Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
124 Cards in this Set
- Front
- Back
the entire group under study as defined by research objectives
|
population
|
|
a subset of the population that is selected to represent the entire population group
|
sample
|
|
an accounting of the complete population
|
census
|
|
the individual “things” that make up the population. It is usually the basic level of investigation
|
sample unit
|
|
A way to identify population members, often a physical list of sample “units,” but sometimes being a method of selection not requiring an actual list.
|
sampling frame
|
|
due to a sampling frame that doesn’t include every unit it should or includes some that it should not.
|
sampling frame error
|
|
due to the random chance of getting sample members that are unusual in regard to the parameter being estimated
|
random sampling error
|
|
4 Reasons for taking a sample instead of a census:
|
1. Faster
2. More accurate 3. May not be possible 4. Cheaper |
|
those is which members of the population have a known chance of being selected into the sample
|
probability sampling
|
|
those where the chances of selecting members from the population into the sample are unknown
|
non-probability samples
|
|
the probability of being selected into the sample is equal for all members of the population
|
simple random sampling
|
|
2 Methods for drawing a simple random sample from a sampling frame:
|
1. Random Device Method
2. Random Numbers Method |
|
type of random sampling done for ease in selecting names
|
systematic sampling
|
|
type of random sampling where people within a cluster need to be heterogeneous and be a small model of the population; done for coest efficiency
|
cluster sampling
|
|
type of random sampling where people within a cluster need to be homogeneous; done for greater accuracy
|
stratified sampling
|
|
type of sampling method where some members of the population do not have any chance of being included in the sample
|
non-probability sampling
|
|
What is the main problem with non-probability sampling:
|
No way to judge the accuracy
|
|
When is it ok to use a non-probability sample:
|
Focus groups
|
|
samples drawn at the convenience of the interviewer
|
convenience sample
|
|
sample drawn that require an '"educated guess" as to who should represent the population
|
judgement sample
|
|
sample drawn that requires respondents to provide the names of prospective respondents
|
referral sample
|
|
sample drawn that establishes a specific percentage of the total sample - for various types of individuals to be interviewed
|
quota sample
|
|
What are the 3 things you must know or decide to calculate sample size:
|
1. Normal distribution
2. Standard deviation 3. Confidence intervals |
|
refers to the amount of error in your estimate of some characteristic of the population. It is how close a random sample’s statistic is to the true population’s value (a "parameter") that it represents
|
sample accuracy
|
|
refers to whether or not the amount of error in your estimate is based upon the sampling method
|
sample representativeness
|
|
Sample size ____________ depend upon the size of the population
|
does not
|
|
involves sample selection method and sample size
|
sampling error
|
|
pertains to all sources of error other than the sample selection method and sample size, such as problem specification mistakes, question bias, or incorrect analysis
|
non-sampling error
|
|
explains why we are able to state confidence intervals for our sample estimates
|
central limit theorem
|
|
the probability that an interval around the sample statistic which is defined by the confidence interval contains the true population value
|
confidence level
|
|
Why would doubling the size of a sample NOT cut the size of the sampling error in half?
|
Because of the nature of the curved relationship between sample size and sample error
|
|
Why is it a lousy idea to say that a sample, which is a certain percentage of the population, is adequate?
|
In almost all cases, the accuracy of a random sample is independent of the size of the population
|
|
The mean of a sample is a _________ while the mean of a population is a ___________.
|
statistic; parameter
|
|
We estimate a population's parameters from a _________________.
|
sample statistic
|
|
hypothesis that the difference in their population parameters is equal to zero
|
null hypothesis
|
|
How should we state the null hypothesis?
|
In a negative form we can clearly reject
|
|
What is the reason we must use a null hypothesis when we do hypothesis tests?
|
We always state the null hypothesis so that its rejection leads to accepting the desired conclusion
|
|
tells us that the values of a statistic when taken from all of the many possible samples which could be drawn from any population will have an average value which is equal to that of the population. All of the possible sample statistic values will be normally distributed around this true mean value. The standard deviation of a statistic such as a mean from this "derived population" is known as the __________________.
|
standard error
|
|
What 2 things are true about standard errors?
