BusML 4202 Midterm 2 tricky concepts reminder The concepts included in this notes are the ones that I get wrong and they are very tricky High chance to be on the exam Attitude and scaling understand the difference between Likert scale and graphing continuum He doesn t ask much in the sample midterm so I don t think there s a need to worry about it Questionnaire design Remember the major steps of setting a survey 1 What should be asked 2 The layout 3 and finishing and refining questionnaire Understand questionnaire definition and objectives Remember a questionnaire should not be a burden to respondents have to be clear avoid making assumptions and the solutions to each problem i avoid complexity use complete sentence to ask questions ii double barreled one question one item iii avoid vague wordings in terms of time i e regularly iv avoid burdensome questions in terms of time frame v avoid making assumptions allow respondents to skip questions vi make sure no overlapping in questionnaire options e g Don t say stuff like 1 10 10 20 vii leaded and loaded questions clue respondents what to answer solution third person technique purpose of funneling and how that relates to the objective of setting up a questionnaire Observation research In an observation study you can observe physical behavior and characteristics but definitely NOT cognitive process no one can do mind reading observational research is less likely to have social desirability bias than survey research that s why observational research is preferred to questionnaire survey research Reason you can have biased wordings in a questionnaire For observation you can have greater control if it is contrived or even rely on machinery for observation main difference between human vs machine observation human they observe machine they record e g physiological observation Dimensions of observational research definition of contrived artificial physiological measure control extraneous factors know the definition of direct and indirect observation Cons of observation research applicable to real world situation external validity Experimental research The disadvantage of one group pretest posttest design is extraneous variable are largely uncontrolled It doesn t even have a control group btw The disadvantage of static group is it has selection bias lack of randomization Matching one to one match can be used to replace randomization Randomization is crucial in posttest only control group design because it helps guarantee experimental and control group are equal understand extraneous variables testing effects history maturation and mortality especially the difference between maturation and history understand how external and internal validity is related to lab and field experiments Review of basic statistics Part I and II difference between population parameter and sample statistics Again is the value of parameter statistics known or unknown Again which one can be depicted by a smooth line Sample population This is very important I can t stress enough of it because he already mentions in the review session and the notes To understand why just read the section about review of inferential statistics The area under frequency distribution histogram can be viewed as probability It is an estimate of the probability distribution of a continuous variable The explanation of it can be found in Review of Inferential Statistics section Normal distribution mean mode median 50 above and below the mean Understand the symmetric rule of normal distribution and understand how to use it in calculations Under Standard Normal Distribution standard deviation and variance 1 Recall variance SD 2 Does Central Limit Theorem CLM apply to any kind of distribution YES I am not gonna explain what Central Limit Theorem is since it is taught in Stat1430 and it is in the book Use of graphical presentations MUST be on the test histogram interval ratio variables frequency tables bar charts pie charts nominal ordinal data open ended questions word clouds In word clouds if the size of words are larger it means they are mentioned frequently among respondents Review of inferential statistics Comparison points Depicted by smooth line 1 Properties of sampling distribution vs sample distribution vs population distribution I Population Yes e g the lines are like rectangular bars There are so many rectangular bars in the population distribution which constitutes a smooth line Population distribution does not ALWAYS have normal distribution If the population distribution is small i e 30 it is possible that it does not have normal distribution Sample distribution No There is no smooth line but just rectangular bars E g II It is hard to imagine what is the meaning of the number of rectangular bars can make a curve smooth One way that may help you this to picture the above rectangular bars breaking down into smaller rectangular bars You may be intuitively understand what is the meaning of this III distribution Yes ing Sampl Note that sampling distribution is NOT sample distribution Sample distribution refers to the fact that you actually collect a sample and it describes how its distribution is like Sampling distribution is you collect an infinite amount of samples and find out the distribution of x bar Theoretically speaking it equals to mean of X the population Thus it does have a smooth line How do you know when it is sampling distribution collect VERY LARGE amount of samples of the SAME size 2 Notation Note that for sampling distribution the standard error is 3 Under standard normal deviation variance 1 It can be tricky if you don t think it For sample distribution standard deviation is s carefully Remember variance the SD 2 4 Sample proportion Even quantitative data can have proportion distribution