Analytics
What 514,629 hedged sentences look like in aggregate. The numbers below count sentences, not papers, because a paper that hedges four times is doing something four times.
What the corpus contains
Everything the pipeline has collected, including what the rest of this page leaves out. The figures elsewhere count specimens only.
| Kind of phrase | Specimens | Plainly factual | Everyday English | Undecidable | Total |
|---|---|---|---|---|---|
| Hedge (nearly significant) | 385,328 | 0 | 12 | 0 | 385,340 |
| Qualified (highly significant) | 132,691 | 0 | 23,164 | 0 | 155,855 |
| Plain (did not reach significance) | 1,611 | 36,740 | 0 | 89,000 | 127,351 |
These proportions are not a prevalence estimate. This is a phrase-matched collection, not a sample of the literature: it finds only the wordings on its list, so a ratio computed here describes how the collected sentences divide and says nothing about how common this writing is in research at large. Measuring that would need a representative sample of papers, which is what Otte et al. (2022) did for randomised trials, finding these phrases in 8.65% of them.
The distribution
40.8% of the p-values quoted here fall in the narrow strip between 0.05 and 0.1. That strip is 5% of the probability scale.
Hedging over time
Bars are sentences per year; the line is the median p-value they reported. Note that the upward slope of the bars is mostly the growth of open-access publishing, not a decline in scientific virtue.
The leaderboard
Ranked by how often each phrasing appears. “Worst offender” is the largest p-value any paper described with that phrase, which is the single best measure of nerve.
| Phrase | Sentences | Median p | Closest | Worst offender |
|---|---|---|---|---|
| highly significantgold | 132,143 | 0.001 | 0.0000 | 1.000 |
| did not reach statistical significance | 111,028 | 0.045 | 0.0000 | 0.050 |
| an increasing trendgold | 63,972 | 0.050 | 0.0000 | 0.996 |
| showed a trend | 53,323 | 0.070 | 0.0000 | 1.000 |
| a decreasing trendgold | 47,181 | 0.050 | 0.0000 | 1.000 |
| marginally significantgold | 26,083 | 0.060 | 0.0000 | 1.000 |
| a clear trendgold | 19,461 | 0.057 | 0.0000 | 0.990 |
| may be significant | 13,098 | 0.050 | 0.0000 | 0.500 |
| borderline significantgold | 11,410 | 0.056 | 0.0000 | 1.000 |
| a positive trendgold | 11,006 | 0.061 | 0.0000 | 0.922 |
| a significant trendgold | 9,775 | 0.012 | 0.0000 | 1.000 |
| approached significance | 8,238 | 0.061 | 0.0000 | 1.000 |
| extremely significant | 7,788 | 0.010 | 0.0000 | 1.000 |
| nominally significantgold | 7,733 | 0.050 | 0.0000 | 0.743 |
| borderline significance | 7,018 | 0.057 | 0.0000 | 1.000 |
| failed to reach statistical significancegold | 6,036 | 0.046 | 0.0000 | 0.050 |
| a strong trendgold | 5,726 | 0.060 | 0.0000 | 1.000 |
| quite significantgold | 5,696 | 0.005 | 0.0000 | 0.200 |
| a nonsignificant trendgold | 5,407 | 0.080 | 0.0001 | 1.000 |
| only marginally significantgold | 5,077 | 0.062 | 0.0001 | 0.766 |
| marginal significancegold | 5,069 | 0.060 | 0.0000 | 1.000 |
| an overall trendgold | 4,872 | 0.055 | 0.0000 | 0.570 |
| a negative trendgold | 4,202 | 0.060 | 0.0000 | 0.979 |
| failed to reach significance | 4,114 | 0.030 | 0.0000 | 0.050 |
| a slight trendgold | 4,079 | 0.090 | 0.0001 | 1.000 |
| a trend toward significance | 3,730 | 0.068 | 0.0000 | 0.902 |
| nearly significantgold | 3,288 | 0.059 | 0.0000 | 0.890 |
| a statistical trendgold | 2,799 | 0.070 | 0.0010 | 1.000 |
| may not be significantgold | 2,775 | 0.029 | 0.0070 | 0.050 |
| a trend towards significance | 2,761 | 0.069 | 0.0000 | 0.910 |
By journal
Journals with at least 20 hedged sentences. A journal with one is not the most hedging journal in science, so the floor is there on purpose. Volume mostly tracks how much a journal publishes.
| Journal | Sentences | Papers | Median p |
|---|---|---|---|
| PLoS One | 33,674 | 29,769 | 0.050 |
| Sci Rep | 27,042 | 23,648 | 0.050 |
| Int J Mol Sci | 8,471 | 7,559 | 0.050 |
| Int J Environ Res Public Health | 7,015 | 6,022 | 0.053 |
| Front Psychol | 6,395 | 5,644 | 0.059 |
| Nat Commun | 5,161 | 4,700 | 0.020 |
| J Clin Med | 5,093 | 4,468 | 0.052 |
| Nutrients | 4,906 | 4,279 | 0.054 |
| Cureus | 4,691 | 4,289 | 0.010 |
| Front Plant Sci | 4,613 | 3,898 | 0.010 |
| Materials (Basel) | 4,558 | 4,069 | 0.002 |
| Foods | 4,228 | 3,548 | 0.010 |
| Front Immunol | 3,862 | 3,472 | 0.050 |
| Heliyon | 3,820 | 3,313 | 0.010 |
| Cancers (Basel) | 3,680 | 3,257 | 0.055 |
| Plants (Basel) | 3,679 | 3,050 | 0.010 |
| Animals (Basel) | 3,468 | 2,995 | 0.050 |
| BMC Public Health | 3,445 | 2,982 | 0.050 |
| Molecules | 3,345 | 3,058 | 0.010 |
| Front Microbiol | 3,343 | 2,920 | 0.010 |
Most persistent papers
Papers that hedge the most times in a single manuscript.