3.3 Research Population & Census
Marcos Antonio de Lima Filho, PhD.
This study compared the design evolution of two distinct industries: the design of commercial passenger aircraft and the design of smartphones. This section describes these research populations and the statistical procedures employed to study them. Instead of examining a portion of the target populations, it was decided that a census could be a better option. This section also discusses the advantages of censuses and the limitations of sample surveys for small and variable populations.
The smartphone industry was the initial area of inquiry. The research later evolved into a comparison of smartphones and commercial passenger aircraft. As advised by Glaser and Strauss (1967/2006), the grounded theorist should maximise the differences among comparison groups. This “increases the probability that the researcher will collect different and varied data, while yet finding strategic similarities among the groups”. The section ends with a brief discussion of comparative strategies.
The Research Population
To enhance the reach and transferability of its discoveries, this research employed a “most different” approach to comparative analysis. This strategy required comparing two distinct industries, which resulted in the creation of two distinct datasets. The first was a sample of 305 smartphone specifications, which later evolved into a census of 13,212 specifications. The second dataset was a census of airliner production and design (see Table below).
The Census of Airliners focuses on commercial passenger aircraft. As such, it does not cover military, cargo, and private aircraft. Each type is designed for a specific mission, thus leaving little common ground for comparison. The census is focused on aircraft built for scheduled services, which are “flights scheduled and performed for remuneration according to a published timetable, which are open to direct booking by members of the public” (ICAO, 2009).
The census of commercial passenger aircraft examined the specifications and production lists of 283 aircraft models introduced between 1932 and 2020. In this period, 53 manufacturers from 19 countries/regions delivered 54,793 airframes to commercial airlines.
The Census of Smartphone Technology examined the specifications of 13,218 mobile phones released by 424 brands. This dataset spans 20 years, from January 2000 to December 2020. This census encompasses the predecessors of the smartphone. This reflects the understanding that the modern smartphone is an evolution of the PDAs (Personal Device Assistants) and feature phones of the past. Therefore, it was essential to include these ancestors as they constitute a critical link in the evolutionary chain.
Census
A census is an attempt to list all elements in a group and to measure one or more characteristics of those elements; the group is often an actual national population, but it can also be all houses, businesses, farms, books in a library, cars from an assembly line, and so on (Cantwell, 2008, p. 90). In short, a census is a count of all the elements in a population; if 4,000 files define the population, a census would obtain information from every one of them (Cooper & Schindler, 2014, p. 338). In this research, the registration of 54,793 airframes defined the total population of commercial airliners. Likewise, 13,212 product specifications represented the total population smartphone specifications.
Besides its application in demographic research, censuses are also conducted to enumerate agricultural production, livestock, road traffic, motor vehicles, electric vehicles, and all sorts of economic activity. In ecological and conservation work, researchers conduct censuses to individually count plants, insects, fish, amphibians, reptiles, mammals, and birds. The technique is applied when it is possible to count the animals or plants in the whole population of interest (Greenwood & Robinson, 2006, p. 13).
The Advantages of a Census
Censuses are often contrasted with sample surveys, and the two share many of the same methods. However, whereas a census is intended to gather information about all members of a population, a survey gathers information from only some of the population members (Lavrakas, 2008). There are several compelling reasons for sampling, including lower cost, and greater speed of data collection (Cooper & Schindler, 2014). However, the advantages of sampling over census studies are less compelling when the population is small and the variability within the population is high:
Two conditions are appropriate for a census study: a census is (1) feasible when the population is small and (2) necessary when the elements are quite different from each other. When the population is small and variable, any sample we draw may not be representative of the population from which it is drawn. The resulting values we calculate from the sample are incorrect as estimates of the population values. Consider North American manufacturers of stereo components. Fewer than 50 companies design, develop, and manufacture amplifier and loudspeaker products at the high end of the price range. The size of this population suggests a census is feasible. The diversity of their product offerings makes it difficult to accurately sample from this group. Some companies specialise in speakers, some in amplifier technology, and others in compact-disc transports. Choosing a census in this situation is appropriate (Cooper & Schindler, 2014, p. 339).
That was the case with the original sample survey of smartphone specifications. The grounded theory process began with this dataset, but due to the high degree of variability in smartphone features, the sample design resulted in inaccurate population estimates. To correct this, the sample of 305 smartphone specifications was expanded to a census of 13,212 smartphone specifications. This census was also feasible due to the availability of historical documents.
Censuses & Innovation Studies
Although censuses of technologies or industries are uncommon in research projects, they do exist. Consulting companies conduct most of these censuses for marketing research purposes. Due to the strategic implications of such data, access to these reports is severely restricted. IDC, for example, charges between $4,500 and $40,000 USD for its census reports. IDC formerly conducted a census of the personal computer industry in the United States (Processor Installation Census), but this series was discontinued in 1994.
Despite its use in market research, innovation studies rarely feature a census design. Even so, some of these census studies have led to seminal contributions. Abernathy and colleagues, for example, proposed various innovation concepts based on a census of the US car industry, such as disruptive innovation, incremental innovation, radical innovation, and others. Similarly, Clayton Christensen’s redefinition of disruption was based on a census of the hard disc-drive industry. Table 3.3.2 summarises a handful of innovation studies that featured a census design. The table compares the variables covered in each study and their respective data sources.
Comparative Analysis & Sampling Strategy
Comparison can be a powerful stimulus for concept formation, as well as for the inductive discovery of new hypotheses and theory development (Collier, 1991). Przeworski and Teune (1970) classify comparative methods into two main types: the “most similar” and the “most different” designs. These authors contend that the former approach runs into serious difficulties by failing to eliminate rival explanations. A most different design, in contrast, deliberately seeks cases of a particular phenomenon that differ as much as possible, since the research objective is to find similar processes or outcomes in diverse cases (George & Bennett, 2005).
When comparing different cases, Przeworski explains that the researcher is forced to distill out of that diversity a set of common elements with great explanatory power (Collier, 1991). This type of comparison can significantly assist the researcher in transcending substantive descriptions of time and place as he or she tries to achieve a general, formal theory (Glaser & Strauss, 1967/2006, p. 55). As a result, the diverse case method is likely to be more representative than other small-N samples (Seawright & Gerring, 2008).
The most different method requires the selection of a set of cases (at minimum, two) which are intended to represent the full range of variance along relevant dimensions (Seawright & Gerring, 2008). Smartphones and aviation fit into different categories of products, as well as industrial segments.
Furthermore, in addition to the categorical differences, the consumer electronics and aviation industries also display extreme differences when compared along continuous variables. In these cases, the researcher usually chooses both extreme values (high and low). The design of consumer electronics and aviation are diametrically opposed in the following dimensions:
Product life cycle (2 year vs 20 years).
Costs (hundreds of dollars vs millions of dollars).
Dimensions (small vs massive).
Competing models (high variation vs low variation).
Competing manufacturers (perfect competition vs duopoly).
Annual Production (1.2 billion vs a couple thousand).
Pace of evolution (fast vs slow).
By encompassing a full range of variation, the main advantage of the “most different” comparison is that it enhances the sample’s representativeness (Seawright & Gerring, 2008). In this research, the constant comparison of organic and industrial evolution, naturally selected and artificially selected, has further enhanced the interchangeability of concepts and their potential for generalisation. Without a “most different” comparison, this thesis’s contribution to knowledge would be far more restricted and less significant.
Last updated
