There is a certain amount of uncertainty (error) associated with every experiment. No measurement is completely free of error. The “true” value of any measure is always unknown. A compromise between the data generated and the time spent has to be made. There are two types of errors called sampling errors and non-sampling errors. Sampling error caused by observing the sample instead of the whole population. Non-sampling errors can occur in all stages due to human errors in data collection, processing, estimation methods and reporting. The errors have to be controlled or else we will end up with wrong results.
Possibilities of Errors in Research Studies in Different Stages:
Planning of the study Collection and maintenance of data Design of the study Execution of the study Data processing Analysis and interpretation of data Presentation of results Reporting and Publications These are few possible events, which causes these errors. Failure to state the problem carefully Failure of the techniques adopted Failure in experimental design and sampling techniques used Errors in processing and classification Errors in response Non-random selections Absence of a standardized procedure Inter and intra Investigator bias Outliers Inferential statistics is the use of statistics to make inference about the population of interest based on a random sample drawn from it. Test of hypothesis help us to make statistical decisions with the help of experimental data.. Significance tests are based on certain assumptions. The data should be random samples from a well-defined basic population and one has to assume that some variables follow a certain distribution – in most cases the normal distribution is assumed.
Statistics helps to:
Quantify the errors/variation and to trace the factors associated with variation. The pitfalls in Planning, execution and analysis of experiments can be described. Statistical Methods are used for description of the findings Statistical methods can be utilized to find the significance of differences between results of studies. It is possible to have checks/study of variations in methods of study in estimation between investigators and between time points
How to Quantify Errors:
In statistics, an alternative word, which can be used in place of error, is variation. How statistical methods are useful for quantification of errors/variation? Measures of central tendencies and measures of dispersion such as Mean, Standard deviation, standard error, measures of skewness are utilized to summarize the errors in measures. The causative factors such as sampling methods, sample size, estimation methods, and quality control methods influences the error distribution. Outliers are a few observations that are not well fitted by the “best” available model and Outliers can errors in the estimates and can be misleading. In practice any observation standardized with residual greater than 2.5 in absolute value is a candidate for being an outlier. In such case one must first investigate the source of data, if there is no doubt about the accuracy of the observation, then it should be removed and the model should be refitted.
These are the few tips to Control Errors and Variation:
- Proper planning of the study
- Use of appropriate statistical design and methods for study
- Appropriate sampling methods
- Reduce investigator bias
- Training for investigators
- Use appropriate and large sample size with consideration of power for the calculation
- Use of Standardized procedures
- Appropriate statistical Techniques
- Computer software for data processing and statistical analysis
Correct interpretation of the study results