Sampling Methods: How to collect data Some important terms Random - PowerPoint PPT Presentation

Sampling Methods: How to collect data

Some important terms � Random - occurring by chance � Population – a group of individuals or items that a study focuses on � Sample – a subset of the population, i.e. individuals selected for the study….Why do we need a sample?

Samples are important because… � It’s much cheaper to collect data from a subset of a population than the whole population. � It also costs less in terms of resources (person-power, computer-power, paper, etc.) to collect data from a subset of a population than the whole population. � It’s also more efficient in terms of time to collect data from a sample.

Simple Random Sampling � all selections must be equally likely � all combinations of selections must be equally likely � A random sample may not end up being representative of the population, but any deviations are only due to chance. Much like in probability, even though something is very unlikely to happen, it still may happen by chance.

Simple Random Sampling approach A yearbook survey at get a list of all 1100 � � CB students at CB and number them population: the � students of CB use a spreadsheet to � generate 100 random Sample size required: � integers from 1 to 1100 100 if any number appeared � more than once, we would have to generate a new number, i.e. you can't survey the same person more than once

Advantages of Simple Random Sampling � This is the simplest method to carry out. � This method will most likely generate the most random sample.

Disadvantages of Simple Random Sampling � This is the costliest method to carry out in terms of resources and $$$.

Systematic Random Sampling � you sample a fixed percent of the population � randomly choose a starting point � then sample every nth individual, where population size n = = = = sample size

Systematic Sampling approach A yearbook survey at CB 1100 � population: the students of n = � 100 CB n is 11, so we can choose to survey Sample size required: 100 � � every 11th person until we reach 100 surveyed Use our list of 1100 students and � generate one random integer (i) from 1 to 1100 a spreadsheet. That number will represent the first person we survey. Suppose i = 397, then start at element 397 and count 11 from them and survey that person. Continue in this way until you have the sample size you need. If you get to the end of the list, � continue counting at the beginning.

Advantages of Systematic Random Sampling � This method will work very well any time your population is in a line, listed somehow, or one element is arriving one after the next. � This a very simple and inexpensive method to carry out if your population is in a line in front of you (e.g. a line up of people waiting to see Star Wars VII).

Disadvantages of Systematic Random Sampling This method requires a lot of resources if your � population is very large (like thousands or millions of elements). It will take a looooong time to count through the list to get your sample. This method is very difficult to carry out if the � population is not listed or lined up. This method will be very expensive if your elements � are very spread out. For example, suppose you want to personally interview a sample of people from the J.K. Rowling fan writing club. You have a list of world wide members, and select 100 of them. You have to fly all over the world to interview them. $$$!

Stratified Random Sampling � divide population into groups called strata (maybe by age, location, etc.) � a simple random sample of each strata is conducted � the size of the sample is proportional to the size of the strata

Stratified Sampling approach A yearbook survey at CB strata will be grades 9, 10 , � � 11, and 12 population: the students of � CB calculate the percentage of � the students in each grade, Sample size required: 100 � that will tell you how many students to survey from each grade since our sample size is 100 get a list of students by � grade and use a spreadsheet to pick students from each of the grades depending on how may students are in that grade

Advantages of Stratified Random Sampling � This method will ensure that every subset (of interest) of the population is represented. � Because each subset is sampled proportionally, an overall average or opinion can be determined.

Disadvantages of Stratified Random Sampling � This requires a lot of resources! � This method generates different sized subsets, so you have to be very careful when comparing them. You must compare PROPORTIONALLY!!!

Cluster Random Sampling � organize the population into groups � randomly select groups � select all people in the selected groups

Cluster Sampling approach � A yearbook survey group by first period � class at CB randomly select 3 or 4 � population: the � classes and survey students of CB everyone in each of � Sample size those classes to do the survey to get the required: 100 required 100 surveys

Advantages of Cluster Random Sampling � This method requires the least amount of $$$, time, and resources. Imagine researchers are surveying Inuit populations. The researchers wouldn’t have to travel to every single town. They can choose a small subset to visit.

Disadvantages of Cluster Random Sampling � This method introduces bias into the survey. Because only a small number of groups of the population are surveyed or tested, only those opinions are represented.

Multistage Random Sampling � organize the population into groups � randomly select groups � randomly sample individuals in the selected groups This method is called “multi”stage because the � researcher must generate “multi” random samples. The first is the random sample from the groups, then the researcher must create a random sample for each group chosen.

Multi-Stage Sampling approach � A yearbook survey � group by first at CB period class � population: the � randomly select 10 students of CB first period classes � Sample size � randomly select 10 required: 100 students from each of those 10 classes to complete the survey

Advantages of Multistage Random Sampling � This method is fairly efficient, especially when data is very spread out. For example, if a researchers are surveying Inuit populations, they don’t have to travel to every single town. They can choose a subset to visit. � More groups are surveyed compared to cluster, so there will be less bias.

Disadvantages of Cluster Random Sampling � This method, like cluster random sampling, introduces bias into the survey. Because only a small number of subsets of the population are surveyed or tested, only those opinions are represented. Less bias is introduced, however, since more groups are surveyed. � It will be more expensive than cluster random sampling, since more groups are being surveyed.

Destructive Sampling � This is simple or systematic random sampling where selected items cannot be reintroduced into the population. They are destroyed either as a result of the testing or after they are tested. � Example: Light bulbs are being tested for quality control. After a bulb is tested it cannot be sold so it is removed from the population.

Advantages of Destructive Random Sampling � This method allows companies to test their product for quality control. This gives their consumers confidence in the product, allows the company to improve their product, and limits the company’s liability for defective parts.

Disadvantages of Destructive Random Sampling � This method decreases the amount of product in circulation, depending on how many elements are tested. It costs the company money to perform the test, and it costs the company money because they are destroying product.

Do you get it????? Get the worksheet, DO IT, and check your answers! Ask if you have ANY questions!!!