Please enable JavaScript.

Coggle requires JavaScript to display documents.

Unit 4 (Samples and Surveys (Parts of Properly Designed Survey (Simple…

- - - - A non-sampling error is a bias that is not from the . Most commonly, it comes from the lack of data or the failure to equally represent an entire population.
      - Examples of non-samplings errors include: nonresponse, response bias, and wording bias.
        
        Response Bias is when the subjects response with a false answer on purpose. They feel that they need to respond with the 'correct' answer. An example is if the school anonymously asked students if they had ever cheated on an assignment. The percentage of yes's would most likely be lower than the actual truth.
        
        Wording Bias is the miscommunication to subjects. They do not understand the question properly, and answer the incorrect way. This creates false data, which is bias. An example is a poll asking "Should the government pay for everyone's education?" The lack of clarity regarding amount of money, length of education, and who 'everyone' is creates a wording bias.
        
        Non-response is the worst of non-sampling error. It is just as described, when a subject does not respond to a survey or census. One example is the nonresponse of the working class to a phone call at 2 PM. If the survey is about healthcare or union rights, their responses will be similar. The lack of their response creates a bias, and does not represent the whole population.
  - - - Voluntary Response Sampling
        
        Voluntary Response Sampling is data from a survey that requires volunteers. These volunteers will not represent the ideas of an entire population. One response is a survey for customers at stores that will give a benefit to those who partake. The volunteers will be looking for a reward. Those who do not really need the reward/coupon will not partake in the survey, creating holes in the data, aka bias.
      - Under-coverage
        
        Under-coverage is the broad term used to describe when some of the population is not represented. It does not describe an entire population, therefore creating bias.
      - Convenience Sampling
        
        Convenience sampling is only collecting data that is convenient. An example is a school wide survey about prom. Only asking people in your classes will be convenient, but will not represent the entire population.
  - - - Selecting a section of the population where every person has an equal chance of being selected
        
        Advantages
        
        Usually not complex
        
        sometimes creates misrepresentation of population
        
        Disadvantages
        
        Takes very long
        
        Random error
    - - Dividing the population into sections that are internally random and randomly selecting whole sections
    - - Classifying the population into groups of similar individuals, surveying each group, and combining them proportionately
        
        Advantages
        
        Individual frequency is proportional to the whole population
        
        Disadvantages
        
        A lot of work
        
        Requires knowledge of the proportions of the population
    - - Prom Survey: Who should go to prom?
        
        We are interested in who the student of Duxbury HIgh School thinks deserve to go to Prom. Specifically we are wondering if non-Duxbury people should be allowed to go.
        
        We will ask everyone the same question:Should anyone who is not a student at DHS be allowed to go to prom?
        
        The next step is eliminating Sampling and non-Sampling error. We need everyone in the school to answer this question. While difficult, we will know the whole population's ideas. We have worded the question simply to ensure there is no wording bias.
        
        To make sure everyone answers, we will make a single file line at the main entrance to DHS. That day, the only way for high schoolers to enter the building will be through one door. To enter, they must answer the question yes or no. We will find out who was absent or tardy that day, and individually email/find them in class to seek their opinion.
        
        This survey can have no coverage issues, as it covers the entire population. The wording is clear, and we have limited bias as much as possible. This survey will create a very clear picture demonstrating the student body's thoughts on who should go to prom.
- - - - Factors
        
        Conditions you will be changing to look for a response
        
        Treatments
        
        the condition or effect applied to an experimental unit or subject
        
        Experimental Units
        
        The individuals in an experiment that are the focus
        
        Humans (and occasionally animals when liberals are offended) are known as subjects
        
        Response Variables
        
        The measurement made at the end of the experiment to determine the results
        
        Blinding
        
        Blinding is giving subject little information. This includes tearing labels of of medicines, not telling the difference between treatments, etc. This is done to eliminate the placebo effect
        
        The placebo effect is the thoughts/predictions of subjects affecting their feelings/outcomes. If testing a new medicine, subjects who are told that they are being given a new medicine my convince themselves they are feeling better. To eliminate placebo effect, blinding is needed.
        
        Placebo Effect
        
        An effect of a treatment that is caused by the subject's belief that it should work, and not its actual effectiveness
      - Replication
        
        Having enough experimental units so that any variation in response can be distinguished from random error
      - Random Assignment
        
        Uses chance to assign treatments to experimental units. This creates roughly equivalent groups of experimental units.
      - Control
        
        Used to eliminate lurking variables. Uses comparative design to ensure that the difference between responses is based on the different treatments
  - - - Example: An observational study looking into the association between shooter video game usage in a state versus the amount of gun violence in that state. There are many other factors involved, but there might be a small positive correlation between violent shooter video games and amount of gun violence. This is an observational study due to the countless lingering variables.
        
        Confounding
        
        When you fail to isolate a variable so you incorrectly interpret the response variable. Lurking variables can lead to confounding. You cannot tell how much each explanatory variable contributes to each response.
- - - - Using a random number table and the active roster for each team (including the players' number), select the two first players from each team based upon their team. Go through the random number table selecting consecutive two digit pairs until there are two pairs that match the number of a player on the roster. These two players from the given team will be used in the experiment. Repeat this 29 more times so that there are 60 total players, 2 from each team. If going through the table for the Houston Rocket's roster, and the first numbers that appear are 13 and 00, James Harden and Russell Westbrook.
        
        Treatments- Each player will shoot 100 threes from the top of the key, behind 23.75 feet, the three point line for the NBA, for each rim height. The response variable is amount of 3 pointers made. We will tell the NBA player that we are studying their shooting mechanics for the new video game, 2K20, and that the rim is at the same height.
        
        10' rim
        
        9'6" rim
        
        10'6" rim