Statistics Final Exam

A researcher is analyzing a study with a treatment group and control group.  The sampling structure consisted of a random sample from the population.  From the random sample individuals were then randomized to receive either the treatment drug or the control placebo.  The outcome of interest is continuous.  Shapiro Wilk tests on the treatment and control groups have p-values of 0.0345 and 0.0732 respectively.  What type of test is most appropriate?

Wilcoxon rank sum (Mann-Whitney U test)

Which of the following are assumptions of the Wilcoxon rank sum test (Mann-Whitney U)?  Select ALL that APPLY

The two samples are independent

The two samples are dependent

Populations are normally distributed

Populations of one or both groups are not normally distributed

The distributions of the two groups are similar

Data are paired

The two samples are independent

Populations of one or both groups are not normally distributed

The distributions of the two groups are similar

Nonparametric tests have more assumptions than parametric tests, and that is why we use parametric tests more often. (T/F)

false

You run a test of variance between two groups for a t-test.  Prism gives you a p-value for the test of variance of 0.003.  You should use a Welch's correction. (T/F)

True

Data which are continuous, normally distributed, and from two groups are always analyzed using an adjusted unpaired t-test. (T/F)

False

A researcher has an analysis to complete.  Their outcome data is continuous.  They are interesting in comparing the outcome among a exposure group and a group a limited exposure group matched on age. The exposure is to an environmental hazard in the area.  The Shapiro Wilk test p-value was 0.0843 for one group and 0.3212 for the other group.    Which test is the most appropriate?

Paired T test

Which of the following ARE assumptions when using a unpaired t-test.  Select ALL that APPLY:

The two samples have equal variance

The populations have a normal distribution in both groups.

That data are matched

The groups are dependent

The two samples have equal variance

The populations have a normal distribution in both groups.

Which test is most appropriate?  The outcome is continuous, and there are two groups (group 1 (n=25) and group 2 (n=25)) of interest.  The groups are matched on age, sex, and race.  The histogram shows the non-normal distribution of both groups.

Wilcoxon sign rank test (Wilcoxon matched pairs)

True or False.  We have limited sample size (n=4 per group), and we want to test for a difference between groups 1 and groups 2 in a hypothetical research scenario.  We read that there is more power in one sample testing, and since we have a low sample size we decide to do this.  We are not sure of the actual direction of the difference of either group.  Was this the correct decision?

False

There is a hypothetical experiment that is testing a technique for decreasing anxiety.  The data collected are continuous anxiety test scores (higher the score the more anxiety) among two groups.  Participants were randomly assigned to the groups, and were not matched on any variables.  There are 30 individuals in each group.  Answer the following questions about the data set.

Are the data independent (unpaired) or dependent (paired).

independent: by randomized design

There is a hypothetical experiment that is testing a technique for decreasing anxiety.  The data collected are continuous anxiety test scores (higher the score the more anxiety) among two groups.  Participants were randomly assigned to the groups, and were not matched on any variables.  There are 30 individuals in each group. The data is normally distributed.

 

Which test is most appropriate?

Unpaired t-test with Welch adjustment

In this hypothetical experiment, respondents (n=15 in each group) were asked to rate their experience ( 1 being least, 10 being best) for a particular class.  One group of respondents had the course with no assistance.  The other group of students had multiple TA's and other online assistance.  The groups were matched on age and gender.  The instructor wants to analyze the data to see if the extra assistance in the course changed someones perception or experience in the course.

Answer the following questions:

Are the data independent (unpaired) or dependent (paired)?

dependent by design

In this hypothetical experiment, respondents (n=15 in each group) were asked to rate their experience ( 1 being least, 10 being best) for a particular class.  One group of respondents had the course with no assistance.  The other group of students had multiple TA's and other online assistance.  The groups were matched on age and gender.  The instructor wants to analyze the data to see if the extra assistance in the course changed someones perception or experience in the course. The data is not normally distributed.

What test is most appropriate?

Wilcoxon sign rank test (Wilcoxon matched pairs)

Which of the following tests use ranks of the data and not the data values?

