Q 542. There are four different types of randomization in statistics - Simple, Blocked, Stratified and Adaptive. What is the difference between these? What is the criteria to use either of them? Provide examples to support your answer.

 

Note for website visitors - Two questions are asked every week on this platform. One on Tuesday and the other on Friday.

• All questions so far can be seen here - https://www.benchmarksixsigma.com/forum/lean-six-sigma-business-excellence-questions/
• Please visit the forum home page at https://www.benchmarksixsigma.com/forum/ to respond to the latest question open till the next Tuesday/ Friday evening 5 PM as per Indian Standard Time. Questions launched on Tuesdays are open till Friday and questions launched on Friday are open till Tuesday. 
• When you respond to this question, your answer will not be visible till it is reviewed. Only non-plagiarised (plagiarism below 5-10%) responses will be approved. If you have doubts about plagiarism, please check your answer with a plagiarism checker tool like https://smallseotools.com/plagiarism-checker/ before submitting. 
• The best answer is always shown at the top among responses and the author finds honorable mention in our Business Excellence dictionary at https://www.benchmarksixsigma.com/forum/business-excellence-dictionary-glossary/ along with the related term

Randomization is an essential tool in statistics that helps to eliminate or reduce the effects of confounding variables and biases in a study. There are four main types of randomization in statistics: simple randomization, blocked randomization, stratified randomization, and adaptive randomization.

 

Here's a breakdown of each type and when to use them:

 

Simple Randomization:
Simple randomization is the most basic form of randomization. It involves randomly assigning subjects to different groups or treatments without any additional criteria. This is often done by using a random number generator or a coin flip. Simple randomization is best used when there are no confounding variables, and the sample size is small. For example, a simple randomization could be used to randomly select 50 students from a university to participate in a study.

 

Blocked Randomization:
Blocked randomization is a type of randomization that is used when there are one or more known confounding variables that may affect the outcome. In blocked randomization, subjects are grouped into blocks based on the confounding variable, and then randomly assigned to different groups or treatments within each block. Blocked randomization helps to ensure that each group or treatment is equally represented in each block. For example, if a study is looking at the effectiveness of a new medication on patients of different ages, the subjects may be blocked by age group (e.g., 20-30, 31-40, etc.) and then randomly assigned to receive the medication or a placebo within each block.

 

Stratified Randomization:
Stratified randomization is similar to blocked randomization but is used when there are multiple confounding variables that may affect the outcome. In stratified randomization, subjects are grouped into strata based on multiple confounding variables, and then randomly assigned to different groups or treatments within each stratum. This helps to ensure that each group or treatment is equally represented across all strata. For example, if a study is looking at the effectiveness of a new medication on patients of different ages and genders, the subjects may be stratified by age group and gender, and then randomly assigned to receive the medication or a placebo within each stratum.

 

Adaptive Randomization:
Adaptive randomization is a type of randomization that adjusts the assignment of subjects based on the results of previous subjects. This is often used in clinical trials to improve the efficiency of the study and reduce the number of subjects required. In adaptive randomization, the probability of assignment to each group or treatment is adjusted based on the outcomes of previous subjects. For example, if a study is looking at the effectiveness of a new medication on patients with a specific condition, and early results show that the medication is more effective in certain subgroups of patients, the probability of assignment to those subgroups may be increased to improve the efficiency of the study.

 

In conclusion, the type of randomization used in a study will depend on the nature of the study and the research question. Simple randomization is appropriate for small studies with no known confounding variables, while blocked and stratified randomization are used when there are known confounding variables that may affect the outcome. Adaptive randomization is used in clinical trials to improve the efficiency of the study and reduce the number of subjects required. It's important to choose the appropriate type of randomization to ensure the study results are reliable and valid.

Randomization is process in which the subjects are randomly assigned to a particular group and their responses to a new drug, overall experience using a new product, effectiveness of a training, etc. are recorded and then analysed further to make an informed decision that is of interest to the researcher.  They are of the following types:

Simple Randomization:

This is based on a single sequence of random assignments. In this type of randomization, from a randomly selected sample of individuals, each individual has an equal opportunity of getting assigned to a group for a particular study and this opportunity is arrived at by either a flip of a coin or by computer generated sequence.

