“…The choice of the mode of data collection depends upon a number of factors including the data that are actually required, cost implications, amount of time available and any data validation requirements. In addition, the mode of data collection will have a direct effect upon the response rate for the study where subjects are being approached to participate, with postal questionnaires tending to have lower response rates, from around 50% to 80% (Cartwright, 1988), (McHorney et al 1994), (Mickey et al 1994), (Fox et al 1992), (Brambilla and McKinlay, 1987), (O'Toole et al 1986), (Siemiatycki, 1979) and face-to-face interviews optimising response rates to in excess of 95% (Donovan et al 1997), (Cartwright, 1988), (McHorney et al 1994), (Mickey et al 1994), (Fox et al 1992), (Brambilla and McKinlay, 1987), (O'Toole et al 1986), (Siemiatycki, 1979), (Draper et al 1999). As reported in section 2.4.2 on information bias, data may be collected from different sources for validation purposes, for example, to use data from medical notes to validate data collected at face-to-face interviews.…”