 

One way ANOVA

T-test

Kruskal Wallis

Friedman Test

Repeated Measures ANOVA

Kruskal Wallis

Friedman Test

A researcher has a design that involves multiple statistical testing.  They run the analysis with and without a Tukey adjustment.  They should report the adjusted p-value in the literature, not the un-adjusted p-value. (T/F)

true

The null hypothesis from an one-way ANOVA tests is that all the means of all the groups are equal. (T/F)

true

Background:  Researchers think that addition of a specific hormone (HA) will increase the beneficial biome of the intestinal tract.

Design:  They propose three groups.  The control group does not receive HA but receives a placebo.  The two intervention groups receive a low dose and a high dose of HA respectively.  They measure 45 different beneficial bacteria species in the intestinal tract as outcomes.  They wish to determine if the groups differ among the 45 different bacteria species. 

Analysis Proposal:  They propose 45 different ANOVA models, and within each ANOVA model they are testing: Control-Low, Control-High, and Low-High.  (So they propose 135 different comparisons)  They set alpha = 0.05

 

What is the percent chance they will see at least one significant p-value by chance alone?

99%

Background:  Researchers think that addition of a specific hormone (HA) will increase the beneficial biome of the intestinal tract.

Design:  They propose three groups.  The control group does not receive HA but receives a placebo.  The two intervention groups receive a low dose and a high dose of HA respectively.  They measure 45 different beneficial bacteria species in the intestinal tract as outcomes.  They wish to determine if the groups differ among the 45 different bacteria species. 

Analysis Proposal:  They propose 45 different ANOVA models, and within each ANOVA model they are testing: Control-Low, Control-High, and Low-High.  (So they propose 135 different comparisons)  They set alpha = 0.05

 

Using the Bonferroni adjustment discussed in class what should these researcher set their alpha to?

0.00037

Background:  Lung cancer is a common oncology condition with high morbidity and mortality.  PET/CT scans are used to measure lung tumor activity.  There is a numeric measure (Standard uptake value (SUV)) that can be used on a PET/CT to estimate tumor activity level.  Lower SUV values mean less tumor activity and higher SUV values mean more activity.  In general, high SUV normally means worse prognosis, lower SUV means better prognosis (but not always).

 

A new treatment for lung cancer is being tested in mice.  Mice that have lung cancer have been selected for this experiment, and they have been randomly assigned to control (control), standard of care (Standard), and the novel treatment (noveldrug).  There are 10 mice in each group.  Researchers are interested in whether tumor SUV is lower among the novel drug group as compared to the control and standard of care.  The SUV values are from rodent PET/CT studies 2 weeks after treatment follow-up.

Assume:  alpha is set to 0.05

Answer the following question:

Which statistical test should be executed for this research question?

Analysis of Variance (ANOVA)

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 6-9 tab of the data set.

 

Background:  Lung cancer is a common oncology condition with high morbidity and mortality.  PET/CT scans are used to measure lung tumor activity.  There is a numeric measure (Standard uptake value (SUV)) that can be used on a PET/CT to estimate tumor activity level.  Lower SUV values mean less tumor activity and higher SUV values mean more activity.  In general, high SUV normally means worse prognosis, lower SUV means better prognosis (but not always).

 

A new treatment for lung cancer is being tested in mice.  Mice that have lung cancer have been selected for this experiment, and they have been randomly assigned to control (control), standard of care (Standard), and the novel treatment (noveldrug).  There are 10 mice in each group, and groups were not matched (yolked) in any fashion.  Researchers are interested in whether tumor SUV is lower among the novel drug group as compared to the control and standard of care.  The SUV values are from rodent PET/CT studies 2 weeks after treatment follow-up.

Assume:  alpha is set to 0.05

Answer the following question:

Report the appropriate test statistic and p-value from the Graphpad output.

F = 20.29, p<0.0001

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 6-9 tab of the data set.

 

Background:  Lung cancer is a common oncology condition with high morbidity and mortality.  PET/CT scans are used to measure lung tumor activity.  There is a numeric measure (Standard uptake value (SUV)) that can be used on a PET/CT to estimate tumor activity level.  Lower SUV values mean less tumor activity and higher SUV values mean more activity.  In general, high SUV normally means worse prognosis, lower SUV means better prognosis (but not always).