|
1. Get smaller with larger samples
2. Bigger when the variation of the population is bigger |
|
the degree of accuracy desired by the researcher and stipulated as a level of confidence in the form of a range with a lower boundary and an upper boundary
|
confidence intervals
|
|
5 Steps of Hypothesis Test:
|
1. Begin with a statement about what you believe exists in the population
2. Draw a random sample and determine the sample statistic 3. Compare the statistic to the hypothesized parameter 4. Decide whether or not the sample supports the hypothesis 5. If sample does not support hypothesis, revise the hypothesis to be consistent with the sample's statistic |
|
used to indicate the number of standard deviations of the statistic that an observed sample value is from the hypothesized value
|
Z
|
|
Z indicates how likely we are to see such a sample value if the null hypothesis is ______.
|
true
|
|
indicates the direction in which you believe the population parameter falls relative to some target mean or percentage
|
directional hypothesis
|
|
used to test the differences for large samples over the size of 30
|
Z-test
|
|
used to test the differences for small samples of under 30
|
T-test
|
|
With Z-tests and T-tests, there can be a _______________ difference which is so small it is unimportant to the marketer because no practical marketing decisions can be based on it.
|
statistically significant
|
|
2 Reasons that apply to the difference between two means:
|
1. Stable
2. Actionable |
|
the significance of a statistical test refers to the probability of seeing a particular sample statistic if the null hypothesis is _____.
|
true
|
|
If the significance of a statistical test is small (usually considered to be .05 or less) we usually decide to ________ the null hypothesis
|
reject
|
|
A t-test is used to test differences between ___ groups.
|
two
|
|
a sort of t-test which is applied when there are three or more groups
|
analysis of variance (ANOVA)
|
|
Cross-tabulations can be used to portray the association of two ______ level variables
|
nominal
|
|
Why is the chi square test the most useful statistical test for marketers?
|
Because cross-tabulations can be used to portray the association of two nominal level variables
|
|
used to show if the data in a cross-tabulation are associated in a non-directional way
|
chi-square test
|
|
Chi-square is calculated using ___________ in the cells of a cross-tabulation, not the percentages.
|
raw numbers
|
|
Chi-square requires the calculation of ___________ for each cell. Observed values are the original raw data.
|
expected values
|
|
If more than ___ of the cells in a cross-tabulation table have an expected value of _ or less, the chi square cannot be correctly calculated.
|
20%; 5
|
|
How can the possibility of having the chi-square fail be avoided?
|
By combining rows or columns of a table
|
|
With the Chi-square test, the ________________ is interpreted the same as for any other statistical test.
|
significance level
|
|
Chi-square test tells us the ______________ of an association but not whether or not it is _____________ to us.
|
significance; meaningful
|
|
A sample is a subset of a…
Known number of things Aggregate group Population A Representative Group |
Population
|
|
If G.E. wants to know how many hours an average light bulb will burn. Why would the firm be forced to use a sample and not a census?
Samples are faster Samples may be more accurate A census may not be possible (impractical) Samples are cheaper |
A census may not be possible (impractical)
|
|
Random Sampling Error is caused by the chance inclusion of particularly unusual units in your sample. This causes errors in...
Responses Representativeness Accuracy Timeliness |
Accuracy
|
|
Random sampling error can be fixed by….
Increasing sample size Getting better sample frames Sampling more carefully Training researchers |
Increasing sample size
|
|
Sampling Frame Error is caused by…
Outdated name lists Under-representative lists Over-representative lists Lack of a practical way to identify who is/is not a member of the population All of the above |
All of the above
|
|
Sampling Frame Error is not caused by…
Outdated name lists Under-representative lists Over-representative lists Lack of a practical way to identify who is/is not a member of the population Random Chance |
Random Chance
|
|
My wife and I won entry into a drawing for a $10,000 prize. In all, 13 families who'd won entry arrived at the drawing. So 13 keys--one of which would open the gate to the grand prize--were placed in a basket, and the families lined up. We believe that the last family in line had the best chance, and we were in 11th position, yet the prize was won by the family in 10th position. We never had a chance to draw a key. Which position do you think had the best chance?(Ask Marilyn, Parade magazine, GR Press, Oct. 29, 2006)
First Tenth Thirteenth All the same chances |
All the same chances
|
|
"Cluster Sampling" is used primarily to…
Save time and money Be more accurate Be more representative Insure a variety of respondents |
Save time and money
|
|
Stratified Sampling is used primarily to…
Save time and money Be more accurate Be more representative Insure a variety of respondents |
Be more accurate
|
|
When WOOD TV news asks you to call in your opinion about closing swimming pools in Grand Rapids, what type of sample are they getting?
Convenience Quota Judgment Referral |
Convenience
|
|
WMMS radio in Cleveland was voted the number one rock radio station in the country by Rolling Stone magazine readers. However, people who were knowledgeable about research suggested that WMMS had influenced the voting results by distributing 800 copies of the magazine (with ballots) to listeners and station employees. What type of sample actually resulted?
