Difference between revisions of "Data analysis: Recommendations"
Line 1: | Line 1: | ||
[[Responsibility: Recommendations|'''Define''']] '''who''' will perform the data analysis. | [[Responsibility: Recommendations|'''Define''']] '''who''' will perform the data analysis. | ||
− | Provide guidance/training how to analyse data (+statistical analysis). Perform blinded analysis if possible. | + | Provide guidance/training how to analyse data (+statistical analysis). Perform [https://eqipd-toolbox.paasp.net/wiki/2.1.7_Blinding blinded] analysis if possible. |
[[Responsibility: Recommendations|'''Define''']] '''who''' will check the data analysis. A [[Quality checkpoints: Recommendations|'''data analysis checkpoint''']] is important to avoid misinterpretation and bias. If the user analyses the data on its own, decide who will check/supervise the analysis (CF staff or PI or specialist). (see [[Responsibility: Recommendations|responsibility]]) | [[Responsibility: Recommendations|'''Define''']] '''who''' will check the data analysis. A [[Quality checkpoints: Recommendations|'''data analysis checkpoint''']] is important to avoid misinterpretation and bias. If the user analyses the data on its own, decide who will check/supervise the analysis (CF staff or PI or specialist). (see [[Responsibility: Recommendations|responsibility]]) | ||
Analysed data should never override raw data. Do not forget to link the analysed data to the raw data, so that they are traceable. | Analysed data should never override raw data. Do not forget to link the analysed data to the raw data, so that they are traceable. |
Revision as of 15:07, 26 February 2021
Define who will perform the data analysis.
Provide guidance/training how to analyse data (+statistical analysis). Perform blinded analysis if possible.
Define who will check the data analysis. A data analysis checkpoint is important to avoid misinterpretation and bias. If the user analyses the data on its own, decide who will check/supervise the analysis (CF staff or PI or specialist). (see responsibility)
Analysed data should never override raw data. Do not forget to link the analysed data to the raw data, so that they are traceable.