 

A new treatment for lung cancer is being tested in mice.  Mice that have lung cancer have been selected for this experiment, and they have been randomly assigned to control (control), standard of care (Standard), and the novel treatment (noveldrug).  There are 10 mice in each group, and groups were not matched (yolked) in any fashion.  Researchers are interested in whether tumor SUV is lower among the novel drug group as compared to the control and standard of care.  The SUV values are from rodent PET/CT studies 2 weeks after treatment follow-up.

Assume:  alpha is set to 0.05

Answer the following question:

Which of the below is the appropriate interpretation?

Yes there is a difference; Standard of care showed a 2.5 higher SUV compared to the novel drug; p=0.0054

Yes there is a difference; Standard of care showed a 2.1 higher SUV compared to the novel drug; p=0.0180

Yes there is a difference; The Novel drug showed a 2.5 higher SUV compared to the standard of care; p=0.0054

Yes there is a difference;  The novel drug showed a 4.6 lower SUV compared to the standard of care; p<0.0001

Yes there is a difference; Standard of care showed a 2.5 higher SUV compared to the novel drug; p=0.0054

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 6-9 tab of the data set.

 

Background:  Lung cancer is a common oncology condition with high morbidity and mortality.  PET/CT scans are used to measure lung tumor activity.  There is a numeric measure (Standard uptake value (SUV)) that can be used on a PET/CT to estimate tumor activity level.  Lower SUV values mean less tumor activity and higher SUV values mean more activity.  In general, high SUV normally means worse prognosis, lower SUV means better prognosis (but not always).

 

A new treatment for lung cancer is being tested in mice.  Mice that have lung cancer have been selected for this experiment, and they have been randomly assigned to control (control), standard of care (Standard), and the novel treatment (noveldrug).  There are 10 mice in each group, and groups were not matched (yolked) in any fashion.  Researchers are interested in whether tumor SUV is lower among the novel drug group as compared to the control and standard of care.  The SUV values are from rodent PET/CT studies 2 weeks after treatment follow-up.

Assume:  alpha is set to 0.05

Answer the following question:

True or False:  Is the following an appropriate interpretation for the difference between the control group and the standard of care group?

SUV values among the standard of care group (n=10, mean: 5.6, SD: 1.9) was compared to the control group (n=10, mean:  7.7, SD: 1.5) using one-way ANOVA.  The standard of care group  was 2.1 (95%CI:  0.3, 3.9) SUV units lower compared to the control group.  (Tukey p=0.0180)

true

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 10-13 tab of the data set.

Researchers are trying to determine if a certain protein helps grow hair in balding individuals.  A sample of balding people is taken and then that sample is randomly assigned to have the control treatment (nothing) for 3 weeks, or the protein treatment for 3 weeks.  After a documented and defendable washout time,  the individuals in the control group would then receive the protein and the individuals originally in the treatment group would go into the control group. Hair growth in mm was measured using the latest technology at an appropriate time point during the regimen.

The data in the data set consists of two groups: control and peptide. Assume alpha = 0.05.

Answer the following question:

Which test is appropriate for analysis based on the data?

Friedman’s test

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 10-13 tab of the data set.

Researchers are trying to determine if a certain protein helps grow hair in balding individuals.  A sample of balding people is taken and then that sample is randomly assigned to have the control treatment (nothing) for 3 weeks, or the protein treatment for 3 weeks.  After a documented and defendable washout time,  the individuals in the control group would then receive the protein and the individuals originally in the treatment group would go into the control group. Hair growth in mm was measured using the latest technology at an appropriate time point during the regimen.

The data in the data set consists of two groups: control and peptide. Assume alpha = 0.05.

Answer the following question:

The appropriate test statistic and p-value is:  F= 6.5, p=0.0118

false

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 10-13 tab of the data set.

Researchers are trying to determine if a certain protein helps grow hair in balding individuals.  A sample of balding people is taken and then that sample is randomly assigned to have the control treatment (nothing) for 3 weeks, or the protein treatment for 3 weeks.  After a documented and defendable washout time,  the individuals in the control group would then receive the protein and the individuals originally in the treatment group would go into the control group. Hair growth in mm was measured using the latest technology at an appropriate time point during the regimen.

The data in the data set consists of two groups: control and peptide. Assume alpha = 0.05.