For e.g. To study the effect of two different blood pressure pills 1and 2, they can now be administered to the group of individuals that are selected via simple randomization.

Block Randomization:

In a block randomization, a randomly selected group of individuals are first divided into different blocks of equal size and then within each block, each individual has an equal opportunity of getting assigned to a subgroup for a particular study and the researchers ensures that the subgroups within each blocks have equal number of individuals.  Is used when the study does not have the entire sample before the study starts and meaningful conclusion can be made even with a comparatively small sample size.

Stratified Randomization:

In this process a randomly selected sample of individuals are divided into subgroups based on the same traits, or peculiarities, or attributes, like economic status or level of education, pre-existing illnesses, etc known as Strata before their response to a treatment is studied.

Stratified randomization prevents imbalances between subgroups for known noise factors that influence the treatment responsiveness. As a result, stratification may prevent type I error and improve power for small trials provided the stratification factors have a considerable effect on treatment responsiveness.

Adaptive Randomization:

Adaptive randomization is a method of making changes to the probability of allocation of an individual to a group that is being studied for treatment responsiveness according to the progress and position of the study at a given point of time. The randomization scheme in this case adapts to accumulating evidence.  The first individual is assigned to a study group through a simple randomization, and the subsequent ones are assigned the study group that is producing better result. In other words, the information of the individuals who have already participated in the study is used to assign the newly recruited individual to a study group. Rather than having fixed number of individuals assigned to each treatment, there maybe an advantage to varying these over time. For e.g. if the treatment responsiveness is better in one study then we might consider assigning more individuals to that study.  This would guarantee that by the time you’ve clearly shown one treatment to be best, most individuals would’ve received that treatment.  Another example could be where we want to study the effectiveness of a training program with the two factors being self-enrolled participants and participants enrolled by their organization.  If self-enrolled participants are showing better results we may want randomly assign more number of self-enrolled participants to the study.

 

Randomization is a technique used in statistics to ensure that study participants are assigned to different groups in an unbiased and random manner. There are four types of randomization in statistics: simple randomization, blocked randomization, stratified randomization, and adaptive randomization.

1.Simple randomization:

In simple randomization, participants are randomly assigned to treatment or control groups without any constraints or restrictions. The simplest method of randomization is to use a computer-generated random number generator or a randomization table to randomly assign participants.

Criteria for using simple randomization:

Simple randomization is useful when the sample size is relatively small, and there are no other variables that need to be controlled. It is typically used when a researcher wants to investigate the impact of a single intervention, such as a new drug or therapy.

Example:

Suppose a researcher is conducting a clinical trial to investigate the effectiveness of a new drug in treating high blood pressure. The researcher could use simple randomization to randomly assign participants to either the treatment group or the control group.

2.Blocked randomization:

In blocked randomization, participants are divided into small groups or blocks, and then randomly assigned to the treatment or control group. The number of participants in each block is usually fixed and predetermined.

Criteria for using blocked randomization:

Blocked randomization is useful when the researcher wants to control for certain variables that may affect the outcome of the study, such as age or gender. It is also useful when the sample size is relatively small.

Example:

Suppose a researcher is conducting a clinical trial to investigate the effectiveness of a new drug in treating a certain disease. The researcher could use blocked randomization to ensure that each block contains an equal number of male and female participants.

3.Stratified randomization:

In stratified randomization, participants are first divided into different subgroups based on certain variables, such as age, gender, or medical history. Then, participants within each subgroup are randomly assigned to the treatment or control group.

Criteria for using stratified randomization:

Stratified randomization is useful when the researcher wants to control for certain variables that may affect the outcome of the study, such as age, gender, or medical history. It is also useful when the sample size is relatively large.

