Plain English Summary
Background and study aims
Recent research has shown that up to 40% of health-related press releases contain exaggerated statements about the findings of the research papers they are about. It has also been found that this exaggeration is spread to later news articles which are reporting on the press releases. There is no real evidence that press releases containing exaggerated statements were more likely to be reported on in the news than those which do not contain exaggerated statements. News exaggeration can have widespread effects for public health as mainstream news, such as television or newspapers, is where the general public find out about science and health. It has been suggested that the exaggerated statements in the news come from the exaggerated statements in the press releases. This study aims to find out whether changing the wording of press releases has an effect of news coverage for health-related research. The study also aims to try to improve the communication between academic research and news content, without affecting how much of the research is reported in the news.
Who can participate?
UK-based press offices that publish press releases on health-related research.
What does the study involve?
Press officers and academic authors write press releases in the usual way, until they are ready for to be published. The press release is then sent to the InSciOut team and is be randomly assigned to one of four groups. Suggestions for changes are then made for the press releases in each group. For the first group, casual statements are changed so that they reflect the design of the study in the paper that is being written about. For the second group, information about the design of the study is included in the press release itself, to show how strong the casual conclusions are that are drawn. For the third group, both the changes for group 1 and 2 is suggested. The final group acts as a control group, involves only minor changes to words which have nothing to do with the main casual statements or study design, e.g. ‘’fizzy water” may be changed to “sparkling water”. Following the suggested modifications for each group, the press releases are sent back to the press office for a final decision about whether the changes are applied when the press release is published. Suggested modifications depend on the type of study as described in the journal article itself. The InSciOut team monitor the news articles that are released to find out how many have kept the suggested changes.
What are the possible benefits and risks of participating?
Benefits of participating include being able to build a rich evidence base of ‘what works’ in terms of accuracy, prominence and quantity of news coverage. There is a possible risk that the changes suggested could reduce news uptake, however this will be monitored throughout.
Where is the study run from?
Cardiff University (UK)
When is the study starting and how long is it expected to run for?
December 2014 to November 2016
Who is funding the study?
Economic and Social Research Council (UK)
Who is the main contact?
Dr Rachel Adams
Dr Rachel Adams
School of Psychology
College of Biomedical & Life Sciences
+44 2902 870708
A multi-armed randomised controlled trial of modifying causal claims and adding study design information to health-related press releases on news coverage
ModPress (modifying press releases)
Modifying causal claims and adding study design information to health-related press releases will increase accuracy of subsequent news coverage relative to the corresponding journal article without reducing news uptake.
Cardiff University Ethics Committee, 23/03/2015, ref: EC.15.02.10.4099
Interventional multi-centre randomised controlled trial
Primary study design
Secondary study design
Randomised controlled trial
Patient information sheet
Not available in web format, please use the contact details below to request a patient information sheet.
Misrepresentation and exaggeration of findings from health-related research in press releases and news coverage.
Press officers will send draft copies of press releases to the InSciOut team prior to their release. The InSciOut team will code the press releases and randomly assign them to either an intervention condition or the control condition. Suggested modifications will be made to the press release, based on these conditions and the corresponding journal article, and will be returned to the press officer prior to public release.
Health-related press releases will be randomly assigned to one of four conditions:
Group 1: Causal statement only condition: Causal statements within the press release will be altered to correspond to the study design described in the journal article. For example, if the title of a press release reported that ‘Alcohol improves social skills’ when the study was a correlational piece of research, the title would be changed to read ‘Alcohol may improve social skills’. Modification of causal claims will always increase correspondence between the journal article and the press release.
Group 2: Design information only condition: Information regarding the design of the study will be included in the main body of the press release to convey the strength of causal conclusions that can be drawn (e.g. correlation vs. randomised controlled trial). For example, if a press release reported a causal relationship between the IV and DV when the study was correlational, a suggested change could be to include the following statement: “This was an observational study which does not allow us to conclude that alcohol causes increased social skills as other factors could be involved. We would need to run an experiment to get causal evidence”. For randomised controlled trials a suggested change might be “This study was a randomised controlled trial which allows us to draw firm conclusions about cause and effect”.
