G. R. Carrol and M. T. Hannan, "Density dependence in the evolution of poplations of newspaper organizations", plus comment by L. Zucker and rejoinder by Carroll and Hannan, American Sociological Review, 54 (1989), 524-548.

Population density -- the number of organizations -- is assumed to be a function of the social processes of legitimization and competition. At low density, their model predicts that the legitimization process will dominate and will lead to high organizational founding and low mortality rates. At high levels of density, competition will dominate, leading to low founding rates and high mortality rates. The model is tested with data from newspaper companies.

Ther model "holds taht both rates for founding and mortality in organizational populations are non-monotonic functions of the number of organizations in the population (density)"

Why do organizational populations follow these trends? Some possiblities are:

* they exploit dynamic resources, growing when abundant and shrinking when scare

* older populations lose their competitive edge as new technologies emerge and social situations change

* opposing processes of legtimization and competition. At low density, growth in numbers mainly legitimizes a population and the organizational form it uses. But as density increases, competition becomes more important (especially when density is high relative to resources).

"Institutional theory stresses legitimacy as taken-for-grantedness rather than legality or official approval. Actors regard it as the "natural" way to organize for social purpose. Thus, rarity of form makes it difficult to be legitimate. However, when there are many organizations with that form, legitimacy doesn't require more organizations.

The researchers then develop a model to predict the population curve based purely on the population itself. This is an indirect model that doesn't rely on measures of competition and legitimacy. They are in effect "estimated" from the number in the population.

Previous Research on Density Dependence

Hannan and Freeman analysis of labor unions and semiconductor manufacturers support the legitimization and competition. Other studies on independent local telephone companies, medical diagnostic imaging, voluntary service organizations,

However, other studies "ran counter to the model". The rate of entry into semiconductor manufacturing was monotonic than non-monotonic. Analysis of entry and exit of wine producers shows limited support.

One main difference with these non-supporting studies is their shorter time frame. Their initial study was over 150 years, while others were only 13-40 years. Some studies also lack data on the "formative years" of the population, which are critical for analysis.

Newspapers and Legitimization

Initial newspapers had much trouble gaining legitimacy, often requiring governmental approval for content. Initial failure rates were high. The first papers were either mercantile or political in nature. and usually one-person operations. Later the Penny Press in the 1830's began the first mass-produced newspapers, which eventually relied on sales and advertising income and divorced themselves from political interests.

By the late 19th century publishing had become a profession.

Competition

Competition began in the 19th century and intensified inthe 20th century. Daily papers grew from 850 in 1880 to 2200 before WW1 and declined to 1750 between WW1 and WW2, being stable after that. In cities there has been a steady downward trend to one daily. Economies of scale become important.

Data

They analyzed newspaper data from Argentina, Ireland, and selected parts of the US.

Results:

Foundings - For all nine populations, density has a non-monotonic effect.

Mortality - The non-monotic effect is stronger for larger populations.

Discussion

"The results show that density-dependent processes of legitimization and competition are powerful explanations of organizational evolution"

The results are more prevalent for larger populations and ones with longer durations.

"Processes in legitimization and change constitue the dominant mechanisms of change in two leading perspectives on organizations, institutional theory and ecological theory. The density model integrates these two theories but with the twist of recasting institutional processes at different levels of analysis.

Critique by Zucker

Zucker contends that actually Carroll and Hannan's paper doesn't bring convergence for ecological and institutional theories. Neither legitimacy nor competition are actually measured. No evidence actually supports the claims that legitimacy or competition are affecting birth or mortality rates.

It also seems that competition should be greatest when there are few players, not when there are so many an extra one has little impact.

The newspaper study also doesn't account for the competion between populations (like newspapers vs TV or radio) that surely affect the data.

The data can also be explained by "shocks" (due to new technology, new demand, etc.) that cause increased births and decreased death rates too.

The equations used assume that all time intervals between events is independent. This assumes that multiple actors in a city wanting to found a firm drop their plans and start over when another enters the market. The formulas also do not depend on historical time, which is counterintuitive.

In general, Zucker thinks that legitimacy only accounts for "only a modest increase in the rate of founding" and that the research doesn't really answer the question or bridge the two theories.

Rebuttal to Zucker

Most researchers can measure key organizational variables directly. Even the institutionalists must make inferences about factors like "institutionalization" from more easily measured variables. There is no possible way to directly measure legitimacy and competition of history.

Actually, they refute Zucker's claim about ignoring historical time, stating that it is included in the analysis.