What is data cleaning?

What is data cleaning?

• A. The removal of parts of data that are likely caused by confounding variables (e.g., false starts or fatigue on reaction times)
• B. The process of scaling data to a range of 0 to 1
• C. The method of imputing missing values with the median
• D. A process to randomize data prior to analysis

Answer: (A)

Data cleaning is about improving data quality by removing or correcting data points that are contaminated by errors or artifacts, so analyses reflect the true signal rather than noise. In a reaction-time study, trials affected by confounds like false starts or fatigue can distort the results, so excluding those trials is a classic data-cleaning step to reduce bias from unintended factors. Other options describe different preprocessing tasks: scaling to a 0–1 range is normalization, not cleaning; imputing missing values is data imputation; randomizing data is used to control bias in design or analysis. While these can be parts of preparing data, the act of removing data points caused by confounds best captures what data cleaning entails.

Data cleaning is about improving data quality by removing or correcting data points that are contaminated by errors or artifacts, so analyses reflect the true signal rather than noise. In a reaction-time study, trials affected by confounds like false starts or fatigue can distort the results, so excluding those trials is a classic data-cleaning step to reduce bias from unintended factors. Other options describe different preprocessing tasks: scaling to a 0–1 range is normalization, not cleaning; imputing missing values is data imputation; randomizing data is used to control bias in design or analysis. While these can be parts of preparing data, the act of removing data points caused by confounds best captures what data cleaning entails.