|Page tools: Print Page Print All RSS Search this Product|
This document was added or updated on 26/03/2020.
REPORTED HEARING IMPAIRMENT
Respondents were presented with a prompt card and asked whether they had any hearing problems that had lasted, or were expected to last, for at least six months.
Respondents who indicated that they had one or more of the following were considered to have long-term hearing impairment in one or both ears:
MEASURED LEVELS OF HEARING
Measurements of hearing were collected from respondents aged seven years and over who did not have a cochlear implant using a voluntary, self-administered test. This was the first time this data has been collected by the ABS.
The ABS worked with the National Acoustics Laboratories, the research division of Hearing Australia, to create a test that respondents could use on the interviewers' laptops out in the field to:
Respondents who completed the test were given their results on a Measurements Card (available on the Downloads tab). The result for each ear was reported on the card as:
There is variability in the descriptors and ranges used to describe level of hearing impairment both within Australia and internationally. The test categorised the results into the following levels of hearing impairment for each ear, which were based on advice from the National Acoustics Laboratories:
Respondents with a hearing impairment in both ears were classified as having a mild, moderate, or severe or profound impairment based on the ear with the lowest level of impairment. For example, respondents with a mild impairment in one ear and a moderate impairment in the other ear were classified as having a mild hearing impairment.
There are limitations with the hearing test — for example, the test is undertaken in the field with background noises, rather than in a soundproof room. However, a comparison of hearing test results with reported hearing impairment can indicate a respondent may have an undiagnosed or untreated hearing impairment.
The hearing test had a relatively high rate of non-response due to its voluntary and potentially sensitive nature.
Response rates for hearing test, by age(a)
The data from those who undertook the hearing test was then used to impute the data for those who had not taken part in the test. Imputation gives a more complete and useful dataset. The following information was produced from the hearing test:
Missing values were imputed using two steps.
Firstly, all persons aged seven years and over were imputed as having a severe or profound hearing impairment in both ears if they reported they:
Then a 'hot decking' imputation method was used. In this method, a record with a missing response (the 'recipient') receives the response of another similar record (the 'donor'). Due to the low response rates, a donor could be used up to five times to populate recipient records. Persons with imputed values from the first step were excluded from this second step of the imputation process.
After some correlation analysis, a number of characteristics (imputation variables) were used to match recipients to donors:
For example, a female recipient who has a hearing aid in one ear, has a profound disability, has completed Year 10 or higher, speaks English well or very well, is not in the labour force, lives in Tasmania, and has no doctor/GP in the local area will match to a donor record who has the same profile (female, has a hearing aid in one ear, has a profound disability, etc).
Using these variables, all of the eligible recipient records were able to be imputed successfully.
Impact of imputation
The table below shows the impact of imputation on summary results for left and right ears.
Measured and imputed hearing test results(a)
Weighting imputed data
The imputation process meant that a new set of replicate weights had to be created for the imputed hearing test data. This is because the imputation process used donor records (and could use them up to five times), which can artificially lower the Relative Standard Errors. The new set of replicate weights adjusted for this and for the additional uncertainty in the estimates due to the use of imputed values rather than reported values.
Jackknife variance estimation for hot deck imputation was used to create the new replicate weights. This resulted in an increased variance when compared to the un-imputed data, which is to be expected.
Each record also retained the original main weight calculated as part of the normal weighting process — see Response rates, sample counts and estimates (appendix).
These documents will be presented in a new window.