Organizing for Society: A Typology of Social Entrepreneuring Models

We use the terms social entrepreneur and social entrepreneurial organization interchangeably throughout the article as our empirical and theoretical focus is organizing.

Mohr and Guerra-Pearson (2010), for example, used relations between categories of relief recipients, classes of social problems, and the type of activities undertaken as indicators of models used by 600 welfare organizations in New York City during the Progressive Era. Also focusing on the Progressive Era, DiMaggio and Mullen (2000) selected the type of actors involved, the category of actions taken, and the object of action (i.e., the audience) to surface distinct models that shaped civic rituals related to National Music Week.

For text sampled, refer to www.​ashoka.​org and www.​schwabfound.​org.

A stratified and weighted random sample was constructed in the case of Ashoka fellows to reflect the distribution across regions and year elected.

We collected all profiles on the web in July 2007.

Unlike other statistical methods for studying configurations such as deviation scores, where the researcher defines ideal types and then calculates distances between the ideal profiles (Delery and Dote 1996), and fuzzy set qualitative comparative analysis (fsQCA), where the selection of attributes is based on theoretical knowledge about their relationship with the outcome (Fiss 2007, p. 1183), cluster analysis makes no prior assumptions about differences in the sample and does not predict outcomes in advance.

Of the hierarchical procedures, Ward’s algorithm has provided superior clustering solutions over other algorithms across distinct applications (Blashfield 1976; Milligan 1980; Mojena 1977). We re-ran k-means cluster analysis using average linkage algorithm to define the initial seed and the results scarcely changed (Cohen’s kappa inter-agreement = 0.76).

The Calinski and Harabasz pseudo-F stopping rule index calculates the ratio of total variation between clusters versus total variation within a cluster. Larger values indicate more distinct clustering. The maximum hierarchy level was used to indicate the correct number of partitions in the data (Calinski and Harabasz 1974). Duda and Hart (1973) proposed a ratio criterion where Je(2) is the sum of the squared errors within a cluster when the data are broken into two clusters, and Je(1) provides the squared errors when one cluster exists. The rule for deciding the number of clusters is to determine the largest Je(2)/Je(1) value (0.8466) that corresponds to a low pseudo-T ² value (10.15) and has a higher T ² value above and below it.

Information about the IHRDA was collected from the organization’s website, http://​www.​ihrda.​org/​, and retrieved on June 27, 2012.

The ACERWC is a committee of 11 experts appointed by the general assembly of the heads of states of the African Union (AU). These experts examine cases against nations, investigate them and decide whether there is a violation of the African Charter on the Rights and Welfare of the Child, which is a charter ratified by the members of the AU. This committee has no legal power, only the ability to make resolutions and declarations to the members of the AU.

Information about Soul City was collected from the organization’s website, http://​www.​soulcity.​org.​za/​, and retrieved on June 27, 2012.

Information about Honey Care was collected from the organization’s website, http://​www.​honeycareafrica.​com/​, and retrieved on June 27, 2012.

Information about Taproot Foundation was collected from the organization’s website, http://​www.​taprootfoundatio​n.​org/​, and retrieved on June 27, 2012.

The Cramer’s V is a χ²-based measure of nominal association which assesses the association strength between two variables where 1 is a perfect relationship and 0 is no relationship. Cramer’s V overcomes the requirement to fill every cell of crosstab matrix.