Group 3: Causal statement + design information condition: Both manipulations, as described above, will be implemented.
Group 4: The control condition: Involve a synonym change to a word that is not relevant to the main causal statements or design of the research, for example, ‘’fizzy water” may be changed to “sparkling water”.
With the exception of the manipulations above all other aspects are equivalent across the four trial arms.
Intervention fidelity will be examined by coding press releases following modifications by the InSciOut team (i.e. before sending the press release back to the press officer) and again after public release. Suggested changes made by the InSciOut Team will be coded for their presence/ absence and for any edits made in the publicly issued press release.
Primary outcome measures
News coverage will be monitored for the week prior to the press release date and for one month following release.
News coverage will be coded using a standardised protocol adapted from Sumner et al. (2014). A researcher blind to the intervention condition and published content of the press release will code the news stories.
1. The number of news stories that contain accurate causal claims, relative to the corresponding journal article, and the number of news stories that contain study design information.
2.The second primary outcome measure is news uptake, defined as the number and length of news articles generated by the press release (print, online and broadcast news).
Secondary outcome measures
1. Secondary outcome measures include two other forms of exaggeration previously found in news articles (Sumner et al., 2014):
1.1. Advice – measured by the number of news stories that contain an in/appropriate level of advice (i.e. no advice, explicit advice not to the reader or general public or explicit advice to the reader or general public
1.2. Human inference – measured by the number of news stories that contain in/accurate information regarding the study sample (i.e. human or non-human participants)
2. The feasibility and acceptability of the trial. These outcomes will be determined by challenges to implementation of the wording intervention and participation in the trial.
Overall trial start date
Overall trial end date
Participant inclusion criteria
Inclusion criteria for press releases:
1. Involves empirical, peer-reviewed research that is directly relevant to human health. Empirical evidence to include surveys, meta-analyses, systematic reviews and case studies.
Inclusion criteria for press officers (from any of the following):
1. British university for topics related to human health e.g. clinical medicine, psychology, psychiatry and neuroscience, public health, health services and primary care
2. British research councils
3. British-based academic journals
4. British charities
Target number of participants
The aim is achieve 100% coverage for all eligible press releases across all participating institutions. The absolute number of press releases included in the trial will depend on the number of institutions who agree to take part. We estimate that we will receive between 300-500 press releases. With 300 press releases our analyses are sensitive to detect a minimum effect size of w =0.187 with 90% power (α =0.05, df =1). For 400 press releases we can detect a minimum effect size of w =0.162, and for 500 press releases we can detect a minimum effect size of w =0.145. Our previous data for exaggeration of main statements from correlational to causal phrases have revealed effect sizes of 0.33 and 0.24 for Russell Group University and Journal press offices, respectively.
Participant exclusion criteria
Exclusion criteria for press releases:
1. Involves empirical, peer-reviewed research that is not directly relevant to human health (e.g. climate change research, astronomy, palaeontology).
2. Do not involve empirical, peer-reviewed research
3. Related to future research
4. Related to grant funding
5. Related to literature reviews, opinion pieces, editorials or commentary
Exclusion criteria for Press Officers:
1. Those from non-UK press offices
2. Those from private/corporate press offices
Recruitment start date
Recruitment end date
Countries of recruitment
Trial participating centre
Tower building School of Psychology College of Biomedical & Life Sciences
School of Psychology
Economic and Social Research Council
Funding Body Type
Funding Body Subtype
Results and Publications
Publication and dissemination plan
The publication date will depend on the trial and other related projects; I imagine this will be between Autumn 2016 - Spring 2017.
Intention to publish date
Participant level data
Stored in repository
Results - basic reporting