Please enable JavaScript.
Coggle requires JavaScript to display documents.
Stats Concept Map (Holistic Survey: (Problem: Is there a correlation…
Stats Concept Map
Holistic Survey:
-
-
-
Example of Results: Males: 02 watched the game, 234 watched the game, 667 watched the game and 24 watched the game. Females watched
Strategy : Get a list of DHS students and split them into two lists based on gender (Male on one and Female on another)
Next number the males 01-700 (For example)
Number Females 01-650 (for example)
Next use a random number generator for the females with numbers 01-700 with no duplicates and get 20 numbers. Then ask my sample if they watched the super bowl or not
Then use a random number generator for the males with numbers 01-650 with no duplicates and get 20 numbers. Then ask my sample if they watched the super bowl or no
Non- Sampling Error
Response Bias- A pattern of incorrect responses in a survey or sample that leads to response bias. ex. Asking a group of students if they studied for the midterm would have response bias because they know that they were supposed to study so even if they didnt they all answered yes.
Non-response- Selecting people to be in sample group and they do not cooperate or you cannot contact them. If people you have to ask after you can not reach the initial people you wanted to contact there is is still bias. ex. elderly citizens are a good sample population because they are at home more often and are able to pick up the phone and do a survey.
Word Bias- The word choice of a question causes people to interpret the meaning of the question in many different ways causing the results to be different. Two very similar questions can get completely different results based on the word choice. ex. Is putting animals down okay?: 4 said yes and 6 said no... Is putting an animal down who is in pain okay? : 9 said yes and 1 said no.
Sampling Error
Voluntary Response Sampling- Sample made up of volunteers such as people who call into a radio station
Undercoverage Sampling- Sampling when members of the general population are not represented in the sample such as sampling voters
Convenience Sampling- sampling a group who is easy to get information from such as people at a local mall
Good Sampling Techniques
Stratified Sampling- technique of sampling where you divide the population into multiple groups based on similarities and differences and draw randomly from each group. Ex. A survey of the Duxbury High School student body. The population would be divided by class and then chosen randomly in each class.
Advantages: This can get good results because it gets information from multiple different groups and is porportional
Disadvantages: This is sometimes very difficult and can take a long time because stratas are difficult to pick consistently.
Cluster Sampling- strategy of sampling in which a large population is divided into multiple random groups called clusters and then SRS is done within the clusters. ex. A survey about teenagers and if they attend public or private high school is taken. Kids are grouped into three random clusters and then random kids in cluster one and three is asked what kind of school they attend.
Advantages: It is very efficient because samples can be taken from groups making the total population smaller
-
SRS- simple random sampling is a a subset chosen from a large sample. It gives each person the same probability to be chosen usually by numbering people (ex. 1-20) and choosing randomly with no repeats. This makes the results more realistic because it is very random.
-
Advantages: SRS is very random and if everyone participates who is asked, then it represents the whole population very well.
-
Observation studies- An attempt to observe and understand a cause and reflect relationship. In an Observation study the researcher can not control any aspects of the study. All they can do is stand by and observe.
ex. A study to see how consuming alcohol effects liver function. The study has three groups of people. People who say they have never drank alcohol, people that drink a moderate amount of alcohol, and people who drink a heavy amount of alcohol. These peoples liver functions will be tested to see if there is a strong link between the two. The observer can not tell the people to drink or not to drink, but they do this on their own and he or she observes.
Lurking Variable- A variable in an observation study that causes the results to differ that is not originally considered in the experiment.
ex. In an observation study the link between organic food consumption and weight loss. Two similarly sized people are taken and one is told to go on a strictly organic diet while the other is told to eat nothing organic. After three months both the organic fed person and the regularly fed person have lost 10 pounds of fat. The lurking variable in this experiment is the persons metabolism. The person who does not eat organic food has a higher metabolism than the other making up the difference in the weight loss.
Experiments- An experiment is a test controlled by multiple variables. Generally there are a few dependent variables and one independent variable. An experiment tests a hypothesis made prior to the testing.
ex. An experiment is made to see how amount of sleep at night effects grade point average over time. All variables in this experiment are kept the same except for one, number of hours of sleep (independent variable) Students with the same starting GPA, and same classes are taken and split into three groups. One group is told to get 8 hours of sleep per night, one is told to get 6 hours of sleep, and another is told to get 4 hours of sleep. Over the course of the year the fluctuation of their GPA is observed.
-
Bias- This is favoritism or error during the data collecting process that leads to skewed results. For example asking who is going to win the world series in Boston is going to get a completly different result than if you asked in Chicago. .
Blinding- Generally in medical trials. This is when the patient who is receiving a drug is not informed exactly what they are taking and how it effects them
Placebo- Generally in medical trials. A neatral treatment that has no effect on the dependent variable in your experiment but the subject might think they are being effected when they are not.
Completely Randomized Design Experiment: I chose 200 random patients from tufts medical center in Boston who suffered from vertigo. I chose these 200 patients out of the pool of 600 patients diagnosed there for this experiment. I used SRS to do this experiment because it was the most random way of choosing. We labeled the subjects randomly 01- 200. 25% of the patients were given nothing, 25% were given a placebo and 50% were given their normal medication for vertigo. We did not tell the patients what they were being given so they would not have a bias answer when we asked how the drug effected their condition. After a week of being medicated we asked how each patient felt and did tests seeing how were effected.