Example:

Suppose a researcher is conducting a clinical trial to investigate the effectiveness of a new drug in treating depression. The researcher could use stratified randomization to ensure that each subgroup contains an equal number of male and female participants of different age groups.

4.Adaptive randomization:

In adaptive randomization, the assignment of participants to treatment or control groups is adjusted based on the outcomes of the study. Participants who have a higher chance of benefiting from the treatment are more likely to be assigned to the treatment group.

Criteria for using adaptive randomization:

Adaptive randomization is useful when the researcher wants to maximize the effectiveness of the treatment by assigning more participants who are likely to benefit from the treatment to the treatment group.

Example:

Suppose a researcher is conducting a clinical trial to investigate the effectiveness of a new drug in treating a certain type of cancer. The researcher could use adaptive randomization to assign more participants who have a specific type of cancer to the treatment group, as they are more likely to benefit from the treatment.

 

In conclusion, the choice of randomization method depends on the specific research question, study design, sample size, and other factors. Researchers should carefully consider the strengths and limitations of each method before selecting the most appropriate one for their study.

Randomization

 

Randomization has been extensively used in experimental design to ensure there are no selection bias and reduces potential for confounding factors (extraneous variable or factor that is not controlled for in the study and can influence the results of the experiment like age, gender, etc.). There are four types of randomization methods commonly used in statistics: Simple, Blocked, Stratified, and Adaptive.

Simple	Blocked	Stratified	Adaptive
Participants are randomly assigned without any earlier determined criteria or blocking factors	This ensures equal distribution of participants across test groups over time. Block size is generally a multiple of treatment groups. Then all possible combination of assignment to test group are found and blocks randomly assigned.	Here participants are divided into groups based on specific factors that can influence the study before using simple randomization within group.	As name suggests, participants are grouped based on results obtained so far in the study. Participants are assigned to groups with higher probability of success
Especially used when sample size is large and there are no known confounding factors to influence the experiment	This is used to balance distribution of participants.	This ensures each group has similar distribution of participants with specific factors reducing risk of such factor affecting study result. Here the sample size is relatively small	Objective is to optimize allocation of participants. Here generally the sample size is large, treatments have unequal effects and early data can inform probability of success of different treatments
·         It is very simple to use. ·         Guarantees equal probability of assigning any participant to a test group ·         Reduces selection bias	·         Improves probability of equal group sizes over time ·         Balanced study result even with small sample size	·         Ensures more even distribution of factors that could affect study result ·         Provides more accurate estimate of study impact	·         Increases efficiency of participant allocation ·         Provides more accurate estimate of study impact ·         Reduces number of participants reqd.
·         However this may result in unequal distribution of participants across groups ·         Not suitable for small sample size	·         However groups may be created that are not comparable for certain factors, introducing some bias ·         Increased complexity of use	·         However this requires research of factors that could affect study result ·         Increased complexity of use	·         However this requires frequent monitoring of study data to adapt ·         May increase risk of bias due to potential for data driven decision making ·         Can be complex and difficult to implement
Examples like flipping coin, picking chits from a lot, using random number generator	For example, for testing new drug for lowing BP blocked randomization ensures the new or and standard drug are assigned equal number of participants	For example, for testing a new drug, this randomization will be used to ensure each group has same distribution of age, gender, disease severity	For example, we have 2 different treatments for a disease. Here adaptive randomization will help allocate more participants to the treatment that appears more effective based on data collected so far

 

 

Randomization, as per statistical theory, is an arrangement of set of objects or people in an unsystematic, random order, to prevent bias in rearrangement.  Randomization attempts to avoid deterministic outcomes, for a sequence of random variables; & instead, infers results in probability distributions.

 

Mathematically, for a discrete function x, the probability distribution is defined by a mass function denoted by f(x).

 

A randomization action needs to contain follow three steps:

 

1.    generation of the random allocation sequence: generation of the random allocation sequence by chance eliminates investigator bias from the experiment.

 

2.    allocation concealment: ensures that all participants and stakeholder in an experiment are not aware of the identity of the element or variable used in study. The step ensures that investigator is not prone to influence by any individual about nature and properties of material under investigation. The direction and scale of research, as- such is not influenced by an internal or external person. 

