STAT 110, Notes for Week of 8/30/16
STAT 110, Notes for Week of 8/30/16 STAT 110
Popular in Introduction to Statistical Reasoning
verified elite notetaker
Popular in Statistics
This 7 page Class Notes was uploaded by runnergal on Saturday September 3, 2016. The Class Notes belongs to STAT 110 at University of South Carolina taught by Dr. Wilma J. Sims in Fall 2016. Since its upload, it has received 5 views. For similar materials see Introduction to Statistical Reasoning in Statistics at University of South Carolina.
Reviews for STAT 110, Notes for Week of 8/30/16
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 09/03/16
STAT 110 – Notes for Week of 8/30/16 Chapter 4 o Sampling Errors: errors that are caused when taking a sample. These errors make sampling statistics different from the actual parameter of the population. Undercoverage: a type of sampling error where certain groups of people are left out of the sample. This results from a flawed sampling frame: the portion of the population from which the sample is chosen. For example, if a researcher wanted to determine how many people in STAT 110 studied for at least one hour before the first test, but the researcher only surveyed people coming out of the men’s bathroom, then there would have been undercoverage of women since the sampling frame only included men. Random Sampling Errors: a type of sampling error where there is a difference between the result of the sample and the actual parameter of the population. These errors are due to chance and exist in all samples because no sample perfectly describes a population. A larger sample, however, helps to mitigate this problem. o Nonsampling Errors: an error that occurs outside of the sample selection. Nonresponse Error: a type of nonsampling error where people that the researchers would like to include in the survey do not respond. One example of a nonresponse error is if researchers tried to determine if sleep is related to academic success for people aged 18-22 but no women responded. Response Error: a type of nonsampling error where a subject(s) in the sample answer the survey questions incorrectly. These subjects may either lie or misremember. One example of a response error would be if a researcher asked high school students if they used recreational marijuana, and some users lied and said they did not use it. Processing Error: a type of nonsampling error where researchers input data incorrectly. One example of a processing error would be if a researcher heard a subject respond that the subject did smoke weed, but the researcher wrote down that the subject did not smoke weed. Data Collection Error: a type of nonsampling error where the researchers may word a question in a leading manner, may give the survey during a time when few people will answer the survey, etc. One example of a data collection error would be if a researcher called subjects and asked them to respond to a question during Labor Day weekend. Since most people are engaging in social activities on Labor Day, the sample would not include all groups of people equally. o In order to (partially) negate the effects of nonsampling errors: Substitute other households for nonresponders. For example, if a woman aged 19 refuses to answer the survey, the researcher should ask another woman aged 19 to answer the survey. Weight the responses according to the groups of people that answered the survey. For example, if 10% more men than women answered a survey, then the women’s answers should be given an extra 10% in importance. o Stratified Random Sampling: Essentially the same thing as a simple random sample, but the subjects are split into different groups, called strata, before the survey is administered. Once the survey is administered to all subjects in all strata, the researchers should take a simple random sample of each individual strata. Then the researchers should average the results of all strata. These strata are chosen based on demographics that the researchers already know, such as gender, ethnicity, religion, age, etc. o Probability Samples: samples where each individual in the population has an equal chance to be chosen. Simple random samples and stratified random samples are both types of probability samples. Chapter 5 o Observational studies are passive, since researchers simply observe; they do not impose treatments on the subjects. Observational studies can only show correlations; they cannot prove cause and effect. o Experimental studies are active, since researchers impose treatments on the subjects; they do not merely observe. Experimental studies can prove cause and effect. o Response Variables: a variable that measures the results of a study. Also known as a dependent variable, since it changes in response to the treatment. If researchers were trying to find out if sleep affected academic performance, then academic performance would be the response variable. o Explanatory Variable: a variable that probably incites the changes in the response variable. Also known as the independent variable, since it is sometimes controlled by the researchers. If researchers were trying to figure out if sleep affected academic performance, then the amount of sleep would be the explanatory variable. o Subjects: the individuals that the researchers study during the experiment. o Treatment: the action that is used on the subjects by the researchers. o Lurking Variable: a variable that affects the response variable but is unknown to the researchers. If researchers are trying to find out the average amount of sadness in an average person, a lurking variable could be whether someone had recently experienced a death in his/her family. o Confounding Variable: a variable whose effect cannot be differentiated from another variable(s). o Randomized Comparative Experiment: where the results from an experimental group are compared to the results of a control group. Both groups receive a treatment. This type of experiment helps to control the effects of lurking variables. Experimental Group: a group of subjects that have the complete experiment imposed on them. For example, if researchers were trying to find out if a new cancer treatment worked, the subjects in the experiment would receive the new treatment. Control Group: a group of subjects that receive a placebo as their treatment during the experiment. Placebo: a treatment that mimics average conditions/conditions without the experimental treatment. A dummy treatment. Placebo Effect: the (sometimes positive) effect of the placebo on subjects. When subjects believe they are receiving treatment to make them feel better, they may start feeling better, regardless of the active ingredients in the treatment. o One-Track Experiment: where all subjects in the experiment receive treatment. The results of this experiment cannot be taken seriously since there is no comparison to a control group. o Two-Track Experiment: where subjects in the experiment are split into an experimental group(s) and a control group. Since there is a control group, the researchers can determine if the experimental treatment has an effect on the response variable. Therefore, the results of this experiment can be taken seriously. o Randomized Comparative Experiments: subjects are randomly assigned to groups with roughly the same number of subjects in each group, these groups get different treatments, and then the researchers compare the responses. Researchers can draw cause and effect relationships from these experiments. o When creating an experiment… Control effects of lurking variables by having similar subjects, comparing at least two treatments, etc. Randomize the groups (assign subjects to groups by chance). Use enough subjects to reduce variation. o A good experiment should have statistically significant results: effects that probably couldn’t happen by chance. o When creating an observational study… Compare random samples of people. Match the demographics of the people in the control group to the people in the experimental group to reduce the effects of lurking variables. Measure and adjust for confounding variables. Chapter 6 o Single-Blind Experiment: where either the subject or the researcher does not know what treatment is being imposed. This eliminates either subject or researcher bias. o Double-Blind Experiment: where both the subject and the researcher do not know what treatment is being imposed. This eliminates both subject and researcher bias. o Some problems in experiments are related solely to the subjects: Refusal: some subjects refuse to answer or skip some questions. Other subjects refuse to be part of the study at all. These problems result in a lack of data. Nonadherers: some subjects do not strictly follow the guidelines of the experiment, resulting in flawed data. Dropouts: some subjects begin the study but do not finish it, resulting in a lack of data.
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'