formal structure in organizations, responsibility, legitimate authority, event complexity, constrained formal communication channels, language, image, as well as cognitive mechanisms of abstraction, selection and categorization. All organizations and organizational units elaborate formal structures which present simple and inherently positive images about their internal ongoings for external audiences or publics. Individuals and units are assigned formal responsibilities by their legitimate authorities within the structure, creating a situation in which the simple and positive images of formal structures must be confirmed and maintained through the generation of formal information by individuals and units. Management control systems are supposed to ensure that unit actions are consistent with their responsibilities.

However, because it is impossible to communicate every detail of a unit's activities, a great deal of information reduction must occur as information moves up the management hierarchy during the reporting process. Information is typically reduced by using language-based reports (or surrogates for language, such as statistics, charts, etc.), to represent formally the ongoings and events within organizational units. Flexibility in the relationship between events and their representations in language changes the nature of the hierarchical control process. Formal structures and formal responsibilities define a system of ambiguous language categories that can be operationalized in a wide variety of ways. Similarly, events and actions are complex, and can be "packaged" or "labelled" to best reflect formal output requirements. This "many to many" relationship between events and language provides opportunities for bias and selectivity to enter formal communication processes. Formal information can be reported in a manner that creates and maintains the best possible image for the unit - to demonstrate formally that the unit is doing as it is supposed to be doing. The result is that management control systems primarily constrain the formal informational outputs of organizational units, rather than actions directly. The degree to which unit actions are also constrained depends on the degree of "representational transparency" - the degree of correspondence between the actions and their representations in formal information.

Overall, formal hierarchical communication in organizations is viewed as a collective process of constructing and maintaining an elaborate organizational image, by constraining the informational outputs of organizational units. The framework considers formal hierarchical communications as a "pseudo-scientific" process, in which legitimate formal structures act as a biased hypotheses, that are operationalized in language and actions by organizational members. These structural hypotheses are then selectively measured and reported on through a positively biased measurement system, which always tends towards hypothesis confirmation. Through flexibilities available in language, organizational members generate information that tends to confirm for higher management that they are doing what they are supposed to be doing.

An in-depth case study, conducted in a high technology industrial organization over a period of about one year, was used to empirically examine aspects of the framework. The case results provide a wide range of evidence in support of the theory and demonstrate at several different levels of analysis how the formal structure placed constraints on communication and language use within the organization studied.

Practical and theoretical implications of the theory and empirical findings are examined. Potential directions for future research are discussed.