 

3.    implementation of the random allocation sequence:  ensures that investigator is not prone to judgment-bias during the course of experimental research study.

 

Randomization measures the probability of attaining specific set of outcomes through 
random section from arrangement of subjects.

 

Randomization in statistical studies, is categorized as—

Simple randomization—Randomization activity comprising of single sequence of random selection of study variable is known as Simple Randomization. Simple randomization ensures equal chance of survey participation to each participant, indicating equal probability for occurrence for each possible outcome. 
Flipping a coin during coin-toss, is most common example of simple randomization. 
A simple randomization is easiest to implement, however is unpredictable on nature & extent randomization.

 

The key disadvantages of simple randomization include unequal sample sizes generated from the activity, along with, non-justifiable baseline characteristics.  

 

In statistical experiments with larger sample size (n>200), simple randomization may result in similar numbers of participants among groups. On the corollary, with small sample size (n<100), simple randomization may result in an unequal number of participants among groups.

 

Blocked randomization—Randomization activity that attempts to randomize subjects into groups resulting in equal sample size, basis the use of smaller blocks & predetermined block assignments. As-such, a balance is restored in sample size, across groups, over time; thus, ensuring balance in trial arms, at all times. 

 

For proper effective randomization, constituent block size need to be determined before-hand and, need to be calibrated as multiple of the number of groups. Further, block size maybe increased incrementally, with all possible balanced combinations of assignment within the block computed there-after, to reach best combination for blocked randomization 

 

A key drawback with blocked randomization is predictability inherent in blocked randomization; when non-varying small blocks are used. The concern is more specific to unbinding subgroups shared during blocked randomization. Also, Blocked randomization is thought to generate imbalance in baseline characteristics, in case of smaller design of experiments.

 

Block Randomization is used to obtain groups with similar characteristics, that enables comparision between the members of the groups. A typical example of blocked randomization includes treatment group comprising patients with similar health characteristics. The effect of therapy and placebo is studied on treatment group and control group of patients obtained using block randomization.

 

Stratified randomization—Randomization activity first stratifies the whole study population into subgroups with same attributes or characteristics, classified as “strata”, followed simple sampling. In second activity, every element within the same subgroup is selected unbiasedly, through chance based random selection. A key advantage with stratified sampling includes balance in trial arms coupled with blocking.

 

The main limitation shared in stratified randomization is over stratification occurring due to incomplete blocks. Use of too many blocking parameters leads to small participant numbers within the block. The outcome may be predictable when non-varying, small blocks are used.

 

In a demographic study, for a sample universe of 500 people, two subgroups can be formed basis stratification randomization. Two covariate of sex (male, female) and individual parameter Education (three levels—undergraduate, postgraduate, doctorate) can be used to generate a total of six block parameters.

 

Adaptive randomization—Randomization activity in which allocation probability is changed according to the progress and position of the study. 

 

There is an underline effort to estimate imbalance of sample size among several covariates. A new element is added to particular group, only after estimating outcome for specific covariates and, previous allocation of elements into groups. Adaptive randomization is shared to minimize the imbalance between treatment groups.
This can be envisioned with help of example, where two covariates are considered basis gender of patient male, female along with count of patients categorized as underweight, normal weight and over weight patient.

 

Two different estimates are maintained for “treatment group” and “control group” of patients. The individual count of the score of patients in different category of body weight, is maintained to include all patients listed as underweight, normal weight and over weight with count of male and female patient included in total score. A new patient may be added to either of the “treatment group” and “control group” of patient basis health characteristics. The new patient may be categorized to either of category of patients, basis the weight of new patient. This helps to reach new count of patients in each the category and further modification may be shared basis progress of patient in response to the therapy.

 

As-such, an adaptive randomization is shared to initiate changes in the allocation probability, based on the therapeutic effect.
 