Answer the following question:

Regardless of what test you performed, and the appropriate test selected above, suppose a statistically naive person just performed common one-way ANOVA on this data.  Does the protein change hair growth in that case?

Protein Does NOT change Hair growth, F=4.9 p=0.0791

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 14-17 tab of the data set.

Researchers have asked patients in a local medical center to rate their care from 1 - 5 (1 being terrible, 5 being wonderful).  Patients who register at the hospital out-patient departments are consented and then randomly assigned to have various specialized training and information on their procedure before having the procedure.  One group gets training with video about their upcoming procedure, and another group gets training through series of pamphlets.  The control group gets no specialized training.  Patients will only be in one group, and rate their procedure experience after the procedure is complete.  Assume alpha=0.05

Data set:  Control = control group, Pamphlets = pamphlet only group, and Video = video and pamphlet group.

 

Answer the following question:

Which test is appropriate for the data?

Kruskal Wallis Test

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 14-17 tab of the data set.

Researchers have asked patients in a local medical center to rate their care from 1 - 5 (1 being terrible, 5 being wonderful).  Patients who register at the hospital out-patient departments are consented and then randomly assigned to have various specialized training and information on their procedure before having the procedure.  One group gets training with video about their upcoming procedure, and another group gets training through series of pamphlets.  The control group gets no specialized training.  Patients will only be in one group, and rate their procedure experience after the procedure is complete.  Assume alpha=0.05

Data set:  Control = control group, Pamphlets = pamphlet only group, and Video = video and pamphlet group.

 

Answer the following question:

What is the appropriate test statistic and p-value?

34.94, p<0.0001

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 14-17 tab of the data set.

Researchers have asked patients in a local medical center to rate their care from 1 - 5 (1 being terrible, 5 being wonderful).  Patients who register at the hospital out-patient departments are consented and then randomly assigned to have various specialized training and information on their procedure before having the procedure.  One group gets training with video about their upcoming procedure, and another group gets training through series of pamphlets.  The control group gets no specialized training.  Patients will only be in one group, and rate their procedure experience after the procedure is complete.  Assume alpha=0.05

Data set:  Control = control group, Pamphlets = pamphlet only group, and Video = video and pamphlet group.

 

Answer the following question:

According to the output, were patients scores any different among those receiving the pamphlets and the controls?

 

Yes, scores were higher in those receiving the pamphlets compared to the controls; adj p-value:  0.6529

No, scores were no different between those receiving pamphlets and control;  adj p-value:  0.6529

Yes, scores were higher in those receiving the pamphlets compared to the controls; adjusted p-value: <0.0001

No, scores were no different between those receiving pamphlets and those seeing the videos;  adj p-value:  0.1235

No, scores were no different between those receiving pamphlets and control;  adj p-value:  0.6529

This question uses the data from the D2L Quiz 7 data.  You will need graphpad.  Refer to Question 14-17 tab of the data set.

Researchers have asked patients in a local medical center to rate their care from 1 - 5 (1 being terrible, 5 being wonderful).  Patients who register at the hospital out-patient departments are consented and then randomly assigned to have various specialized training and information on their procedure before having the procedure.  One group gets training with video about their upcoming procedure, and another group gets training through series of pamphlets.  The control group gets no specialized training.  Patients will only be in one group, and rate their procedure experience after the procedure is complete.  Assume alpha=0.05

Data set:  Control = control group, Pamphlets = pamphlet only group, and Video = video and pamphlet group.

Answer the following question:

The researcher concludes that the education material with the videos (n=19, med: 4.0, IQR: 2.0) is helpful for patient satisfaction, but paper based education (n=19, med: 2.0, IQR: 2.0) is not helpful, compared to control (n=19, med: 1.0, IQR: 1.0).  (Video adj p-value:  <0.0001, paper based adj p-value: 0.6529)

This is a correct interpretation. (T/F)

true

What is the overall problem with conducting multiple statistical comparisons?

inflation of alpha or type one error

Using the equation given in class:  100(1-0.95^c)

If a researcher designed an experiment that included three dosage groups of a peptide and a control (4 total groups), and they were interested in testing for differences among any group.  (They want to test all groups compared to one another)  What would be their resulting type one error %?

