3280 Ch. 7-9

• Based on your personal experience, why would you participate, or not participate in a survey?

I would because now i understand how much true samples from surveys can help research for a company or person

• I mailed out a lot of surveys, but very few came back. What could I have done to increase the response rate?

Increase Incentives or number of surveys sent

• National and local news media often report the results of professionally conducted national surveys on topical issues. Many also encourage readers and viewers to write in or call in with their opinions on topical issues. How do you think the opinions captured by professionally conducted national surveys and by "write in" or "call in" surveys might differ in nature? If you were a legislator trying to make a policy decision on both kinds of public opinion data, which would you prefer?

I think people could be swayed by media and and write in however i think people might be say more on the phone because if they are typing it they might tone it down but if they are on the phone a tone can be observed. this can help answer important questions a researcher might have on different issues. some people are more willing to voice their opinion if they feel they are being heard then just a survey online.

• Would you trust or not trust the results of surveys you have participated in or been invited to participate in? Why?

Trust because i am truthful. however anonymity plays into how honest people are in most cases

• I want to get the opinions of just 10 people; can I use a survey?

Yes

• Is the Internet a good way to survey people?

Yes

• Will talking to fellow students give me the best insight on students' opinion?

Yes, Purposive or judgmental sampling: Sampling based on specific criteria the researcher may have.

• How many people or media samples should you study to get credible results??

400 at a 95% confidence level with an error factor of +/- 5% and 665 at a 99% confidence level with an error factor of +/- 5%

• How do I decide which statistic is the most appropriate?

Confidence Interval The possible range of values for a variable in a population, calculated from a sample of the population. Confidence Level The probability of a calculated value occurring. Example - There is a 95% probability that the mean lies between 0.98 and 10.42. Parametric Statistics Used where we can assume a normal distribution of values in the population. Statistical Significance Significance means that there is a better than random chance that a relationship exists.

• Don't people just ignore most surveys?

Depends on the survey and how it is administered

• How can I know my public opinion survey really has captured public opinion?

Inferential Statistics:Using the sample to project to the population; Inferential statistics (inductive statistics): Set of statistical procedures that allow a researcher to go beyond the group that has been measured, and make statements about the characteristics of a much larger group. Estimation: Generalize the results obtained from a sample to its "parent" population Significance testing: How likely differences between groups and relationships between variables occur by chance Use the results from the sample to describe the population. Estimating population characteristics (called PARAMETERS) on the basis of characteristics found in a sample. Estimation procedures: PARAMETRIC STATISTICS Assumptions Allowing Estimates of Population Parameters from Sample Statistics:Normal distribution;Random sample Each population member has an equal chance of being selected for the sample Central Limit Theorem: If the sample is large enough, the distribution is almost always approximately normal. If you have to choose between random sample versus larger sample size, ALWAYS CHOOSE RANDOM SAMPLE How close you are to correct. Also called the CONFIDENCE INTERVAL If I have a 5% margin of error, that means that my answer is within -5 or +5 percentage points of the "true" answer in the population. The size of the confidence interval is influenced by: The variability that exists on the topic in the population being studied The confidence level used The size of the random sample

• Why is the normal curve an important concept?

Inferential statistics assume that the values of a variable in a population are normally distributed. Assuming a normal distribution of values in a population, we can calculate the probability that a sample of that population has captured its characteristics. In a normal distribution - 68% of values occur plus or minus one standard deviation (SD) from the mean. 95% of its values occur plus or minus two SDs from the mean. 99.7% of its values occur plus or minus three SDs from the mean.

• What unique insights do inferential statistics provide?

Inferential statistics help us - Estimate the probability that a sample represents the population it came from. Decide whether groups differ significantly on a variable. Decide whether there are significant relationships among or between variables. Inferential statistics assume that the values of a variable in a population are normally distributed. Assuming a normal distribution of values in a population, we can calculate the probability that a sample of that population has captured its characteristics.

• "Data mining" is the statistical analysis of large amounts of data to find relationships within the data. For example a video rental company might use data mining to recommend movies to individual customers based on their rental history. How accurate do you sense such data mining predictions can be? What might make them more or less accurate with respect to predicting specific behaviors?

It might be fairly accurate, A confidence interval might help strengthen results

• I need to ask some very personal questions; is a survey the best way to get such information?

No

• I want people to respond to questions in their own words; will a survey let them do this?

No

• Can statistics provide results in which I can have 100% confidence?

No, only 99% confidence

• Does a random sample of people mean that each person in the sample will be uniquely different?

Sampling in which every member of a population has an equal chance to be selected and in which selection is determined by "luck of the draw" rather than a decision by the researcher.

• What does random sampling really mean?

Sampling in which every member of a population has an equal chance to be selected and in which selection is determined by "luck of the draw" rather than a decision by the researcher.

• Most of us often get "pop up" invitations to participate in a brief survey when we are on line. How do you think we get selected? Do you think such surveys are legitimate research or just a sales tool and how might you decide which is which?

Significant: Important/ Momentous "How is 'x' related to other variables?" Inferring relationships between variables within a population on the basis of relationships found in the sample Parametric Statistics: Significance tests that assume that the variable is normally distributed in the population. Nonparametric statistics: Significance tests that do NOT assume this Is the difference statistically significant? Difference Analysis: Used to see if there is a difference between groups of people/texts Independent variables Nominal categories Correlations Statistical Relationships Between Variables Measure of Association Correlation Coefficient Ranges from -1 to +1 0.00 means 2 variables are unrelated

• Will a survey of 1,000 people provide more insight on human communication than interviews with 10 people?

The best way to reduce the size of the confidence interval is to increase the size of the random sample selected. LAW OF LARGE NUMBERS: The smaller the sample size, the less accurately it represents a true population parameter. Most researchers use smaller samples and accept larger confidence intervals in their population estimates

• What does statistical significance really mean?

The probability that a computed statistic such as a t test or correlation is not due to chance.

• Your local television station runs a survey on a current hot issue on its web site. How would you assess the quality of information obtained from such a survey? To what extent would you be prepared to make generalizations from the results you see reported?

You could see the demographics from who they are getting responses from and the types of questions being asked to asses whether or not they would be valid results Cross-sectional surveys Surveys taken at one point in time, as opposed to trend studies. I would look at the population size and the amount of homogeneity reported in the study and then if there was a great deal i could most likely make generalizations about the results reported

• Which of the following would you place most faith in - a calculated statistical probability that you have a significant research finding or an intuitive, personal conviction that you have an important finding?

a calculated statistical probability that you have a significant research finding

• Inferential statistics frequently depend on the assumption of a normal distribution of data. Based on your experiences with, for example, GPAs, test scores, commuting distances or time spent online how valid is the assumption of a normal distribution in everyday life?

extremely valid

• Often we get "pop up" invitations to participate in a brief survey when on line. This can be effective in some cases and ineffective in others. Think about why this might be true.

most people ignore popups or might not be as honest if it doesn't mean anything to them.

• Might careful sampling produce a better reading of a population than a census, which enumerates an entire population?

you could get a closer look at the population you are actually wanting to see other then the entire population which might merit results that are not needed in the study