Statistics Module 2

*problems with closed questions

-It forces an answer; and the most logical/correct response may not even be one of the choices --can avoid this by asking the question as an open-ended questions, then taking the most common responses to make a closed question (a pilot study) -magnitude not measured or is often misleading; ie: 65% of respondents believe going to class most important. Does this mean that study skills is not important??? -therefore, A Likert scale (1 2 3 4 5) or a ranking (out of 10) might be more accurate or produce better results

Open vs Closed Questions (Should Choices Be Given?):

-Open Question - a question (or statement) in which a respondent is allowed to answer in their own words. Eg. -'Do you have/please list' any other thoughts you have on what it takes to be successful as a student at Ohio University. Eg. -What are the most important attributes for successful students? -Closed Question -a question in which the respondent is given a list of alternatives from which to choose their answer Eg. -What is most important for academic success at Ohio University? a) study skills of the student b) going to class c) competence of the instructors/faculty d) money

Difficulties and Disasters in Sampling

-The following can occur even when a sampling plan has been well designed.

Defining A Common Language:

-Validity, Reliability, Bias, Variability, Natural Variability

Bias

-a biased measurement is one that is systematically off the mark in the same or in one direction. Consistently off the mark. -eg. A weight scale that was set too low at the factory would continuously score you a few pounds under your actually weight—this would be a biased view of your own weight. A weight scale that fluctuated between too high and too low (both directions) would be considered unreliable

Meta-Analysis

-a quantitative review of a collection of studies all done on a similar topic -when many studies are analyzed together -helps to try and find patterns or effects that are not conclusively available from individual studies eg: -comparing available research on AIDS from many sources and researchers and even over time (longitudinal)

Reliability

-a reliable measurement is one that will give you or anyone else approximately the same result time after time, when taken on the same object or individual. -consistency -physical measurements tend to be most reliable as long as you have the proper tools to measure eg. Height, IQ vs happiness, anxiety, stress -in fact, attitudes, emotions, etc. are not only hard to define but are typically less reliable eg. -A scale to determine the weight of a person -an IQ test that produces a score of 80 the first time and 130 the next would not be very reliable (in fact, it may not be valid either. Most commonly used IQ tests score within a couple of points is re-administered

Sample Surveys

-a subset of a large population is questioned on a set of topics -the results of this subset are used to make generalizations of the larger population -typically cost efficient, time efficient, accurate

Validity

-a valid measure is one that actually measure what it claims to measure eg. -you cannot measure anxiety or happiness using an IQ test. -how valid will your PSY 120 exam be—will it actually measure what it supposed to measure?

Case Studies

-an in-depth examination of one or a small number of individuals -the researcher observes and interviews that individual and any others who know about the topic of research eg: serial killer Ted Bundy -typically not used to make inferences about the population—just want to investigate a rare situation precisely

Systematic Sampling

-divide list into as many consecutive segments as you need. Randomly choose a starting point in the first segment, then sample at that same point in each segment. -often a good alternative to random sampling eg: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 -however, be careful—can lead to a biased sample

Stratified Random Sample:

-divide the population of units into groups (strata) and then take a random sample from each -can try to detect patterns/differences in strata -eg: -natural groups/regions in population -time zones; US West, Midwest, South, East -benefits: we can find individual estimates for each stratum, cost effective -can utilize different interviewers to conduct interviews in own region -sometimes more accurate estimates (especially if natural variability reduced)

Simple Random Sampling

-every member of the population has an equal chance of being selected into the sample -Need a list of all units of the population. Further, need a source of random numbers.

Multistage Sampling

-making use of a combination of sampling plans -ie: might stratify to a region then cluster

Random Digit Dialing

-many national polling organizations now use this -make a list of all possible telephone ' exchanges' (area code and first three digits). Using white pages, approximate the proportion of all households in country that have each exchange. Then have a computer generate a sample that has approximately those same proportions. Next, use the same method to randomly sample banks within each exchange. A bank consists of the next two numbers. Finally, to complete the number, the computer randomly generates the last two digits between 00 and 99

Experiments

-measure the effect of manipulating the environment in some way—then measure the result of the feature being manipulated on an outcome eg: receiving a drug, agreeing to a special diet, ie: does caffeine, tums, steroids, increase athletic performance -in most cases you have two groups and get volunteers randomly assigned to one of the two groups

Cluster Samples:

-much different, usually easier that stratified -divide population units groups (clusters). Select a random sample of clusters and measure each member of the selected cluster. -analysis will be different; since similarities may exist among the members of the clusters, and these must be taken into account eg: -to survey the opinions of its customers, an airline company made a list of all its flights and randomly selected 25 flights. All passengers on those flights were asked to fill out a survey.

