Question

The organisation ‘Australian Hearing’, an agency of the federal government, published a report on ‘binge listening’. As part of the research for the report, they carried out a survey, which was described as follows:

'One thousand Australians aged between 18 and 35 years in metropolitan and regional locations in all states across all education levels participated in a 15- minute online survey. The questions related to their exposure to noise during leisure activities, their perception of noise, perceived causes of hearing loss and attitudes towards hearing protection. The research was conducted by Inside Story.'

Which one of the following propositions is reasonable, in the circumstances and given only the information provided in the paragraph?

a) Although there is not much detail in the description here, it is reasonable to assume that a survey conducted by a federal agency was well-conducted and that the results will be reliable.

b) The large sample size of 1000 means that the results will be unbiased.

c) The coverage of all areas and states in Australia and all education levels means that the result will be reliable.

d) The online mechanism used to provide responses was a reasonable way for the survey participants to provide their views.

Answer 1

a) Although there is not much detail in the description here, it is reasonable to assume that a survey conducted by a federal agency was well-conducted and that the results will be reliable.

Not reasonable. A survey conducted by a federal agency may be unreasonable if it is not properly designed.

b) The large sample size of 1000 means that the results will be unbiased.

Not reasonable. Bias is the tendency of a sample statistic for systematic overrepresentation or underrepresention of a population parameter. A small sample size may inadequately represent all sections of a population thus making it biased. A large sample may help in reducing this bias. However, a large sample may still have systematic underrepresentation or overrepresentation of some sections of a population so even with a large sample, there can be significant bias present. For example - Supose there are 3 age groups of people (Child, Adult and Old) and a survey about food habits is to be conducted. If the survey is conducted by taking a much larger number of children than adults and old people, the results would be biased towards the food habits of children, even if the sample size is large (out of total sample size of 1000, 900 children, 50 adults and 50 old are surveyed).

c) The coverage of all areas and states in Australia and all education levels means that the result will be reliable.

Reasonable, provided the sample size is also large. Reliability is consistency of test results, that is, if ihe test is conducted several times under similar conditions with the same population then it should yield similar result. In this example, the population is the 18 to 35 year olds in Australia. A survey has been conducted across all states, regions and education levels in this population. Hence it represents all sections of the population. Within a region and education group, people's attitudes towards noice will remain constant on an average. Hence, if the survey is conducted again in similar conditions (again including all states, regions and education levels), it should give similar results as the first survey. This is true no matter how many times the survey is conducted including all the states, regions and education groups. So we can assume that the results of the survey will be reliable.

The survey size should also be large in order for results to be reliable.

d) The online mechanism used to provide responses was a reasonable way for the survey participants to provide their views.

It depends. It may be unreasonable if some of the people don't have access to the internet and hence can't take the esurvey. Non-inclusion of these people may introduce a bias in the survey. However, conducting an online survey is cheaper and easier for the federal agency as a physical survey would involve deploying surveyors and approching poeple individually for taking the survey.