Journal of the New Zealand Medical Association, 21-January-2011, Vol 124 No 1328
Choice of the channels for delivery of health information to the public is a critical decision that faces agencies interested in improving the health status of a population. Trust in different sources of information and media through which the information is disseminated can be a major factor in determining the effectiveness of any health promotion.1,2
Research in recent years has focused on the use of the Internet3 as a health information source4 (HIS) and its acceptability as a medium of communication.4–6 However, relatively little attention has been paid to the wide variety of other sources individuals can draw upon for information. Instead, research reinforces how the Internet is preferred in use7 as a HIS but that it is not highly trusted.8
Findings also are mixed regarding whether increased frequency of use changes this.9,10 It is recognised however, that trust toward a HIS can improve adherence to medical advice provided.11 Therefore, it is important to gain a comprehensive understanding of how people trust the wide variety of sources that are used to deliver health information. In this paper we examine the expressed trust that people have in a range of 24 different sources and types of media. In terms of overall approach this research complements that undertaken in the US by the Health Information National Trends Survey (HINTS). HINTS focuses on how personal characteristics influence perceived information needs and examines the consequent effects for choosing appropriate information channels for health information.12
‘Trust’ is taken to mean the “message received is true and reliable and that the communicator demonstrates competence and honesty in conveying accurate objective, and complete information”.13 Previous studies demonstrate trust toward information sources influences both usage14 and frequency of searching.9 Research on trust in sources for health information has also produced interesting links to personal characteristics such as education, ethnicity and sex which can then be used to assist in the selection of different media and potentially increase the effectiveness of health communications.7
A key way in which this study differs from previous research on trust in health information is the range of sources investigated. Firstly, we have a comprehensive list of different types of health professionals, including physiotherapists, nurses and alternative sources such as homeopaths. Additionally our range includes different types of institutional sources, including Government agencies and some charities, as well as media types and friends and family. Many of these are consistently omitted from studies on health information.7,15,16
The data reported in this paper were collected during a replication of the Maibach et al. American Healthstyles study16 adapted for and conducted in New Zealand and funded by Sport and Recreation New Zealand (SPARC) and the Cancer Society.17 The questionnaire covered comprehensive information regarding health status and beliefs as well as a data on physical activity and fruit and vegetable intake. Measures of trust were replicated from the Maibach survey with appropriate updates and adjustments for the local context.
A total of 24 relevant sources and media were included in the survey (see Table 2) and respondents were asked to indicate their level of trust toward each source with respect to obtaining health information. This level of trust was measured on a five point scale ranging from don’t trust at all (1) through to trust a lot (5). While this approach to measurement obviously cannot capture the variation that lies within any particular category, for example a patient is likely to trust one doctor within their general group practice more than another, it is still clear that people do hold overall attitudes to different types of information sources and that measurement at this global level allows the comparisons across the wide assortment of sources and media that is intended in this paper.
The survey was mailed to a sample of New Zealanders drawn from the electoral roll, achieving a 61% response rate and yielding a total of 8,291 respondents for analysis. Those identifying themselves as Māori were ‘oversampled’ by 26% in order to compensate for the normally lower response rate from this group. Specifically, individuals were required to self report themselves as Māori or of Māori descent to be eligible for the Māori electoral roll, which was then used as a sampling frame for this group.
A summary of the main demographic characteristics of the sample is given in Table 1 which shows more female respondents than would be expected. The categories for other variables in the table have been collapsed for reporting purposes and are presented to show the main features of the sample. Respondents were allowed to nominate more than one ethnic grouping, an option chosen by 4.7% of the sample. People choosing more than one ethnicity were removed from the analysis for comparisons on this variable. Different Pacific Island groups (Cook Islands, Niuean, Samoan and Tongan) were recorded separately in the survey but were subsequently amalgamated to one group as no differences were identified between them in analysis on the trust ratings.
Table 1. Characteristics of study participants
Exploratory factor analysis, using principal axis factoring and direct oblimin rotation, was conducted on the trust ratings for the 24 different HISs in order to identify underlying patterns in the ratings of different information sources. Polychoric correlations estimated in LISREL were used as inputs for the factor analysis and all other analysis was conducted using PASW v18.0 software. Summary variables were generated to represent the factors by computing an average score for all the variables that loaded with an absolute value greater than 0.6. A multivariate general linear model was used to compare these trust variables across categories of age, gender, education, income, ethnic background and occupational status.
Before exploring the results of the factor analysis it is useful to briefly consider the median trust ratings and ranges for each of the information sources (see Table 2). Medians are reported as opposed to means because four of the rating scales deviate significantly from a normal distribution.
Table 2: Median trust scores for each health information source
Range = 1–5: “don’t trust at all” (1) to “trust a lot” (5); n = 6,541. SPARC: Sports and Recreation New Zealand.
Overall the pattern of responses for these average trust scores looks reasonable and consistent with expectations from previous research. For example, similar to the HINTS survey,12 personal professional sources, in particular the person’s own general practitioner, but also their nurse and trained dieticians are seen as very trustworthy sources of health information.
The three other HIS that are rated especially highly are the three major national charities that were included in the list: The Heart Foundation, Cancer Society and Diabetes New Zealand. Amongst these the rating for the Heart Foundation is significantly higher than the other two. This is possibly a reflection of its high public profile and the repeated exposure that it achieves through schemes like the product endorsements that it offers to ‘healthy foods’. The Heart Foundation mark of approval is a recognized symbol on many food products and offers continuous reinforcement of its name as a supplier of health-related information.
While the remaining sources all have a median of 3, the mean scores suggest that they fall into three broad groups. The highest rated of these three is a mixed range of sources including official public health bodies, professional in health-related occupations including pharmacists and gym/personal trainers. Friends and family are rated next and the lowest set of ratings are those offered to the mass media sources, though the idea of books and journals contains more credibility with the sample than do the other sources.