Problems with open-ended questions

-results could be difficult to summarize eg . My dissertation -210 responses to two open-ended questions that I had at the end of my survey instrument -a lot of time, effort to categorize all of this information -some responses "I" had to interpret what they were trying to tell me

Observational Studies

-when we merely observe things about our sample (in an appropriate way) -the ̳manipulation' occurs naturally rather than being imposed by the experimenter eg: what happens to weight, blood pressure, when quit smoking

Margin of error

1 divided by the square root of n -*where n represents the number of people in the sample -the sample proportion differs from the population proportion by more than the margin of error less than 5% of the time. Eg: with a sample of 1600 people we usually get an estimate that is accurate to within 1/40 = 0.025 = 2.5% Therefore, the margin of error for this survey is plus or minus 2.5 percentage points. If 60% of the respondents support Bush's tax plan, this means that you would expect that between 57.5% and 62.5% of the entire population would support this tax plan. Only once in 20 times would you not get this interval result (following this same method)

Disasters in Sampling

1) Getting a volunteer sample -volunteer responses vs volunteer samples -relying on a volunteer sample is a waste of time -magazines and television stations are famous for this—ask if you have an opinion to respond to a certain question -however, studies have shown that highly opinionated people tend to answer these more; therefore, the responding group is NOT representative of some larger group *most media outlets now claim that their results are ̳unscientific' when reporting results 2) Using a convenient or haphazard sample -using the most convenient group or a group that you decide on the spot can also be worthless. Again, it may not represent any larger population eg: -asking students with 777/888 SSN's their aptitude with regards to American history (they are all international students)

Difficulties in Sampling

1) Using the wrong sampling frame (population from which the sample is drawn) -the sampling frame includes unwanted units and/or excludes desired units eg: using a telephone directory to survey the general population as to their opinions in an upcoming election; list may exclude numbers for teachers, doctors, and those to poor to afford a telephone; further, might include someone like ME—not eligible to vote 2) Not reaching the individuals selected -even if a proper sample of units is selected, the intended units/individuals may not be reached eg: -Consumer Reports sent to the subscriber of a magazine. This individual too busy therefore, someone else in the household fills it out -telephone surveys tend to reach a disproportionately high number of women 3) Getting no response or getting a volunteer response -response rates should be reported in the research summary. The lower the response rate the less the research can be generalized to the population -often a follow-up letter/phone call is necessary to get people to respond. However, excessive ' prodding' could produce an undesired response—someone who ompletes the instrument just to 'get it done'

Pitfalls encountered when asking questions on a survey or in an experiment:

1. Deliberate bias -appropriate wording should not indicate a desired answer eg. -Lawyers in court -Do you agree that..... - "Do you agree that abortion, the murder of innocent beings, should be outlawed?" 2. Unintentional bias -sometimes questions get worded in such a way that the meaning is misinterpreted by the respondents. Eg. -" But daaad, I did not mean it that way! -Any body in this class take drugs? -prescription, illegal, over-the-counter 3. Desire to Please -most survey respondents have a desire to please the person who is asking the question 4. Asking the Uninformed -people do not like to admit that they don't know what you are talking about when you ask them a question. Eg. - "Uh.....Hmmm....Yes, oh course I know about that!" 5. Unnecessary Complexity -if questions are to be understood, they must be kept simple eg. - "Shouldn't former drug dealers not be allowed to work in hospitals after they are released from prison?" 6. Ordering of Questions -if one question requires respondents to think about something that they might not have otherwise considered, then the order in which questions are presented can change the results eg. Q#1 - "To what extent do you think teenagers abstain from sex for fear of getting aids?" Q#2 - "List the top three reasons teens abstain from sex." **on your university exams, watch for this....sometimes questions later in the test can bring insight/answers for earlier questions** 7. Confidentiality and Anonymity -people sometimes answer questions differently based on the degree to which they believe they are anonymous eg. -―Are you sure my responses will be kept confidential?"

Seven Critical Components to a Good News Report

1. The source of research and funding-government vs private or independent companies -universities; companies (consumer behavior) 2. The researchers who had contact with the participants -who and what message was conveyed 3. The individuals or objects studied and how they were selected -volunteers; experts; paid?; possible biases? 4. Exact nature of the measurements made and the questions asked -MC; T/F; fill in blank/; wording eg. ' eat' breakfast 5. The setting in which measurements were taken -mail; phone; in person; anonymous?; the day before spring break (ie when and where) 6. The extraneous differences between groups being compared -extraneous (ie regular differences b/t people) differences may account for results 7. The magnitude of any claimed effects or differences -how large are the observed effects? (ie large differences or small differences)

Sampling Frame

Is a list of units from which the sample is chosen. Ideally, it includes the whole population

Unit

Is a single individual or object to be measured

Census

Is a survey in which the entire population is measured eg: Census 2000 USA

Sample Survey

Measurements are taken on a subset, or sample, of units from the population

Measurements, Mistakes, and Misunderstandings It's All in The Wording

One of the seven critical components of a good news report: #4 Exact nature of the measurements made and the questions asked -needs some further discussion -all/most of your research could be incorrect/misleading or may have measured something completely different than you intended to measure

Important Language:

Population vs Sample Unit Sampling Frame Sample Survey Census Margin of Error Margin of error

Common Research Strategies:

Sample Surveys Experiments Observational Studies Meta-Analysis Case Studies

Methods of Sampling:

Simple Random Sampling Stratified Random Sample Cluster Samples Systematic Sampling Random Digit Dialing

Margin of Error

The measure of accuracy of a sample survey

Variability

Unpredictable errors or discrepancies that are not easily explained eg. Height, blood pressure—measure one time and then immediately following and might have some slight differences in your measurements ie: 120/78; 118/80 -the dispersion of scores around a central point (the mean). Eg: individual exams scores vs. the class average

Natural Variability

Variability is inherent in nature. We are bound to get some variability because everyone is different. Need to know how much variability to expect due to natural causes