26.5%

You perform an appropriate one-way ANOVA test.  GraphPad gives you a p-value of <0.0001 on the over all ANOVA F test.  You were testing mean differences among 5 different groups.  This p-value indicates that ALL the groups are different from one another. (T/F)

false

To perform a Pearson's Correlation Coefficient or Linear Regression, only the explanatory variable needs to be normally distributed. (T/F)

false

A regression equation in the form of Y = -4.35X +25.4 would show a negative correlation. (T/F)

true

A researcher is interested if a new formula (NP45) will increase cell survival after exposure to a detrimental dose of a environmental toxin.  It is unknown how much NP45 to use so they design an experiment to have cells exposed to various randomly chosen levels of NP45.  All the cells are exposed to the LD100 dose of the environmental toxin.  The explanatory variable is the dose in mg of NP45, and the outcome variable is time (hours) until cell death.  They determine time in hours is not normally distributed.  Which test should they used to determine an association between NP45 dose and time until cell death.

Spearman Rank Correlation Coefficent

A regression equation results in Y = 1.75X +3.2.  Assuming you are not extrapolating data, what would the expected Y value be if a new X equaled 4.2?

10.55

A clinician is interested in prediction the resting energy potential for young children.  They task a research assistant with figuring out a way to achieve this.  The research assistant knows of an causal relation between age and gender and resting energy potential.  Which test should the researcher use to estimate a prediction equation for resting energy potential?

linear regression

A researcher collects an explanatory variable and an outcome variable (both continuous).  They plan on linear regression because they wish to predict the outcome based off the explanatory variable.  The range of the explanatory variable is from 16 - 39, and the range of the outcome variable is 48 - 88.  The results show a regression equation of Y = 1.72 X + 20.6.  A new observation has a X value of 5 and they want to use the equation to predict the outcome.  Which is correct?

 

The predicted Y is 29.2

They should not extrapolate as 5 was not in the range that created the model.

The predicted Y is -9.07

They need to use Spearman correlation coefficient instead

They should not extrapolate as 5 was not in the range that created the model.

Regression should be used for non-normally distributed data. (T/F)

false

A research assistant runs a Pearson Correlation coefficient on their data. All the assumptions of the test were valid.  They arrive at a correlation coefficient of 0.82.  Which of the following is correct about this correlation between the explanatory variable and outcome?

 

The explanatory variable causes the outcome.

The explanatory variable is negatively associated with the outcome.

The explanatory variable shows a weak positive association with the outcome.

The explanatory variable shows a strong positive association with the outcome.

The explanatory variable shows a strong positive association with the outcome

The following question involves the use of the 15 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 10-14 Tab Data and GraphPad.

The explanatory variable (exp) is the BMI of different patients from a cardiology clinic.  The outcome variable (outcome) is diastolic blood pressure of the patient at peak stress on a Bruce protocol exercise stress test.  A student is trying to show that there is an association between high BMI and higher diastolic stress blood pressure. 

The data from the Outcome variable is distributed normally. (T/F)

true

The following question involves the use of the 15 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 10-14 Tab Data and GraphPad.

The explanatory variable (exp) is the BMI of different patients from a cardiology clinic.  The outcome variable (outcome) is diastolic blood pressure of the patient at peak stress on a Bruce protocol exercise stress test. 

A student is trying to show that there is an association between high BMI and higher diastolic stress blood pressure.

Based on the testing of the data you have completed so far, which test should be conducted?

Pearson’s Correlation Coefficient

The following question involves the use of the 15 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 10-14 Tab Data and GraphPad.

The explanatory variable (exp) is the BMI of different patients from a cardiology clinic.  The outcome variable (outcome) is diastolic blood pressure of the patient at peak stress on a Bruce protocol exercise stress test. 

A student is trying to use BMI to predict the diastolic blood pressure at peak stress from an assumed causal association. 

What type of test should be run on this data?

linear regression

The following question involves the use of the 15 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 10-14 Tab Data and GraphPad.

The explanatory variable (exp) is the BMI of different patients from a cardiology clinic.  The outcome variable (outcome) is diastolic blood pressure of the patient at peak stress on a Bruce protocol exercise stress test.  A student is trying to show that there is an association between high BMI and higher diastolic stress blood pressure. 

