Title | Statistics Chapter 6 - Lecture notes Lecture 6 |
---|---|
Course | Introduction to Statistics |
Institution | Laurentian University |
Pages | 9 |
File Size | 328.5 KB |
File Type | |
Total Downloads | 51 |
Total Views | 223 |
In-depth notes on sampling distribution and hypothesis testing including a question that will be on the exam as well as pictures in the notes ...
Statistics Chapter 6 - Sampling Distribution Exam question: When we are interested in drawing inferences from collected data, there are 3 frequency distributions of interest. Name and describe each of the 3 distributions. Describe the relationship between the distributions and how they are useful to inferential statistics ● From a sample we can calculate the mean and standard deviation ● Population also has mean and standard deviation ● We use samples to understand population ● The mean of the sample is based on sample size ● To compare M to the population directly it doesn't work ● z=x-μ ● The normal curve is a construct ● If the data is normal and symmetrical ○ We understand a lot about the nature of the curve itself ● All the possible M’s of a given size that could be derived from the population it came from Frequency distributions of sample means ● A sample mean on average, is close to the mean of the population from which it came from ● Inferences about population parameters can be made from sample statistics ● Ex: if our assumptions about the data being normally distributed are valid, then the sample mean, on average, reflects the mean of the population ○ Some error - on average ● The means from different from samples obtained from the same underlying population are all estimates of μ ● What needs to be determined is the probability that a particular sample actually comes from a particular population ● Compare the sample mean that we have collected during the experiment to a distribution of all possible sample means from the population ● NOTE: this is not a comparison of our sample mean and the distribution of individual scores within the population ● There is always going to be some variability between the sample statistics we obtain from the experiment and the population parameter ○ Sampling error: the amount by which the sample estimate of the measure (say the mean or SD) differs from the underlying population parameter (μ or σ) ○ In general, the difference between the samples are simply due by chance
The Sampling Distribution ● Another theoretical distribution ● It is a distribution that gives the probability of all possible values of a sample mean ● When collecting sample statistics, we can exploit the distribution of means to male inferences about the population ● Ex: population scores: 2, 4, 6, 8. The μ = 5 and σ= 2.24 ○ We can create a distribution of all the possible random samples when n=2. We assume that sampling with replacement is happening ○ Takes 16 samples to get every combination
● Note that ○ The sample means tend to “pile up” around the population mean (recall from a previous slide that μ = 5) ○ The distribution is approximately normal in shape ● Another way of thinking about the distribution of means is that it is a probability distribution ● Mean of a sampling distribution is always the same ○ Ex: the probability of pulling a sample that has mean = 3 is: ● We can calculate the parameter of the sampling distribution of the mean is:
Standard error of the mean ● The SEM is the standard amount of difference between a sample (M) and the population mean (μ). This represents the average amount of difference we can expect between the population mean and the sample mean ● The M and μ will usually be relatively close; however, there is always going to be some variability between these values for each sample. The σM quantifies the variability ● Consider the original population of 4 scores, the μ=5 and the σ=2.24 σ2=5.01 ● Ex:
● Sample size increases, there is less variability (or less error) associated with predicting the population mean
● The central limit theorem: ○ Is a statistical principle that defines the μM, σM, and the shape of the theoretical sample distribution ○ In essence, it tells us that for a population with a given σ, μ, the distribution of sample means for sample size of n will approach a normal distribution with a mean of μ and a SD of σ divided by the square root of n ○ The distribution of sample means is the collection of sample means for all possible random samples of a given n that can be obtained from the population ○ A sampling distribution is a distribution of statistics obtained by selecting all of the possible samples of a specific size from a population ○ There are 2 fundamentally important facts that male the central limit theorem useful to us: ■ It describes the distribution of sample means for any population, irrespective of the population, shape, mean, or SD ■ The distribution of sample means “approaches” a normal distribution very quickly with increasing sample sizes (n) Principles of central limit theorem ● The mean of the distribution of means is always equal to the mean of the population: ○ μM=μ ● The standard error measures the standard amount of difference that is expected simply due to chance :
● If the underlying population is normally distributed, then the distribution of means is also normally distributed ● If the underlying population distribution is not normal, Ex: The population mean for the GRE is 558 with a standard deviation of 139. A researcher gives a training course to 100 candidates to see if their scores will improve. After the course, students take the GRE and score, on average, 585. Does the training course improve performance on the GRE? ● μ = 558 ● σ= 139 ● ●
N = 100 M = 585
●
● ● ● ● ● ● ●
● ●
●
μM= 558 σM =13.9 z= 1.94 Sample that is less that 558 -> 50% Probability of getting 585 or less -> 0.9738 Mean greater than 585 -> p= 0.0262 Does the training course improve marks on the GRE ○ No Ex: a researcher undertakes a study of a treatment for depression. There are 120 depressed patients who are getting a new form of therapy ○ Population 1: all possible patients in existence that have met the criteria for depression and who are getting a new form of therapy ○ Population 2: all possible patients in existence that have met the criteria for depression and who do NOT receive the new therapy ○ The goal of the experiment is to see if after 4 weels of the new therapy, patients that have a new form of therapy have less depression than patients who do not get the new therapy Let's assume that the researcher is using a standardized test of depression like the Beck Depression Inventory or the Hamilton Depression Scale Since the DV is a standardized test, we know the statistical parameters of the underlying depressed population ○ μ = 69.5 ○ Σ = 14.1 Given the characteristics of the central limit theorem, we know the parameters of the distribution of means related to this population: ○ μM = 69.5 and