Simple random convenience judgment Quota stratified |
convenience
|
|
When a researcher wanted to develop a sample of pregnant women, he asked some of them to identify pregnant friends. This is known as a…
Cluster sample Referral sample Judgment sample Cluster sample Convenience sample |
Referral sample
|
|
When selecting a sample for a new car study, a researcher divided the entire sampling frame into males and females and randomly picked 100 men and 100 women This was what type of sample?
Cluster sample Referral sample Stratified sample Convenience sample |
stratified sample
|
|
If the average score on a quiz is 50, and the standard deviation is 10, about what percent of the students scores will be between 40 and 60?...
55% 68% 95% 99% |
68%
|
|
If our sample has 32% black M&M's in it, what is the approximate true proportion of black M&M's? , (with a 68% level of confidence, where Sp is .10 ?)
Between 22 and 42% Between 0 and 55% Between 12 and 52% We don't know |
Between 22 and 42%
|
|
What is the “confidence interval” in the estimate of some statistic from a sample?
Plus or minus the amount of error around the estimated value The amount of certainty you have that your estimate is correct The Z value The representativeness of your estimate |
Plus or minus the amount of error around the estimated value
|
|
How do you select a Z value for the sample size equation?
By selecting a confidence level By selecting a confidence interval By seeing what other studies have done You don't select it since the equation defines it. |
By selecting a confidence level
|
|
Assume that you want to estimate the proportion of a population (p) that has black hair. For which country would you need the biggest sample?(Assume you have no prior estimate of p)
USA (pop. = 301 million) China (pop.=1.3 billion) Same size for Both Can't determine |
Same size for both
|
|
Assume that you know that last year the percentage of people with black hair in China was 95% and in the USA it was 40%. Knowing this, in which country would you need the biggest sample?
USA China Same size for Both Can't determine |
USA
|
|
Estimating a Percentage: What is n in the previous slide?
Z=1.96 (95% confidence)p=42q=100-p=58e=5% or .05 (Note: is .0255 and so e= 1.96 * .0255=.05 500 374 255 196 |
374
|
|
If there are millions of possible samples (each with its own mean or proportion) taken from a population, how are these values distributed?
Each value occurs once We can’t tell until we look It depends on the population A normal distribution |
A normal distribution
|
|
When you want to estimate a proportion value for the population, what must you decide on before calculating the needed sample size?
Confidence level you desire Amount of error you will allow Standard error of the statistic Both #1 and #2 |
Both #1 and #2
|
|
The standard error of a statistic is really the ______ of the derived population of estimated values for all of the possible samples with that statistic from some parent population.
Frequency Median Standard deviation reliability |
Standard deviation
|
|
When U.S. courts try accused defendents, they are testing a null hypothesis. How should that null hypothesis be stated?
The accused is innocent The accused is guilty The accused might be innocent The weight of the evidence could go either way. |
The accused is innocent
|
|
The prosecution tried to convince the jury that it was impossible for footprints from shoes of the same brand and size as O.J.’s to be found at the murder site and for O.J. to be innocent. So the prosecution wanted the jury to believe that the chances of a true null hypothesis were what?
100% 50% 5% 0% |
0%
|
|
The defense argued that the police planted evidence in order to convict O.J. They raised the same issue, “how likely is it that footprints from shoes of the same brand and size could have been found at the murder site and O.J. be innocent? So the defense wanted the jury to perceive the probability of the evidence was what, given a true null hypothesis?
100% 50% 5% 0% |
100%
|
|
According to the prosecution, How likely is it that Nicole's blood could be found in O.J.'s Ford Bronco and O.J. be innocent?
100% 50% 5% <0% |
<0%
|
|
According to the defense, How likely is it that O.J.'s Akita attack dog could be at the murder scene and NOT bark at the "real killer," if it were a stranger?
100% 50% 10% <1% |
100%
|
|
The reason to state a hypothesis in “null” form is…
Most researchers are negative We only need one piece of evidence contrary to the null hypothesis to prove our desired hypothesis It is usually too difficult to find confirming evidence for a hypothesis The null hypothesis is usually exactly what we want to demonstrate |
We only need one piece of evidence contrary to the null hypothesis to prove our desired hypothesis
|
|
Is there a significant difference in the rating of GROW by women who did vs. did not start a prior business?
