Generally we encounter the problem while planning design of experiment for any research trial or experiment regarding number of experimental units to be included in each group? as a regular practice for medicinal or nutritional trials in large animals we consider 6 per group as sufficient however i want to know is there any standards to decide these numbers or any correlation studies for effective number of experimental units and accuracy of result...?
When we talk about breeding or genetic study we know that large number of units are desirable for higher accuracy...but what about other trials please...?
Important question. I think the number is correlated with the standard error and consequently with statistical significance
Dear Ali, it is important to calculate the sample size for each experiment. The sample cannot be arbitrarily determined.
I suggest that you use one of these programs to calculate the sample size:
G*Power or PS: Power and Sample Size
Thank you for the question I think that the size of the research sample is selected in experimental research so that it bears the characteristics of the original society and represents the research community
sometimes it depends on the availability of material, but at least triplicate
Thanks All...That means there is no specific guidelines.. and as we follow six animals is sufficiently good....my doubt was when we conduct nutritional study in poultry we took in thousands where as in case of large animals we feed only 4-6 animals...?
Dear Sajd Ali,
I tried to help writing that there are specific guidelines for calculating the power of the study according to the sample size. The science that makes such calculations is called STATISTICS. It has long been calculated that the most common number of animals in experimental research is around 6, and researchers use it as valid data. As you know, statistics are an inseparable part of every research and later article, so it is necessary that before each research you count the number of animals for your experiment.
Good luck!
The statistics can be very clear - IF the laboratory director sets MEDICAL goals and Acceptable Risk Criteria.
This is a link to a series of posts on LInked In https://www.linkedin.com/in/zoe-brooks-risk-mgmt-for-clinical-labs-qc-047a4b26/detail/recent-activity/shares/
It describes and discusses the recommendations for risk management from ISO, CLIA and CLSI and each post links to a survey that I would appreciate you taking!
We will be modeling this new process in January and I am looking for collaborators to examine and submit data. The stats require a computer, but the science is crystal clear and removes almost all opportunities for human error.
It's not straightforward. Your objective in the experiment is not clear, either. I suggest you consult a sampling statistician or give summary of your objectives. Statisticians on the platform might help.
Thanks for the suggestion. These informal Linked-In Posts are not intended to do that yet. We are discussing each of the best practice recommendations. There will be a new post tomorrow describing the full 4-part process. There are 2 surveys and then 2 webinars.
Webinar 1 Jan 10th:
Learning Objectives - This Best Practice Webinar will help you reduce risk by:
1. auditing if your medical laboratory is following best practice for Risk Management in Statistical QC.
2. challenging traditional statistical assumptions relative to recommended best practice.
3. considering the potential value of updating traditional statistical QC to validated best practice for risk management.
We will be collecting specific risk drivers that drive the number and cost of Medically-Unreliable Results so researchers like you can see the impact.
I am inviting statisticians or researchers interested in this topic to collaborate. Are you interested in reviewing data with our team?
Dear Sajid Ali,
Indeed, there are some rules for determining the number of repetitions (experimental units: EU) that depend on the objective of the research work. However, available means (budget, experimental material, etc.) is limited for most of the time. Thus, the number of EU is also limited and this will have an impact of the achieved accuracy. So, the experimenter should look for a balance between accuracy and number of EU. In statistics, you may have 2 different objectives: either estimate a population parameter or test a given hypothesis. For each one of these 2 situations, you may have many cases. For example, for estimation, you may be interested in a mean, a proportion, etc. For hypothesis testing, you may do a conformity test: is the population mean equal to a given constant, you may compare 2 or more means, 2 or more proportions, etc. To determine the number of EU, you should consider, in addition to the objective and the accuracy, the levels of the types I and II errors (alpha and beta). There are many formulas that are given as functions of these 4 factors (objective, accuracy, alpha and beta or 1-beta as the statistical power).
- Badges
- Science topic
- Similar topics
- Chemistry, Organic
- Organic