Missing data can be a problem when there is non-response or the data is not collected for the variable mainly not at random.
A common feature of many of these surveys is non-response, and a major concern for analysis and generalisation is that the level of non-response can result in substantial biases in the population HIV prevalence estimates [10].
A major disadvantage with postal questionnaires, however, is non-response (or loss to follow-up).
The second component of the missing data issue is non-response to either the baseline or follow-up questionnaires.
The first stage of attrition is non-response, a loss of subjects to recruitment and baseline data collection.
He commented: 'not only are non-response and test incompletion in PISA distinct from proficiency, they are also nationally distinctive characteristics that change over time' (p. 33).
The response rates of the NTU and CSEW are very high, but we do not know exactly how this is measured, for instance, we do not know if framing errors such as 'wrong address' are excluded as being non-response.
another limitation of the analysis are non-response bias, however the relatively high response rate (98.8%) might minimise this bias.
In the elderly sample, there was a loss of 14.2% (6.1% were refusals, 3.1% were non-contacts and 5.0% were non-response for some other reason).
Furthermore, there was non-response on the dependent variables and consequently another 5 internists, 21 surgeons, and 118 radiologists were excluded.
