If we agree (hopefully) that a social enterprise survey is useful not only from an academic perspective but also as a first step to create much needed data to intervene and develop the eco-system, the second question soon arises: How do we identify the respondents to our survey? If the results from this survey are to be credible it is crucial to collect a sufficient number of responses from enterprises. As we all know, social enterprises are difficult to identify, in particular in a country where this field is just emerging and where there is no widely accepted definition of what a social enterprise is. Therefore, the two questions we were confronted with right at the beginning of our work were:
- First, how do we get a sufficient number of responses from the set of organizations that we identified so that the survey results carry some weight?
- Second, how do we identify the right kind of organizations for this survey limiting both the risk of including the ‘wrong’ ones and the risk of excluding the ‘right’ ones?
Top down, bottom up, test-based, activity based, self-identification
I looked at how researchers have recently carried out social enterprise surveys in other countries. A few started from a clearly identified database that broadly matched the respective research objective (top down). The majority, however, had to build a sample population from a variety of sources due to a lack of suitable sample frame (bottom up). The latter is more difficult to administer but includes a wider range of potential respondents. Within these two approaches, respondents can be identified based on a researchers’ test-based definition or through the respondent’s self-identification, or ‘user-driven definition’. In many cases researchers use a proxy to identify a social enterprise such as:
- The organization’s legal structure; the Community Interest Company in the UK or B-Corporations in the US,
- The organization’s involvement in certain activities (. housing associations, credit unions, community enterprises), or
- The setting of (arbitrary) thresholds on criteria such as ‘at least 25% of income most come from trading activities’.
However, such an approach may easily overlook social enterprises that operate outside of such relatively rigid proxies and criteria. Furthermore, in many countries the availability of such proxies is limited or they simply do not exist (as yet). In the case of self-identification respondents would be asked, for example, whether they would consider themselves a social enterprise. While at lower cost, more easily to administer, and more inclusive than test-based approaches, such an approach may also be subject to misinterpretation by the respondents.
How should we go about it for our research?
In Turkey, there are a number of considerations that limit the choice of survey approaches (as in many other countries with a newly emerging social enterprise sector). These include, for example:
- The notion of social enterprise has been introduced very recently and no consensus has been reached on what constitutes a social enterprise,
- Many organizations may function effectively as social enterprises without being aware of it,
- Social enterprises operate under a variety of legal structures such as associations, foundations or companies as a specific legal structure for social entrepreneurship does not exist, and
- An encompassing industry association for social entrepreneurship, industry body, research organization or think tank that could provide us with a database that also does not exist.
In addition, my research explicitly looked at social entrepreneurs with an environmental objective or ‘green entrepreneurs’ as we call them in our research. This increased the heterogeneity of the sample and thus the complexity of the required approach. Taken all these constraints into consideration I decided to identify the sample population in a bottom-up approach to take the diversity of organizations into account. This includes searching out databases of intermediaries such as civil society associations, social enterprise support organizations, universities, initiators of business plan competitions and grant managers in and outside the government. However, I found that we could not access these databases as much as we would have liked due to the confidentiality concerns of many intermediaries. I therefore decided to rely heavily on the self-identification of potential respondents, referrals of existing respondents and intermediaries based on our explanation of our research objectives and our criteria of target organizations. We also planned to actively promote the survey through our project website, our partners’ networks and social media.
Within the survey questionnaire itself I decided to focus on three broad dimensions of social enterprise to be able to filter out some organizations that are clearly outside of the focus of our work: social and environmental impact, innovation and sustainability (see part 4). Identifying a high number of suitable candidates, however, is not enough. Ultimately, what counts to make the survey results credible is the number of questionnaires that are completed. In the next part of this blog series I want to say a few words about the challenges I observed around the response rate.