Overall, the majority (61 %) of participants in our study were willing to participate in health research and only about 11 % said that they were unwilling.
Participation in our study was higher for those who had previously participated in health research across two methodologies, a case-control or cohort health study, than for the general population sample. This high rate of participation by people recruited from our previous studies suggests that their participation in a health study had not diminished their willingness to participate. In the case of the largely male cohort, the responders continued to be willing to take part in research, although men usually show lower interest in participating in health research than women, as seen among the general population participants in this study. Case-control participants had a higher participation rate in this survey than the cohort participants which may be a result of the shorter time between the case-control study and our survey. It may also reflect the reduced interest in participating in a study which is less personally relevant than the cohort study. There may be value in ensuring that all study participants are canvassed at set up as to their willingness to participant in future studies, as was done in the WABoHS study, and these individuals may provide a pool for future study recruitment.
A greater proportion of the case-control study participants (43 %) had a long term disease or disability than did the general population sample (25 %) or the cohort study population (30 %) but an individual’s case-control status was unknown in this study. Those who were cases may have been motivated by their experience of a serious disease.
Some questions such as education level or long term disability may have been subject to reporting bias but the remoteness index and SEIFA were coded from area measures and are not likely to be biased.
A limitation of our study was the 44.5 % participation rate for the general population sample. Those who refused to take part in our study may also be unwilling to take part in other health research and/or might have taken part if it had not been a telephone survey. Response rates may have decreased in bona fide health research because of confusion with increased telemarketing ‘surveys’ or because of reduced volunteering more generally in society [4]. An investigation of public opinion in Canada published in 2010 included 3000 participants selected from the telephone directory with 58.3 % participation of those eligible, while Molster et al. (2007) [18] obtained a 78 % response rate from 600 participants in a study of the views of Western Australians on the collection of identifiable data [26].
This form of research clearly involves some circularity, in that those who took part are de facto, willing to participate. However, it is difficult to see how we can explore people’s views without asking them and there were informative differences in rates of willingness associated with different questions and different subgroups of participants. Interviewees were advised about the overall aims of the study before giving consent to participate and it may be that the particular topic might have induced them to participate when ordinarily they would not have agreed. Researchers have suggested that factual willingness to participate in health research is greater than hypothetical willingness. That is, a higher proportion of people actually take part in research when asked than say they are willing do so in answer to a hypothetical question [27]. In support of this, 11 % of the general population sample who took part in our survey said they were unwilling or very unwilling to take part in health research.
We have no information about the individuals who did not answer the telephone or who answered but did not take part in the study. The telephone numbers were obtained from a commercial list without such information as age or sex. In view of this we cannot assess any non-response bias. We have compared the demographic descriptors of our general population sample to those of the general population to assess how different they are to the general population however. Our general population survey respondents were more likely to be female and older than the Australian population. Participation differed with measures of socio-economic disadvantage and by individual level educational attainment. These demographic differences in participation have been found in other studies [12, 28–32].
We found a very modest increase in response when personalised letters were sent before the telephone survey to those for whom an address was available. We did not randomise for those with and without an address and these groups may differ in regards to willingness to respond. Randomisation would have enabled us to assess whether receiving a letter enhanced participation in our study. This is consistent with a systematic review [33] and subsequent study [34] which found that personalised questionnaires and letters increased the response for postal questionnaires. However, a recent randomised controlled trial in Australia, showed that letters did not increase the participation rate in an epidemiological study using telephone interviews [35].
People were most likely express willingness to participate if the research was about treating or preventing a disease, or if it was about a disease that they or a family member or friend had experienced. Expressed willingness was increased when the question provided more information about the reason for the study, (hospital planning, treatment or prevention of a disease) or about the source of the funding, (government or charity) or if the participants were to receive a copy of the results. These findings seem to indicate that people are more prepared to take part in studies where the reason for the research is evident and the findings will be made available.
Australians in our study were less likely to be willing to participate in research sponsored by a pharmaceutical company than by a university or charity. Willison et al., in a survey of Canadian general practice patients, also found concern about their records being used for research by drug companies [36]. Canadians were also less willing to take part in research sponsored by other private sources such as a private research firm [18] or a private insurance company [36]. In a systematic review of randomised controlled trials of postal questionnaires, those originating from universities were about 30 % more likely to be completed and returned than were questionnaires from other sources [33].
Many of our participants, like those surveyed in Canada [18], wanted a copy of the summary results from the research. Kaufman et al. [37] found reduced willingness to participate if respondents were told that they would not receive their individual research results and their reduced willingness was related to privacy concerns. Very few of our participants raised concerns about privacy during the interview.
Difficulty in identifying and contacting potential participants is a barrier to participation. We used randomly selected telephone numbers for all the general population participants, but fewer than 40 % of these participants indicated that this would be a contact methodology that would make them favour participation (Fig. 1).
Our study asked participants about their preferences for survey delivery, i.e., online or mail and we found that younger participants expressed a preference for online surveys but this has not always been successful strategy for recruitment in practice [4]. A recent web-based survey of mental health among Australian University students had a 25 % participation rate [38]. Newer survey methods such as web based questionnaires have advantages and disadvantages in comparison with traditional survey methods such as telephone or postal questionnaires but have been used considerably less commonly in comparison and there are concerns about nonresponse bias and reliability of the data collected by this method [39]. Our study has provided contemporary information using an established method to inform future recruitment to health research studies.
Finally, the preference in every group for an in-depth 4-page over a brief 1-page study explanation came as a surprise, given the concern often expressed over the lengthy, potentially confusing information that is often required to be provided to participants. If the contrast had been between 1, 4, and 12 pages, the implications of this finding might have been clearer.
Reduced participation over time has been encountered across all epidemiological study designs, in terms of non-response as well as refusal, and a low response rate may reduce the statistical power and generalisability of results to the wider community. Therefore it is important to maintain current understanding of people’s willingness to participate in health research and factors that affect their willingness to participate.
Despite declining participation rates, the fact that participation in health research studies is seen positively is a good sign. Our findings have several implications for future research with a view to improving participation in health research studies. We suggest that the bone fide credentials of health researchers or their institutions are identified early in the interview process to distinguish them from telemarketing ‘surveys’. Further that the reason for the call is communicated clearly and is understandable to participants. The source of funding should be identified, and if the funding source is not a profit-making entity this should be emphasised. Participants should be made aware of the ways that the findings will be made available. Participants should be offered a range of methods of data collection (where feasible) eg offering the options of telephone, postal and web based survey. Different demographic groups may need to be targeted with different recruitment methodologies. A one page summary plain language statement should be provided alongside a longer (4–5 page) explanation for those who wished to have more information. Some individuals may have been motivated to participate by their knowledge of a serious disease or disability. Information materials should adequately describe, in lay terms the nature of the disease and the implications for sufferers of the disease. This may be particularly important for controls. Finally, researchers should ask participants at the time of enrolment to studies to agree to be contacted for future surveys.