Some of what you need to know… Chapter 11 Some of what you need to know…
Chapter 11 Random does not mean “unexpected”. Humans are not random. Like, ever. Example exceptions: Seizures, Gametes Randomness refers to there being a collection of known possible outcomes, but which cannot be determined ahead of time. According to the College Board, the best way to randomly select something is to write down the options you have on separate pieces of paper, put them in a hat, shake the hat up, and select however many you need without looking.
Hunting For The Unicorn Sampling Procedures
Representative Samples Representative Samples are like unicorns. They are imaginary. They are awesome. We do not usually need a truly representative sample, but instead we need a sample that has a representative average. That is an average that is close to the true population average.
Sampling We sample because it is usually easier and more cost-effective than surveying everybody. Also: It takes less time. It is usually representative enough. It is sometimes more possible than surveying everybody.
Census A census is when every member of the population is surveyed. A census is generally pretty much a bad plan. You never get everyone, it takes a lot of time and effort, and by the time you are done some of your information may be out of date.
Sampling When we sample, we want to randomly select our sample. The randomness helps us to scatter lurking variables we were unaware of or not accounting for. This makes our sample more reliable since other factors we did not know about are now less likely to affect the averages in a disproportionate way.
Quantity Is Quality The bigger your sample, the more representative your average. It is more important that you get a large enough sample by numbers than by percentage. A sample of 100 students is about as good at estimating something about Highland as a sample of 100 U.S. marines would be for estimating something about the entire Marine Corp.
Sampling The larger group that we pull a sample from is called the population. The actual group we survey is called the sample. We hope that our sample is fairly representative of the population. It is most accurate to say “representative enough” of the population.
It’s All Greek To Me When we talk about populations, we will use Greek letters for the variables. The mean is µ and the standard deviation is for a population. When we talk about samples, we will use the alphabet that is more familiar to us. The mean is and the standard deviation is s for a sample.
Sampling In taking random samples, the purest way to get randomness is the hat strategy. We get the whole population and then choose at random. This can be time consuming or even impossible. When it is not impossible, we call this a simple random sample, abbreviated SRS.
Error and Bias Samples will be a little different than the population. This difference is called sampling error. This error is ok. If we do something with our data collection method that creates a less representative sample, this is called bias. Bias is not ok, and is very bad.
Sampling There are alternatives to simple random sampling. They are: Stratified Sampling Cluster Sampling Systematic Sampling Multistage Sampling Convenience Sampling Convenience sampling is really bad.
When Good Unicorn Hunts Go Bad Bias Rears Its Ugly Head!
Identifying the Population This can be a really tricky process. Consider the Highland Student Body. Do we include students who come from other schools? Do we include students who started the year here, but transferred? Who transferred here later? Are currently expelled or suspended? Do we include students who are foreign exchange students? Students studying abroad?
Identifying the Sample To make a good sample, you need to be aware of who exactly is in your population. When you randomly determine a sample, you need to stick with the originally selected people, even if they are hard to get a hold of. Adjusting for new people based on convenience alone makes it a convenience sample.
Two Houses, Both Alike In Dignity There are two general categories of bias: Nonresponse Bias Response Bias A nonresponse bias results when people in the sample do not respond. A response bias is something in our survey that affects the way people respond.
But Wait…There’s More There is also the idea of undercoverage. Basically this means that some group is not getting its fair share of representation. This cannot always be completely avoided, but it is considered bias if the way we do the study creates the undercoverage.
Nonresponse Bias Voluntary response bias is a kind of nonresponse bias created specifically when people choose whether or not to respond. Other factors besides simple choice can keep someone from responding, such as time or schedule constraints, language barriers, or just being ill.
Response Bias A wide variety of things can influence how people respond. The most common is how questions are worded. Also, the order of questions. Even the title of the survey makes a difference. If there is an expectation for the answer you want to hear, this also can cause bias.
Sampling Methods Explained How We Hunt Sampling Methods Explained
Chapter 12 – Sampling Methods The most ideal method of sampling is called a simple random sample. You take everyone in the population and randomly select however many you need. The downside of this method is that you need a full roster of the population, which is often difficult to get or maybe even impossible. This usually only happens in textbooks.
Chapter 12 – Sampling Methods The two next most common sampling methods are cluster sampling and stratified sampling. These methods involve splitting the sample into groups, which is actually often convenient as data frequently already comes in groups. For example, schools can be broken down by homeroom, people in the U.S. can be broken down by state, city, county, or zip code…and so on.
Chapter 12 – Sampling Methods In stratified sampling, you pull a few people out of each group. These groups, by the way, are called strata. Stratum is the singular term. In cluster sampling, you select a few groups and sample each person in those strata. Sometimes these can be merged into a combination of the two, which is called multistage sampling.
Chapter 12 – Sampling Methods There is also systematic sampling, which involves taking a list and selecting every n-th person. This is only a valid method if you have a list in a randomized order, but since most lists are in alphabetical order, this tends to be an inferior method.
Chapter 12 – Sampling Methods Finally, there is convenience sampling, which is sampling based on whoever or whatever was convenient to study. Don’t do it! Portal to HELL!!! Metaphorically, that is. Still, though, it is gateway to potentially unmitigated badness.
Assignments Ch 12: 2 problems from 1 to 10 and also problems 11 and 12. Due Friday The experimental design write-up will be due the week after break. These are the only two assignments in Unit 3. Read Chapter 13 through the top of 298 for Thursday. Read the remainder of Chapter 13 for the Monday after break. Quiz Friday over chapters 11, 12, and parts of 13. There will be a quiz the week after break over experimental design (which is chapter 13). These are the only two quizzes in Unit 3.
Quiz Bulletpoints (This Week) Know how to randomize. Know the different types of bias. Know the difference between controlled variables and blocking variables and be able to suggest some of each in a given scenario. Know the difference between statistical significance and practical significance. Know the difference between factors, levels, and treatments.