○ After 4 weeks of the new therapy, the subjects are all re-evaluated using the standard test that was used to make the initial diagnosis. The 120 subjects in the experiment are the sample. The mean depression score on the standardized test is 66.8. The million dollar question is “did the therapy make a difference”? ○ We need to determine how the sample mean compared to the population mean
○ The researcher has found the the treatment group has a score that is 2.09 SD’s below the mean of the depressed population. The researcher needs to decide if the treatment is effective is the treatment enough to make a clinical difference? Hypothesis testing ● Means from different samples from the same population are estimates of μ ● The most important question when considering inferential statistics is “what is the probability that a particular sample mean comes from a particular population”? ● Compare the sample mean to a distribution of all possible sample means in the population and not to a distribution of individual scores from the population Logic of hypothesis testing: ● We assume the experimental manipulation has no impact on the DV. that is, our manipulations make no difference ● If this basic assumption is wrong, then the conclusion that the manipulation does male a difference can be supported ● The confusion logic is necessary as mathematically, we generally only know about the circumstance where there are no differences ○ μM and σM ● Ex: a person claims that she can identify people who have above average intelligence with their eyes closed. We take her to a full stadium and blindfold her. We ask her to pick out someone with greater than average IQ ○ We know that IQ is normally distributed in the population with μ = 100 and σ = 15 ○ We know that if she picks out someone with an IQ of 145 (3 DS above the mean), it is unlikely that we can attribute her performance to chance ○ Since 145 is 3 SD’s aboce the mean, we know that the relevant z-score is +3.0 ○ Is we consult the tables, we find that the probability of picking someone with an IQ of 130 as only 2.3% of scores are higher than that ○ What if she selects someone with an IQ of 115? Are we as impressed? There is a probability of picking this score or higher equal to 15.9% ■ We might consider this an inconclusive result
○ The deal with the inconclusive, we might decide a printer on an IQ score that, if she successfully picked a person that has that IQ, we would be impressed by ■ We could say that we want her to pick people out with a 2% probability of being wrong. That would mean that we said “ pick out people with IQ’s greater than 130 The Jargon ● Null hypothesis ○ H0 - a statement that suggest that the treatment of interests will have NO outcome on the results of the experiment ○ The H0 acknowledges that we are assuming that the experiment will not describe or influence an existing natural phenomenon ● The alternative hypothesis ○ H1, ○ Logically, the complete opposite to the H0 ○ The H0 and H1, must be mutually exclusive and exhaustive Steps in hypothesis testing ● Ex: we are researchers who want to investigate the effectiveness of a new stress reduction method that we have developed: ○ Suppose a subject is randomly selected from the general population. The subject is treated with the new stress reduction method. Then, his or her stress is measured ○ We will assume that we know the population parameters related to stress within the population from which the subject was taken: ■ μ= 40, σ= 10 ● We can reframe the research question into null and alternative hypotheses about the specific population that we are interested in: ○ Population 1: all the people in the world who have not been trained in the new method ○ Population 2: all people in the world who have been trained in the stress reduction method ■ H0: Those trained in the method do not have a lower stress score ■ H1: those trained in the new method do have lower stress scores ● We now need to think about the information that we do not have about stress in the general population ○ μ= 40, σ= 10
○ We know absolutely nothing about the other population (population 2) other than the stress levels of the subjects selected from population 1 ○ Does the performance of the subject most probably represents the general population (population 1 as represented by the H0) or is their score so extreme that it is unlikely to represent the general population and therefore, most likely represents another totally different population - the population of all possible people who received the new stress reducing treatment (population 2 represented by H1) ● Decide the point at which we are comfortable with saying “no that score is too extreme from the general population” ○ Establish the “cutoff” sample score at which the H0 should be rejected in favour of the H1 ○ Accept a 1% possibility of making a mistake in declaring our new treatment a success ○ Another way of saying this is that for our treatment “successful” the subjects has to have a score that has the probability of being from the general population of 0.01 or less ■ In the jargon, we are setting a=0.01 ○ We will need to know the point on the distribution that will “cut off” the lower 1% of the distribution ○ Consulting the z score tables, the z score that is consistent with 1% is ± 2.33
● Suppose that the subject scored 15 on the standardized stress scale that we know the population parameters for ● Z score for this raw datum
● Contrast the subjects score with the general population to see if the new treatment caused the subject to experience decreased amounts of stress such that their experimental score is unlikely to represent the general population ○ Calculated that the subjects performance yielded a z score of -2.5. This is more extreme than our cut off point
○ Therefore, we reject H0 that there is no difference between the subjects score and the scores of the general population ○ Since the H0 is rejected, the H1, must be supported. We conclude that the subjects score is unlikely to have come from the population of people who did not get the new treatment ○ We further conclude that we are accepting a 1% chance of being wrong in our determination ○ The most interesting conclusion is that our new stress reduction method leads to lower stress scores (p...