A student is trying to show that there is an association between high BMI and higher diastolic stress blood pressure. 

Compute the appropriate test measure.  Which of the following represents the R estimate and p-value.

R=0.76, p=0.0010

R=0.45, p=0.0345

R=0.23, p=0.1235

R=0.88, p<0.0001

R=0.76, p=0.0010

The following question involves the use of the 15 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 10-14 Tab Data and GraphPad.

The explanatory variable (exp) is the BMI of different patients from a cardiology clinic.  The outcome variable (outcome) is diastolic blood pressure of the patient at peak stress on a Bruce protocol exercise stress test.  A student is trying to show that there is an association between high BMI and higher diastolic stress blood pressure. 

Based off the results you found:  Is there an association between BMI and peak stress diastolic blood pressure?

Yes, there is an association and it is a positive association.

The following question involves the use of the 8 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 15-17 Tab Data and GraphPad.

A researcher has collected data on patients that have come into a local health care facility in the last 4 hours.  They are recording the patient's overall feelings about the day (on a 5 point Likert scale).  A one indicates they are having a great day and a five indicates a terrible day.  They wish to see if this measure of happiness is associated with patient satisfaction scores (score).  The idea of this research stems from the question if health care providers have complete control over patient satisfaction.

Is the Likert variable normally distributed?

not normally distributed

The following question involves the use of the 8 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 15-17 Tab Data and GraphPad.

A researcher has collected data on patients that have come into a local health care facility in the last 4 hours.  They are recording the patient's overall feelings about the day (on a 5 point Likert scale).  A one indicates they are having a great day and a five indicates a terrible day.  They wish to see if this measure of happiness is associated with patient satisfaction scores (score).  The idea of this research stems from the question if health care providers have complete control over patient satisfaction.  (Set up Score as the outcome and Likert as the explanatory variable)

Which of the following tests is most appropriate?

Spearman Correlation Coefficient

The following question involves the use of the 8 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 15-17 Tab Data and GraphPad.

A researcher has collected data on patients that have come into a local health care facility in the last 4 hours.  They are recording the patient's overall feelings about the day (on a 5 point Likert scale).  A one indicates they are having a great day and a five indicates a terrible day.  They wish to see if this measure of happiness is associated with patient satisfaction scores (score).  The idea of this research stems from the question if health care providers have complete control over patient satisfaction. (Set up Score as the outcome and Likert as the explanatory variable)

Based on your analysis so far.  Report the appropriate test statistic and P-value.

R= -0.67, p=0.0786

The following question involves the use of the 12 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 18-20 Tab Data and GraphPad.

A research study focused on the prediction of a causal relationship between BMI (outcome) based on the number of hours per week spent exercising (explanatory variable).  Exercising was defined as completing any activity out side of work related activities for health reasons.  BMI was used in its raw continuous form and not categorized. It included walking, hiking, running, strength training, and yoga type activities.

Which type of test would achieve the goal of the research study?

linear regression

The following question involves the use of the 12 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 18-20 Tab Data and GraphPad.

A research study focused on the prediction of a causal relationship between BMI (outcome) based on the number of hours per week spent exercising (explanatory variable).   Exercising was defined as completing any activity out side of work related activities for health reasons.  BMI was used in its raw continuous form and not categorized. It included walking, hiking, running, strength training, and yoga type activities.

Report the appropriate results and p-values:

R² = 0.35, Y=-0.37X + 39.6, p=0.0442

The following question involves the use of the 12 observations from the D2L data set for Day 8 Quiz.  This question uses Quiz 8 Question 18-20 Tab Data and GraphPad.

A research study focused on the prediction of BMI based on the number of hours per week spent exercising.  Exercising was defined as completing any activity out side of work related activities for health reasons.  BMI was used in its raw continuous form and not categorized. It included walking, hiking, running, strength training, and yoga type activities.

Based on the results, what would be expected from someone with 5.5 hours of exercise?

BMI = 37.6

A chi-square test should be used with paired or dependent categorical data. (T/F)

false

Which of the following terms describes the ratio of the probability of an undesired outcome in one group compared to another group?