Simple randomization

Simple randomization is unpredictability that relies on one sequence of the arbitrary assignments. This method keeps the allocation of such a subject to something like a specific group completely random. Flipping a coin is the most popular and fundamental easy randomization method.

 

For instance, when there are two experimental groups (controlled versus treatment), each participant is based primarily on which side of the coin comes up heads (control) or tails (treatment). Alternative strategies involve rolling a die or using a shuffled deck of cards (for example, even-control or odd-treatment). For the straightforward randomization of participants, it can alternatively utilize a random number table from a statistical book or computerized software for numbers.

 

Randomization in blocks

 

"The block randomization approach is intended to randomly assign people to groups so that the sample sizes are equal. Using this technique, sample size distribution among groups is maintained over time. Because of the tiny size and balance of the blocks as well as the planned groupings, there is always a similar number of participants in each group. The researcher chooses the block size, which should be multiplied by the group count as such there will be groups of 2 treatments, size of block will be either 4, 6 or 8. The optimal way to employ blocks is in smaller increments so that researchers will smoothly maintain balance. Once the size of block is establish, all feasible balanced assignment combinations inside the block are considered. It is necessary to determine an equal number for each group within the block. The patients are then divided into the groups using a random selection of blocks.

 

Randomization by stratification

 

The stratified randomization process utilizes care of the issue of balancing and regulating the impact of covariates. By using this technique, groups of subjects' initial characteristics can be balanced (covariates). The researcher must specify the covariates after considering the potential impact each covariate may have on the variable which is dependent.By establishing an independent block for every combination of variables, stratified randomization may be achieved, and participants are then randomized to the correct block of covariates.

 

Simple randomization is used inside each block to divide individuals into the groups after each subject has been identified and allocated to a block.

 

Randomization with adaptation

 

Simple randomization, with or without accounting for the classification of prognostic variables, may contribute to the imbalance of significant variables between treatment groups in clinical studies of small to moderate size. Covariate imbalances are crucial because they have the potential to affect how a findings of the study is interpreted. Covariate adaptive randomization has been put out by a number of researchers as a valid alternative to randomization in clinical research.

 

The consecutive assignment of a new subject to a particular treatment group in randomization with adaptation takes into account the designated covariates and participant assignments from earlier trials. When using covariate adaptive randomization, the sample size divergence of various covariates is measured using the minimization method.

 

Industrial setting for randomization

 

Many businesses have employed randomization to make sure that operations operate as efficiently as possible. For instance, several airlines schedule flights using randomization. Which aircraft and members of the crew will fly on which routes are chosen using randomization techniques. This helps to avoid overbooking and guarantees on-time flight arrival.

 

Randomization methods are frequently used in the manufacturing sector to assess various raw material and processing combinations. For instance, an automaker might try with several oils or lubricants during a production process to see which mixture suits their requirements the best. Finding the best option for their functioning is the aim. Randomization is also used in the banking and finance industries to enhance processes. To determine the best strategy for lowering the risk involved with card payments or automated payments, banks may utilize randomization algorithms. Companies can create strategies that reduce potential costs while maximize earnings by evaluating multiple scenarios.

 

Randomization, though, is not always carried out perfectly. These are some recommendations for optimal practices.

 

1. Not all variables, such as a customer's background or attitude, may be taken into account during randomization, which could result in distorted or inconclusive results.

 

2. Not all process improvement methods are well suited to randomization. Think about using other methodological approaches as appropriate

 

3. Randomization prevents bias from seeping into test results, but it leaves room for potential confirmation bias. When someone looks for evidence to support their opinions or prejudices, confirmation bias may result. Randomization can make sure that all inputs are equally represented, eliminating any potential bias, by gathering a set of inputs, including materials, and thereafter randomly assigning them to a certain output, such as the item being made.

 

Wow. I am truly amazed at the quality of answers that I have got for this tricky and somewhat difficult question. All answers are a must read. There are 2 winners for this question - Anupam Goswami for summarizing the various methods in a tabular format and Nunhuck Oosman for highlighting the drawbacks of randomization.