3.3 How do we study the SME, entrepreneurship and migration?

You have looked at the mixed methods approach in MIAG, combining both quantitative and qualitative data-collection techniques. So far, you have looked at the macro-quantitative dataset and used different indicators to explore the relationship between migration and growth. You then examined extracts from semi-structured interviews. These can help you to delve deeper behind the high-level macro-data that is telling us ‘what’ is happening, to unpack the more complex picture of ‘why’ causal connections are taking place. Here we turn to the third aspect of the methodology, the business survey.

A survey is a ‘means for gathering information about the characteristics, actions, or opinions of a large group of people’ (Pinsonneault and Kraemer, 1993, p. 77). They are a primary data-collection tool that involves devising a list of questions around the research topic. The questioning is often closed – ‘yes or no’ answers, or multiple choice – but surveys can include open questions that allow the respondent to give more varied answers. They differ fundamentally from an interview in that they are not the more free-flowing conversation you find in semi-structured interviewing. Kraemer (1991) notes that the distinguishing characteristics of surveys are that:

• they employ quantitative data to describe a chosen population
• survey research is collected from people, and so will be subjective
• they use data from a selected portion of a population to generalise back to wider society.

The wider field of study around entrepreneurship has, to date, largely relied on quantitative data (Jack et al., 2010); qualitative methods are used, but are limited relative to the tendency towards gathering statistical data. The Global Entrepreneurship Monitor (GEM), for example, says that it is ‘the world’s foremost study of entrepreneurship since 1999’. GEM is a network of national country teams that deploys surveys to collect data on the nature and characteristics of entrepreneurship and entrepreneurship ecosystems.

This is a trend that we see carried into the migration and development literatures. The study of entrepreneurial migrants has favoured focusing on capturing quantitative data through the macro-transfers and flows of finances through remittance, or the micro-level impact on the individual migrant or their household. MIAG has been original in its approach of the survey paying particular attention to the firm level. There are strengths and weaknesses attached to any data method and surveys are no exception.

Watch Video 3.4, in which MIAG team members Dr Dinar Kale (The Open University) and Solomon Mugane (African Migration and Development Policy Centre) talk about how they developed the survey for MIAG and analysed the data.

Show transcript|Hide transcript

Transcript: Video 3.4 Developing and analysing the MIAG survey.

Solomon Mugane:

So, Dr Dinar, why did MIAG require a survey?

Dinar Kale:

Thanks, Solomon, for that very interesting question. One of the unique aspects of MIAG’s study has been its approach to the survey. In migration studies, you will find that surveys have historically been a very common way for both researchers as well as policy-makers to get a lot of information very quickly on a large number of migrants.

Most of the times, these surveys have tended to focus on individual migrants. The data is typically about looking at migration history, experience and how it has impacted their lives as well as their families’ lives. Where we differ in our use of survey was we were looking at migrant business owners, and specifically those who owned and operated small/medium enterprises, SMEs.

MIAG survey differentiates and defines businesses using OECD’s SME and entrepreneurship outlook report categories. Firms are understood to be independent or subsidiaries of multinational companies that can be part of both formal or informal economy. And these categories include the sole trader, family or partnerships.

Now in total, in MIAG project involved conducting a survey of 1200 participants spread over four African countries. The primary purpose of conducting a survey of migrant and local firms was to establish the nature and extent of their contribution towards inclusive growth. This survey was aimed at bridging the gap between the secondary quantitative data, and the primary data collected through semi-structured interviews.

The secondary quantitative data look at in Weeks 1 and 2: you will have noticed that allowed us to describe a particular type of phenomena. So it gives you a descriptive analysis of what is happening. But that data doesn’t tell you, or doesn’t allow you, to really start understanding why a particular trend is happening.

One survey is still largely an exercise in quantitative data-gathering. It allows us to unpack more of the finer details of what was happening below the quantitative trends in a quantitative way. For example, quantitative data shows that there is an increasing trend of south–south migration, but it doesn’t tell you, what is the nature or what is the purpose of that migration.

That’s where the survey data feeds. Survey data allows us to understand the nature and purpose of this migration trend. Further, the cross-referencing of survey data with the interviews helps to create a much richer and deeper analysis of the phenomena in this study. Finally, there is a clear lack of survey data that is for some business owners operating at small and medium-scale firms, contributing firms.

The MIAG survey data brilliantly or excellently contributes to creating a unique database that was absent in the migration research conducted in academia and policy think tanks. These were primarily two critical reasons why MIAG needed a survey.

Solomon Mugane:

Dr Dinar, how did we develop the survey aligning with the OECD framework logical, culturally key information?

Dinar Kale:

Thanks for that question. And it was indeed a relatively complicated process. The development of the survey, we did it in two stages. First stage involved locating the participants, while the second, and the more complicated and critical step, involved creating a survey instrument.