Yes…we are nearly 95% certain of this Yes...but there is only about a 5% chance that we are correct in this. No…the significance number is way too small We can’t tell for sure |
Yes…we are nearly 95% certain of this
|
|
Assume you want to know the average intelligence of GVSU marketing majors. You take a sample of the majors and get a mean of 116 and a standard error of the mean of 5 (s ). How do you express the result? (using approximate figures)
I am 95% confident the IQ is 116 +/- 5 I am 68% confident the IQ is 116 +/- 10 I am 95% confident the IQ is 116 +/- 10 I am 95% confident the IQ is 116 exactly |
I am 95% confident the IQ is 116 +/- 10
|
|
Assume that you meet a guy at a party who tells you that he has an IQ of 99 and he is a GVSU marketing major. You know that in the population of marketing majors, the average marketing IQ is 116 and the standard deviation is 10 (σ is 10) which is bigger than the standard error of the mean, σ =5). Is he telling you the truth about his being a marketing major?
Is this person’s score of 99 likely to have come from a population with a mean of 121 and a standard deviation of 10? (We wonder if he is either too smart or too “dumb” to be a marketing major) Yes, a very high chance Yes, but a low chance, though not so low that we reject the idea that he is a marketing major. No, It is nearly impossible. No, we know with 100 percent certainty he is not a marketing major. |
Yes, but a low chance, though not so low that we reject the idea that he is a marketing major.
|
|
If we want to show that the guy we met is NOT a marketing major, what is the null hypothesis?
The person is a marketing major The person is not a marketing major The person's IQ is too low to be a marketing major The person's IQ is too high to be a marketing major |
The person is a marketing major
|
|
Do we reject hypothesis that the guy is a marketing major?
No, not at the 95 % level of confidence Yes, at the 95% level of confidence We cannot tell whether or not he is a marketing major |
No, not at the 95 % level of confidence
|
|
What if we had believed that the guy we met with the 99 IQ is less intelligent than the average marketing major? (1 tail test)
We reject the Ho that he is as smart as marketing majors. At the 95% level of confidence, he is not as smart as average marketing majors He is most likely a marketing major, accept Ho. He is not that much less smart than the average major. |
We reject the Ho that he is as smart as marketing majors. At the 95% level of confidence, he is not as smart as average marketing majors
|
|
Assume that prior research has shown that 32% of all business school graduates in Grand Rapids are "low ceiling" professionals. Then we do a sample survey showing 26% of Seidman business graduates are low ceiling professionals. If we hypothesized that Seidman graduates are below average in their proportion of low ceiling professionals (a “one tail test”), are we correct?
What do we need to know to calculate Z? Standard error of the proportion Standard error of the mean The proportion on non-low ceiling professional The proportion of high ceiling professionals |
Standard error of the proportion
|
|
Assume that we find that the standard error of the proportion is 3.5%
So what is the Z value for testing our one directional hypothesis? 4.8 2.0 -1.71 .5 |
-1.71
|
|
So, our Z value tells us that the proportion of Seidman graduates with low ceiling jobs is…
Significantly below average (99% confidence) Significantly below average (95% confidence) Significantly different from average at the 95% confidence level. Significantly different at the 99% confidence level. |
Significantly below average (95% confidence)
|
|
Here is the difference between a one and two directional (two tail) test that is important. You are more likely to find a significant result with a one tail test. If your hypothesis were that the proportion of Seidman students was different (larger or smaller is not specified) and we had a Z of 1.71, would we say that there is a significant difference at the 95% confidence level?
Yes, if it is significantly greater there must be a difference No, the Z is less than 1.96 Yes…the differing z is arbitrary Excuse me while I air out my brain. |
No, the Z is less than 1.96
|
|
If I subtract from 2011 the year in which a respondent says his/her firm became a client of BDO, what type of data will result?
Nominal Ordinal Interval Ratio |
Ratio
|
|
What type of data does the following question produce?
2. Were you personally involved, either partly or completely in the selection of BDO as your organization’s accountant? Nominal Ordinal Interval Ratio |
Nominal
|
|
What type of data does the following question produce?
When BDO was selected as your organization’s accountant, what service promises by BDO influenced your choice? (open ended) Nominal Ordinal Interval Ratio |
Nominal
|
|
What type of data does the following question produce?
4. Which of these services does BDO supply to your organization? _____ Tax accounting _____ Auditing _____ Consulting _____ Other Nominal Ordinal Interval Ratio |
Nominal
|
|
How many variables are required to capture all of the information provided by this question?
