Independence of observations
Independence of observations means each participant is only counted as one observation
The statistical assumption of independence of observations stipulates that all participants in a sample are only counted once. If a participant was to appear multiple times in a sample, each time as an independent observation, then the statistics would be artificially skewed in their favor and not be representative of a true sample of independent participants. Independence of observations makes sure that each participant's variance affects the overall analysis just once.
Statistical designs and independence of observations
Between-subjects designs can be grossly affected by a violation of this assumption. Comparing groups that are artificially weighted towards ONE PARTICIPANT counted multiple times in one or both groups yields false inferences.
In within-subjects designs, the participants are independent of each other and there are several observations of the outcome "within" each person or across time.
Multivariate analyses that account for demographic, etiological, prognostic, confounding, or clinical variables cannot meet the statistical assumptions of normality and linearity if there are multiple observations of the same participant in a sample. This is especially true if the aforementioned variables are measured at a categorical level.
There is no statistical test associated with testing independence of observations. It is up to you, as the researcher, to meet this assumption by counting each participant in your study as one independent observation.
In within-subjects designs, the participants are independent of each other and there are several observations of the outcome "within" each person or across time.
Multivariate analyses that account for demographic, etiological, prognostic, confounding, or clinical variables cannot meet the statistical assumptions of normality and linearity if there are multiple observations of the same participant in a sample. This is especially true if the aforementioned variables are measured at a categorical level.
There is no statistical test associated with testing independence of observations. It is up to you, as the researcher, to meet this assumption by counting each participant in your study as one independent observation.
Statistician For Hire
DO YOU NEED TO HIRE A STATISTICIAN?
Eric Heidel, Ph.D. will provide statistical consulting for your research study at $100/hour. Secure checkout is available with PayPal, Stripe, Venmo, and Zelle.
- Statistical Analysis
- Sample Size Calculations
- Diagnostic Testing and Epidemiological Calculations
- Psychometrics