The first step, we relied on business directories, street surveys in areas where manufacturing, IT and service sector firms are more concentrated. Now I will elaborate on the process of creating the survey instrument. There were a number of steps involved in that. The first was conducting a thorough literature review to identify the key attributes of inclusive growth.

We also found that the OECD report, which title was ‘The policy action on inclusive growth’, provides a good conceptual framework for our research. It was agreed that the survey instrument will align with the conceptualisation of inclusive growth identified in the OECD report.

We then reviewed the survey instruments used for collecting the similar type of data by international organisations, policy think tanks and academic scholars. We, based on all activities, then developed a rough draft of a survey. That was finalised after a few iterations, involving feedback from the project members as well as the pilot testing with some participants.

The final survey was divided into nine different sections. It focused on collecting some background data on the respondents and also, critically, the role of their businesses in facilitating inclusive growth. We categorised the questions related to conceptual framework of inclusive growth as a core and compulsory question.

These questions were selected covering what we thought we would need to be able to analyse how these businesses were contributing to inclusive growth. We adopted a structured survey instrument, where the questions asked were standardised, so that each respondent is exactly asked the same question in the same order.

The goal is that the questions are asked as consistently as possible, without interference by the interviewer, a perspective that assumes that the legal, social research can be truly objective. The survey questions included a mix of different styles of questions and responses, and these responses can be categorical or numerical. For example, the categorical responses can be yes or no, or one from a number of choices, while the numerical responses can be a continuous range (for example, ‘What’s your age?’) or discrete (like, ‘What’s your turnover?’) or scale (‘On the scale of 1 to 5, how do you rate government’s migration policies?’).

Further, we adopted a quota sampling method, where we divided the sampling frame into three sub-groups of business owners, both nationals and locals. We then set a target of 50 locals and 250 migrants, split in each country as a control to capture the key various characteristics of migrant businesses and inclusive growth relationships. This survey data provided rich information that allowed us to bridge and triangulate the quantitative interview datasets.

Solomon, you played a really important role in cleaning, organising and analysing the data. Can you please elaborate in how you cleaned, organised and analysed this data?

Solomon Mugane:

Thank you, Dinar. That was very informative. And I’ll take on my part. Insights, decisions and policies are as good as the data they are based on. It is therefore imperative that data must be of good quality. My role on the project was to handle the data processing, cleaning and management of the analysis that would help ensure the quality of real survey datasets.

The way the collection tool we used (that is, KoBo Toolbox) works, all the submitted data streams into one central point. Caution must be taken to avoid mixing up or losing data. The first critical task I had to do was to disaggregate the data by country, as the basis for cleaning it. This would help in assessing and determining the quality and the different aspects of country data independently.

The uniqueness of the survey was mutual exclusivity of data for a business. Only one submission was required from the business, and preferably, from the business owner. Any duplicates were deleted.

The survey targeted international migrants and international-owned businesses within the individual countries. Any other outliers were irrelevant and has deleted. Others deleted were columns that bore raw data. For all the open-ended questions, relevant answers were grouped into categories to ease analysis. However, some open-ended responses remain relevant independently, like the country of origin, villages, names of businesses, time, et cetera. Answers to critical questions like country of origin were fixed or structured, then conventional naming, typos and incorrect characterisations. The age of the business was counter-checked against the year the business was started and corrections made.

The multiple selection questions required a technical approach, as choices are space-separated in one cell. For example, in a question with options A, B, C, D, E and F, and where A, C and E are the selected choices, the data appears in one cell as A C E. The choices were split into different columns. Any specified columns and blank cells worked on as explained above, and the identity metric columns deleted.

The recent staffing had been captured as the actual number of employees. That, nonetheless, was and is incomparable. The data was converted into proportions for ease of use. Upon cleaning of all columns, the data was validated for use in analysis, and creation of additional variables, e.g. age groups.

During analysis, comparing different background groups and the role they play in inclusivity and growth was key. This called for creation of a variable based on the country of origin of the micro-business owners. They were classified as either intercontinental south–south, north–south or regional migrants, again as the control group, local business owners. Analysis involved, first of all, descriptive statistics for individual questions, and customisation of the new migrant variables against all questions with special focus on the critical questions.

For multiple-choice questions, the descriptive statistics were based on all choices and not just the first choice. Excel was used in all these, but other packages could be used as well.

Thank you.

Now, Dr Dinar, could you introduce us the next activity that would survey data analysis?

Dinar Kale:

Thanks for the question, Solomon. And thanks for a really good explanation about how you clean, organise and analyse the data. So, in the next activity, students will be looking at a small sample of survey data that we have collected to investigate three questions about the contribution of migrant owner businesses to inclusive growth.

Three different sets of data, focusing on key attributes of inclusive growth, have been selected for this activity. These include the factors supporting the expansion of business, the number of employees, and salary and other benefit options for the employees. Students will be comparing data from Kenya and Ghana. Thank you. All the best.

End transcript: Video 3.4 Developing and analysing the MIAG survey.

Video 3.4 Developing and analysing the MIAG survey.