Loading...
Please wait, while we are loading the content...
Similar Documents
Spin and Boasting in Research Articles S
| Content Provider | Semantic Scholar |
|---|---|
| Abstract | S OME AUTHORS EXAGgerate the importance of their research and unfairly denigrate other studies. This occurs only in a minority of articles we review but is frequent enough that we have collected examples and grouped them into categories. We confess to committing some of these literary sins ourselves; improving writing is a lifelong process. Examples include: " [X] has reached an alarming proportion, " " an awe-inspiring number, " and " drastic increase. " These feverish words lack scientific meaning. Instead, give relevant quantities for a particular place and time: counts of events, incidence rates, or prevalence estimates. Consider describing how these quantities have changed over time. Let the reader decide whether anything is alarming, awesome, or drastic. Writing that something is " a critical priority, " even if true in some sense, is just an editorial opinion unlikely to persuade. Referring to " the obesity epidemic " is a cliché. Hackneyed phrases do not make the writer appear thoughtful, are boring for the reader, and take up space. Consider whether the reader needs to once again hear that obesity is common, diabetes is increasing, and that the cost of medical care is a problem. We think not. Researchers should use a minimum of adjectives to describe their topic. To promote their work, authors have declared a " crisis of credibility " and a " large research gap. " Some dismiss past studies for the following specious reasons: v " Inconsistent " or " mixed " results: These fuzzy words explain nothing. We do studies in finite samples, so some difference in results is almost inevitable. If a statistical test of homogeneity indicates the difference is more than expected , a new study is unlikely to change that. Consider meta-analytic methods to describe the variation and seek explanations, such as dissimilar study populations or designs. v " Methodologic flaws " : These words are not helpful because essentially all studies have flaws. The new study will have its own flaws. If there is evidence that a prior study is biased, by all means show this. A less than optimal design is not proof of bias. v " Small sample size " : Some small studies provide useful evidence. The first report that diethylstilbestrol taken by pregnant women was related to vaginal cancer in their offspring had only 8 cases and 32 controls. 1 If all else is equal, a larger … |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://tarrlabwiki.cnbc.cmu.edu/images/6/61/Spin.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
