What if you want to find out whether entrepreneurs are different in Finland and in Germany, or inform the UK government on the sentiments of the export industry on their anticipations on the impact of Brexit. How would you go about it? Where would you start?
Probably, you would first start asking questions on who or what it actually is that you want to learn about. In other words what is the population that your results should apply to.
But how will you get information on this population? If you cannot include all of them, you must select. Should you do the sampling so that everyone has an equal chance, a probability, of getting selected, take those that are most conveniently reached, or snowball your way forward, i.e. ask the participants to suggest further participants? How many is enough? In quantitative research, sample sizes are larger than in qualitative research; this is because most quantitative studies seek to make generalisations to a wider population.
Let’s say you have a good sample of respondents that represent the population, how would you collect data on them? Should you rely on their subjective accounts, or find something more objective? You could either collect primary data yourself, or turn to the already existing secondary data, such as archives or databases. What would be benefits and drawbacks of surveys and questionnaires, or of doing field research?
Remember, in order to do statistical analysis, you will need the data in numerical form, so observations and interviews must be structured, coded and quantified. How can you measure the elusive concepts, such as attitudes, capabilities, orientations, and emotions – or the innovativeness of a firm, or the entrepreneurial culture of a country? In order to do statistical analysis, we must transform the concepts into measurable variables.
How we measure and what kind of numbers we assign to these variables may also vary. Do we simply divide people, firms or other entities into categories? Or, is our measuring truly continuous, or is it discrete in a sense that only certain values are possible? The level of measurement will then also affect the way our data can be analysed. Different calculations can be done with binary variables (two groups) or nominal variables (more than two groups), ordinal variables (assigned order), interval variable (quantified differences) and ratio variables (percentages). As you might already realise, the level of measurement and the variables we receive also depends on the way we collect the data and the way we formulate questions.
Without even diving into the deep philosophical debates on whether there is a reality that we should discover and assign with true values (realism), or whether whatever we measure is what it then becomes central (nominalism), there are a number of issues to consider in regard to the quality of measurement. First of all, does the response in a questionnaire really tell about, for instance, ‘job satisfaction’? Is it a valid measure for that? And, what if we ask this again, would we get the same answer, or will the person tick a different box depending on the weather outside or due to her/his character – in other words, is the measure reliable? How could you test the reliability and validity of a selected measure?
The saving grace is that research does not have to represent the reality to the full extent – it is actually meant to summarize and simplify it in such a way that we can make wise decisions, and accumulate knowledge. Thus, some measurement error always exists, and this is ok. However, the data is the bedrock of our research and no fancy analysis can correct the damage (or even carried out) if the data is flawed, and contains too much random noise or systematic bias. How could you avoid bias, and gain the best possible basis for further analysis?
Kuckertz, A. & Wagner, M. 2010. The influence of sustainability orientation on entrepreneurial intentions — Investigating the role of business experience. Journal of Business Venturing, 25(5), 524-539.
Population and Sample
1 - Objective and Subjective
2 - Primary and Secondary
3 - Observation, Survey, Experiment
4 - Survey
Measurement and Scales
Validity and Reliability
Summary of Quantitative Data Collection
Exercise 5.1 – Comprehend
Take a look at two databases that are freely available and offer interesting data for entrepreneurship and innovation management researchers: Global Entrepreneurship Monitor (GEM) European Innovation Scoreboard (EIS).
Give a brief (max 1 page) account on your impressions on one or both of the databases. What kind of data is available? How is the data collected? What do you find interesting in the database?
If you are interested to learn more about databases, see for example:
Wennberg, K. 2005. Entrepreneurship research through databases: Measurement and design issues. New England Journal of Entrepreneurship, 8(2), 9-19.
Exercise 5.2 – Critique
Read the Data, Measures and Limitations sections of the article by Kuckertz and Wagner (2010), and answer the questions on it: What kind of data does the study use? How well do the authors describe the process of data collection? Do they assess the validity and reliability of the data, and its potential limitations? Would you have ideas for improving the data (e.g. do you think they should have used different data, or collect more data)?
Please check. Did you gain an understanding of the following?
- The meaning of population and sample
- Some data collection methods used in quantitative research
- Process of measuring and how to assess the quality of measurement
If you can answer everything with a confident Yes! then you have achieved the learning objective of this session.