Suppose I have two variables, A and B measured for the same participants multiple times. If at one time, the participant's Variable A was an outlier and it was removed, should I also remove this point for variable B.
For example:
Variable A; Response Time, measured 15 times (T1-T15)
Variable B: also measured in the 15 times (T1-T15)
If Participant 1 had for response time (T2) an outlier, and I removed the outlier, should I also remove participant 1 T2's value for variable B ?
You either remove an observation or you don't. You can't remove half of an observation. Look at multivariable data not as just one variable at a time. Google outlier detection procedures for a full discussion..There are many references available. Best wishes, David Booth
Hi Habiba Ajour
Outliers are data points and before throwing them away you should discover why are different: gross error, special trend, etc.
In this case, you have to deal with the outliers using a multivariate approach. Checking the variables one by one is not the right way.
You can create a boxplot to identify your outliers and find the following
The IQR method to find outliers by R CODE the quantile() function to find the 25th and the 75th percentile. The IQR function also requires numerical vectors.
Eliminating Outliers BY R Using the subset() function, you can simply extract the part of your dataset between the upper and lower ranges leaving out the outliers.
Then The boxplot without outliers can be found.
- Badges
- Science method
- Similar topics
- Methodology
- Research Methodology
- Qualitative Research