1
$\begingroup$

I'm reviewing research proposals. The proposal is, in human eyes, practically the same as a 2017 degree thesis. The thesis guide professor is the principal investigator of the current project. I used an online site (https://copyleaks.com/) to verify the amount of similar text and it indicates 18%. I would like to add as many evidence to my decision not only to reject but also raise an expression of concern for plagiarism.

My question is: would it be correct to apply a proportion test to indicate that the proportion found, of 18% of equal text in an 18000-word document, is scarcely attributable to chance?

I used an online calculator for the calculation, available in https://www.medcalc.org/calc/test_one_proportion.php with these data enter image description here

$\endgroup$
5
  • $\begingroup$ Why did you choose $p=0.05$ as null hypothesis value? That seems artificial. I think this is a case where you need to compare with an "empirical null", you need to find what is the typical percentage of similar text between such documents in a class of documents relevant for comparison. Say, if this is a lab field, it might be natural to copy some sections such as "methods and materials" and maybe others. $\endgroup$ Jun 11, 2019 at 14:30
  • $\begingroup$ Thank you David, it seems that between 10 to 20% would be acceptable, with most opinions towards a maximum of 10%. See researchgate.net/post/… $\endgroup$ Jun 11, 2019 at 14:31
  • 1
    $\begingroup$ @sergiouribe If 20% is still considered acceptable, then 18% is too. No statistics needed. But the main problem I have is the reliability of that index (after all, it's summarizing into a single number some phenomenon that is way more complex than that) $\endgroup$
    – David
    Jun 11, 2019 at 14:33
  • $\begingroup$ Indeed @kjetilbhalvorsen . It's arbitrary and that's part of my question. I could consider that it is natural for a lab to continue with a line of research and to base projects based on previous work, but in the project the previous thesis is not cited. The project is essentially the thesis but with a major n. $\endgroup$ Jun 11, 2019 at 14:34
  • 1
    $\begingroup$ @sergiouribe: * the previous thesis is not cited. The project is essentially the thesis but with a major n.* (something was cut off there) This is the main argument then, and not the hypothesis test. For that you would need to gather data as I indicated. $\endgroup$ Jun 11, 2019 at 15:49

1 Answer 1

0
$\begingroup$

In order to determine whether 18% is too big or not, it would be needed to have a precise definition of how that index is calculated. Fortunatelly, we have a calculator that does it for us, but I don't know to what extent the source should be trusted (i.e: more detail on how the calculation was done is required).

Assuming the p-value shown there is accurate, there is very strong evidence that it did not happen by chance, but an analysis on the methods used is of course critical (both for the index calculation and its conversion to a p-value)

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.