How much can you self-plagiarize

[BillKet]

Hello! I am writing a paper resulted from a given experiment, but I already published a paper before from the same experiment. The topic of the papers (the obtained results) are different (hence why there are 2 papers), but the introduction and especially the methods sections are basically identical (it is what we did with the data that is different). Can I just copy paste parts from the old paper, or slightly rephrase them, or is that considered plagiarism. And if so, how should I proceed? The experimental setup is absolutely identical in both cases. How should I proceed?

 

[Dale]

You need to cite yourself to avoid plagiarism, just the same as if you had replicated someone else’s experiment. You also need to explicitly identify that you are re-analyzing data that was already published. This has statistical consequences.

 

[BillKet]

You need to cite yourself to avoid plagiarism, just the same as if you had replicated someone else’s experiment. You also need to explicitly identify that you are re-analyzing data that was already published. This has statistical consequences.

Thank you for your reply! Yes, I will cite myself. My questions was more along the lines, do I need to rephrase certain paragraphs (that basically have exactly the same content) or can I just copy and paste them (with the right citation to myself)? Actually I am not re-analyzing data. This data is from the same experiment, but it was not related to the first analysis.

 

[Dale]

Actually I am not re-analyzing data. This data is from the same experiment, but it was not related to the first analysis.

Ah, ok. Then you don’t need to do more than cite yourself.

do I need to rephrase certain paragraphs (that basically have exactly the same content) or can I just copy and paste them (with the right citation to myself)?

That is up to you. I would probably not do either in favor of simply referring to the technique by reference. In other words, I would only summarize the experiment with all details done by reference, similar to how I would cite someone else whose experiment I were replicating. But that is up to you.

 

[DaveC426913]

(it is what we did with the data that is different).

Actually I am not re-analyzing data. This data is from the same experiment, but it was not related to the first analysis.

Ah, ok. Then you don’t need to do more than cite yourself.

Sorry, In your OP you say you "did something different with the data", yet you then say "Actually I am not re-analyzing data."
That seems contradictory at first glance. @Dale, do you not think so?

 

[Dale]

That seems contradictory at first glance. @Dale, do you not think so?

It did confuse me, at least.

For clarity, suppose that in your experiment you collected 100 pieces of data and analyzed 75 of them with some technique. Then if you are analyzing the remaining 25 pieces of data there is little concern (except if you selected those 25 systematically from the whole sample). If you are analyzing some of the 75 pieces of data then it is a statistical consent even if the analysis is completely different.

 

[BillKet]

It did confuse me, at least.

For clarity, suppose that in your experiment you collected 100 pieces of data and analyzed 75 of them with some technique. Then if you are analyzing the remaining 25 pieces of data there is little concern (except if you selected those 25 systematically from the whole sample). If you are analyzing some of the 75 pieces of data then it is a statistical consent even if the analysis is completely different.

@DaveC426913 @Dale I am using data from the same experiment, but the 2 data sets are independent. For example, from a molecular spectroscopy experiment, one paper would be about the lifetime of certain low laying states, while the second one would be about the ionization potential. The data was all taken within the same experiment (of course the wavelength of the laser was adjusted, but everything else was the same), but the 2 data sets are completely different and you can't use one data set to statistically improve the other.

 

[Dale]

you can't use one data set to statistically improve the other

That isn’t the relevant statistical issue, but from what you describe the data should be fine. From what you describe it sounds more like two experiments run in parallel.

The issue is about multiple comparisons. If you conduct further analysis of the same data then you increase the p-value of your previously published study. (You also increase the p-value if you perform additional analysis that you don’t report)