Table 3 gives the main results of the exploratory factor analysis. For ease of interpretation only loadings above 0.4 are displayed. Five factors had an eigenvalue of over 1 with a sixth factor having an eigenvalue of .989. The scree plot showed a marked drop after that with the next eigenvalue at .667. The 5-factor solution essentially combined Factors 3 and 4 in table 3 but the resulting factor was also highly correlated (.645) with Factor 2, with many sources cross-loading. The 6-factor solution presented below is much more interpretable with more face validity.
Parallel analysis was also conducted following the procedure published by O’Connor.18 This also produced an ambiguous result with the sixth factor failing to meet the equivalence criteria by 0.06. Ideally future work in this area would test both five and six factor structures on a different sample using confirmatory factor analysis. In total the factors explained 77.6% of the variance. The lowest communality for any single variable was 0.408 (trust in naturopaths).
The distribution of factor loadings in Table 3 has considerable face validity. There are a few items that cross-load between factors, especially three and five, but most sources only load on one factor and those (such as dieticians and SPARC) which are more distributed are understandable. Factor 1 was identified as grouping mass media together as sources for health information.
The variables that load most heavily on Factor 2 are the three major New Zealand health charities that were included in the list. SPARC and Regional sports trusts also contribute to this factor. These are the two organisations that cross into more than one factor which would seem appropriate since, while they have overlaps with charities and other official bodies they also fall outside the official ‘health industry’ sources. Both can act as a source of funding to support local community physical activity initiatives.
Factor 3 is plainly related to official health sources and, apart from the small cross-loading by pharmacists on this factor all the other contributing variables are official health bodies who could contact people outside a normal primary care situation. Conversely Factor 4 is centred on the primary personal professional health contacts experienced by most people: doctors and doctor’s nurses. Pharmacists, and local hospitals also contribute to this factor.
Table 3. Rotated factor loadings from the exploratory factor analysis
Factor 5 is linked to alternative or ‘non-medical’ personal HISs and the final factor constitutes family and friends as an independent grouping. Thus the factor analysis suggests that, in terms of trust, the New Zealand public perceive six major groupings of sources for health information.
In Table 4 below we report associations between these trust factors and a number of demographic characteristics. All the results reported below are statistically significant at p ≤0.001 though, based on partial eta squared, effect sizes would all be regarded as small.
Table 4: Variations in trust factors by demographics
The variations across demographics reveal some interesting patterns though immediate explanations for all the relationships are not obvious. Why females exhibit more trust than males is not absolutely understood although it has been recorded in other contexts such as the internet and trust games in experimental economics20,21. The links with work status are the weakest of all those examined and possibly the least useful from a policy perspective. A possible explanation for lower trust in alternative health professionals by those identifying themselves as sick or invalid may simply be lower levels of contact with some of these sources – for example sports organisations and gym trainers.
The two sources that retired people rate more highly than other work status groups are both personal sources as opposed to other the factors that contain at least some impersonal items. The associations with age, income and education all follow the same pattern with media and alternative sources being less trusted as all three increase. Intuitively this seems sensible and as would be expected. Ethnicity is rather more complex. Two of the groupings in the survey are composite groups—‘British/European’ and ‘other Asian’ but both still show some significant variations across the factors. Overall the biggest differences are found between the different Asian groups and the rest of the sample.
While these groups still trust personal health professionals such as doctors and nurses more than other sources for HIS they do express more trust in the media and alternative health professionals. Arguably the latter may be a feature of a wider view of medicine that is sometimes attributed to Asian countries. An interesting finding is the difference in trust accorded to friends and family.
In New Zealand recognition and involvement of whānau (extended family) has been a significant issue in relation to Māori in recent years. Our data suggest that this is not just a feature of Māori but more a difference between those of European heritage and all other ethnic groupings. It is possible that the latter groupings reflect cultures that pay more attention to the extended family and are less individualistic in their value systems than European, especially Anglo-Saxon, cultures.
The purpose of this paper is to investigate how the source of information used to about health might affect the trust that people have in the information. As such it differs from most of the work in trust and health information that manipulates an individual message and identifies the effects and interactions of the different message components. Trust was assessed by single statements that asked people to judge sources at a general level. While this approach does not capture variation that exists within any of the categories—for example some radio programs on science may be more trusted than information coming from a radio talkback show—it does allow for comparisons across a wide variety of media types and the results of the factor analysis suggest that the data is capturing systematic variations in a reliable way.
While some cross-loading is evident, the factor solution is quite clear and each factor has a distinctive set of sources. Not surprisingly, health professionals with whom people have personal contact are the most trusted sources across the whole spectrum. But clearly not all information can be delivered through these channels.
For many of the population their contact with these health professionals is sporadic and usually motivated by some specific need which may be far removed from an information message that policy makers or higher level planners want to put across to the population. Therefore it becomes important to understand the mix of media that might be required in order to effectively communicate trustworthy health information and it is clear that the optimum mix varies across the population. Choice of channels could well be as important as the message itself.
Competing interests: None.
Author information: Rob Lawson, Professor, Department of Marketing, School of Business, University of Otago, Dunedin; Sarah Forbes, PhD Candidate, Department of Marketing, School of Business, University of Otago, Dunedin; John Williams, Lecturer, Department of Marketing, School of Business, University of Otago, Dunedin
Correspondence: Professor Rob Lawson, Department of Marketing, School of Business, University of Otago, PO Box 56, Dunedin, New Zealand. Email: firstname.lastname@example.org
issue | Search journal |
Archived issues | Classifieds
| Hotline (free ads)
Subscribe | Contribute | Advertise | Contact Us | Copyright | Other Journals