4. Which of these services does BDO supply to your organization? _____ Tax accounting _____ Auditing _____ Consulting _____ Other One Two Three Four |
Four
|
|
What type of data does the following question produce?
5. “My organization is very satisfied with BDO’s services.” (strongly disagree to strongly agree on a 5 point scale) Nominal Ordinal Interval Ratio |
Interval
|
|
What type of data does the following question produce?
7. Would you personally recommend BDO to an executive you know at another organization? ___yes ___no Nominal Ordinal Interval Ratio |
Nominal
|
|
What type of data does the following question produce?
How many employees does your firm have in West Michigan? _______# Nominal Ordinal Inteval Ratio |
Ratio
|
|
How do you state the proper null hypothesis to test your supposition that the proportion of firms with more than 200 employees is over 80 percent? (one sample, one variable)
The proportion of firms with over 200 is 80%. The proportion of firms with over 200 is 80% or more. The proportion of firms with over 200 is 80% or less. The average number of employees is under 200. |
The proportion of firms with over 200 is 80% or less.
|
|
How do you state the proper null hypothesis to test your supposition that the average number of employees of West Michigan BDO clients is over 200. (one sample, one variable)
The average number of employees is 200. The average number of employees is over 200. The average number of employees is 200 or less. The average number of employees is under 200. |
The average number of employees is 200 or less.
|
|
The average number of employees is 319. Again treating our data as a sample, the standard error of the mean is 52.23 given by
Then Z= or 319 – 200 divided by 52.23= 2.28 With a Z of 2.28 should we reject the null hypothesis? (Use the one sample t-test for this) No Yes Can’t be determined |
Yes
|
|
Here is the SPSS reported chi-square. Can we reject the null hypothesis of no relationship?
Yes No Can’t tell |
No
|
|
Here is the SPSS reported correlation. Can we reject the null hypothesis of no relationship between overall service quality and satisfaction
Yes No Can’t tell |
Yes
|
|
So if domestic owners bought 3.22 cans of car wax while imported car owners bought 3.02 and the standard error of the difference between the two means, is .08, then what is the "Z" value of the test for the difference between the two groups of car owner?
.08 .01 2.50 .20 |
2.50
|
|
So with the Z value you calculated for the previous question, what do you conclude? The usage difference between domestic and import car owners groups is…
Not significant @ .05 Significant @ .05 We can’t determine Significant @ .05 only for a two tail test |
Significant @ .05
|
|
If the average usage difference between domestic and foreign car owners is statistically significant, an important consideration for managers whether or not the difference is large enough that it warrants treating the groups as different segments. In other words, is the difference…
logical Likely to really exist meaningful objective |
meaningful
|
|
The previous slides about car wax reported on a test of the difference between two…
proportions ordinal numbers means percentages |
means
|
|
Approximately at what "level of significance" is the difference in car wax buying between domestic and imported car owners when the Z=2.5? (you can get the answer from a Z table by looking at Z=2.5. Consider whether this is a directional (one tail) test or not. A Z table is provided on p.565.)
.0001 .005 .012 .05 |
.012
|
|
So consider this null Hypothesis, H0: There is no difference in the percentage of students who studied the book and passed an exam and those students who did not study the book and passed the exam. What is the alternative hypothesis if we believe that studying affects passing?
There will be a difference between “studiers” and “non-studiers” in the % of students who pass the exam. The % of students who pass will be higher for students who studied vs. the % of students who did not study. The average number of students who pass will be greater for those who study vs. the average number of those who do not study. The average amount of studying will be greater for students who passed the exam. |
The % of students who pass will be higher for students who studied vs. the % of students who did not study.
|
|
To test the previous hypothesis about the effectiveness of studying the book vs. not-studying the book, our best choice is to use a…
1 tail (directional) t-test 2-tail (non-directional) t-test correlation Chi-squared |
1 tail (directional) t-test
|
|
The resulting Z score is approximately what?
.06 6 .6 .10 |
6
|
|
By saying that the significance level of the test is .000, we are actually saying that
There is zero significance to our test The chances of getting the results we did and the null hypothesis being true are nearly zero The chances that we have accurate results are zero The chances of the alternative hypothesis being true are zero. |
The chances of getting the results we did and the null hypothesis being true are nearly zero
|
|
The significance of the previous chi squared test is shown as .000. This means the results are…
Not significant Highly significant Marginally significant Unknown, due to the zero values. |
Highly significant
|
|
So do we accept or reject the null hypothesis in the previous slide?
Accept Reject Tentativley accept the althernative Withhold judgment |
Reject
|