Candice Li
21 October 2019
Philosophy, The University of Sydney
One inductive approach to make an argument is statistical generalisation. An inductive approach provides evidence to support the argument while it does not guarantee the conclusion. According to Manley (2019), statistical generalization refers to a conclusion drawn from the features of a sample and attributed to a population. I suggest that this is a relatively weak argument which can easily contain many flaws and might not have enough support to show the features of a population. In this essay, I will give an example and illustrate the limitations of this type of argument.
The Foundation of Young Australians (2018) reports that young Australians are facing barriers for seeking full-time work based on a survey which shows that 50% of 14,000 Australian young people are not working full-time. This survey observed 14,000 young Australians aged 15-25 years old and collected the information of their education and employment status. The new result turned out in 2016 that 50% of 25-year-old Australian young people observed are not working full-time, which is 7% lower than that in 2006. (The Foundation of Young Australians (FYA), 2018) Then the report draws the conclusion that obtaining full-time jobs is increasingly precarious and young people are facing barriers for seeking full-time work, and it analyses the difficulties and accelerating factors for getting full-time work. I suggest that this statistical generalization contains some flaws and might not reflect the truth. I will discuss it in two parts, which includes the relationship between detected property and target property, and the similarity of sample and population.
Below is the standard form of this generalization:
P1: 50% of 14,000 25-year-old Australian young people are not working full-time.
P2: Not working full time indicates that people are facing barriers for seeking full-time jobs.
C1: 50% of 14,000 25-year-old Young Australians are facing barriers for seeking full-time jobs. (From P1 and P2, induction)
P3: These 14,000 25-year-old young Australians are similar to all Australian young people.
C2: 50% Australian young people are facing barriers for seeking full-time jobs. (From C1 and P3, induction)
C3: Australian young people are facing barriers for seeking full-time jobs (From C2, induction)
The first part I will argue is whether the detected property indicates the target property. Detected property means the observed features, which is ‘not working full-time’ in this argument. While target property means the features attributed to the population in the conclusion, which is ‘facing barriers for seeking full-time jobs’. Whether the detected property indicates the target property is one of two important factors to make a good generalization. The other factor is the similarity of the sample and population, which I will discuss in the next two paragraphs. In this example, the detected property is not a very good indication of the target property because not working full-time does not necessarily refer to facing barriers for seeking full-time jobs. There are many other reasons that can lead to this result. According to O’Donnell and Zion (2019), younger Australians are attracted by more creative, interesting and long-lasting jobs. They suggest that many people in creative industries prefer flexible work styles although the pay and working conditions are unstable. (O’Donnell and Zion, 2019) Thus, one of the alternative possibilities is that the younger Australians are interested in more creative and flexible work styles instead of full-time jobs, which can result in the decline of the proportion of those working full-time. It is also mentioned by the FYA report (2018) that 25-year-old young people in 2016 are better educated than before. In hence it is likely that more young people are working part-time or not working while studying. And some people are not working full-time because they choose to do volunteer work or care for children at home. (FYA, 2018) These are some of the possibilities that they are not working full-time but also not facing barriers for seeking full-time jobs. Therefore, it is inappropriate to get to the conclusion that young people are facing barriers for seeking full-time jobs because of the result that 50% of 14,000 25-year-old Australians are not working full-time. Thus, the detected property does not properly indicates the target property.
The second part I will illustrate is the similarity of the sample and the population. A good statistical generalization has samples that are relevantly similar to the population. In this argument, the sample is 14,000 25-year-old Australians and the population is Australian young people. The first aspect of the similarity I will discuss is the size of the sample. In this example, it is not a relevantly big enough sample. This survey has followed 14,000 25-year-old young Australians for 10 years. The .id website (2016) shows that there are 2,988,404 young people ranging from 15-24 in Australia in 2016. Although 14,000 seems a big number, it is only approximately 0.47% of young people in Australia. Based on the law of large numbers, the larger the sample is, the more possible that it reflects that feature of a population. (Manley, 2019) Hence, the possibility that it reflects all Australian young people is not much higher than the possibility that it does not. Consequently, it is not strong evidence to support that it is a big enough sample which can show the feature of all young people in Australia. It is also not appropriate to draw to the conclusion that young people are facing barriers for seeking full-time jobs because 50% of young people are considered as facing barriers.
On the other hand, another reason that the sample is not similar to the population is that it can easily contain biases in sample selection. The first bias it might contain is sampling bias, which is created by the way people sample the population. (Manley, 2019) It is very likely that the report does not pick a sample that can fully reflect the whole population. When analysing the accelerating factors for seeking full-time jobs, the FYA report (2018) simply compares those who have different socioeconomic status and genders, which does not contain other important factors such as different areas and races. However, these easily ignored factors can affect the result. For example, people living in suburban areas may have different full-time working proportions as those living in city areas. Different states can share a higher or lower proportion than the average. And the percentage of people working full-time may differ from different races and classes. Thus, it is difficult to ensure that this sample reflects the features of the population. According to Manley (2019), a representative sample that can avoid sampling bias should have the same species of variety as its target population. Therefore, in this argument, the possibility that the sample is self-selected which does not have the same sort of variety as its target population is higher than that the sample is a relevantly representative sample. To solve this problem, one approach mentioned by Manley (2019) that can be taken is stratified random sampling. It helps increase the possibility that the sample reflects the population. To use stratified random sampling, they can divide the population into subgroups with all the features that are relevant to the result. Then they should choose the sample from each subgroups with the proportion of the sample matching the proportion of the whole population. However, in this case, it may still contain inaccuracy and is very hard to fully reflect the population even it is using stratified random sampling. Hence, the sample selected is not relevantly similar to the population.
In conclusion, the argument made by FYA reports seems that contain many flaws because the detected property is not a good indication of the target property and the sample is not relevantly similar to the population. Therefore, as an inductive approach, statistical generalization might have limitations and not truly reflect the features of the whole population.
Reference List:
Australians, F. (2018). The new work reality. Retrieved from Foundation for young Australians website: https://www.fya.org.au/report/the-new-work-reality/
O’Donnell, P., Zion, L. (2019). Precarity in Media Work. In D. M. & P. M. (Eds.),Making Media: Production, Practices, and Professions(pp. 223–234). Amsterdam University Press.
Manley, D. (2019).Reason Better: An Interdisciplinary Guide to Critical Thinking.Toronto, ON, Canada: Tophat Monocle.
Five year age groups. (2016). Retrieved from: https://profile.id.com.au/australia/five-year-age-groups
网友评论