Relative Risk

A study on a new prevention treatment for coronary artery disease has participants randomly assigned to receive either the prevention or to receive a placebo.  The research team follows all these people over 10 years and determines if coronary artery disease is present after the 10 year follow up.  The following contingency table is the result: 

 

	CAD	No CAD	Totals
Treatment	20	93	113
Control	47	66	113
Totals	67	159	226

True or False:

After 10 year follow-up, the proportion of those receiving the prevention treatment who contracted CAD (20/113, 17.7%) was significantly different compared to the proportion of those who contracted CAD and did not receive the prevention treatment (47/113, 41.6%).  (X² = (some number), p<0.05)

Note, I did not want to give the exact test statistic value as it may give away other results on the quiz. (T/F)

true

A study on a new prevention treatment for coronary artery disease has participants randomly assigned to receive either the prevention or to receive a placebo.  The research team follows all these people over 10 years and determines if coronary artery disease is present after the 10 year follow up.  The following contingency table is the result: 

 

	CAD	No CAD	Totals
Treatment	20	93	113
Control	47	66	113
Totals	67	159	226

 

Copy this data into GraphPad.  Which of the following represents the appropriate p-value of the appropriate test?

0.0002

A study on a new prevention treatment for coronary artery disease has participants randomly assigned to receive either the prevention or to receive a placebo.  The research team follows all these people over 10 years and determines if coronary artery disease is present after the 10 year follow up.  The following contingency table is the result: 

 

	CAD	No CAD	Totals
Treatment	20	93	113
Control	47	66	113
Totals	67	159	226

 

All the observed cell counts are above 5 and none are under 1 so we do not need to compute expected cell counts, and we should just use chi-square. (T/F)

false

A study on a new prevention treatment for coronary artery disease has participants randomly assigned to receive either the prevention or to receive a placebo.  The research team follows all these people over 10 years and determines if coronary artery disease is present after the 10 year follow up.  The following contingency table is the result: 

 

	CAD	No CAD	Totals
Treatment	20	93	113
Control	47	66	113
Totals	67	159	226

 

Based on your calculations, what test is appropriate for this data?

Chi-Squared test

A study on a new prevention treatment for coronary artery disease has participants randomly assigned to receive either the prevention or to receive a placebo.  The research team follows all these people over 10 years and determines if coronary artery disease is present after the 10 year follow up.  The following contingency table is the result: 

 

	CAD	No CAD	Totals
Treatment	20	93	113
Control	47	66	113
Totals	67	159	226

Which of the following represents the EXPECTED cell count for those with CAD and were Treated?

33.4

Suppose the data below come from a study on a treatment on an outcome.  Use this data and perform a Fisher's exact test in GraphPad.  Report the p-value and if there is a difference between treatment and control.

	Outcome	No Outcome	Totals
Treatment	2	15	17
Control	4	13	17
Totals	6	28	34

p=0.6562, There is not a difference.

A researcher is planning an analysis.  Their outcome is an categorical variable and they are interested if this outcome differs among to classifications of cell types.   Based on this description what test should they use to determine differences in central tendencies of the outcome among the two classifications? (Note: the expected Cell Counts are all adequate)

Chi-square

A fellow researcher comes to you and wants your opinion on their study design.  They are using survival analysis, and they have not decided to power their study since they have access to 30,000 individuals of the population of interest.  Their outcome is death from a very rare disease (only about 0.75% of the population has/gets this disease).  Additionally, this disease is fatal 45% of the time.

Excluding other issues, based on the list below what would you tell them.

 

30,000 people is very large sample, you should be powered based off that.

Sample size does not drive survival analysis, but number of deaths/events do.  Their disease is very rare and not always deadly, so the number of deaths will be very low.  So they may be under powered.

As long as they use a log rank test it should be fine.

Where in the world did you get access to 30,000 people!?!  (dont pick this one its not the right answer, hopefully it made you smile)

"Sign me up for this study we are going to get published, 30,000 sounds great!"

Sample size does not drive survival analysis, but number of deaths/events do.  Their disease is very rare and not always deadly, so the number of deaths will be very low.  So they may be under powered.

Which of the following describes the type of data needed to conduct survival analysis with a Log Rank Test.

Time to event/death, Censoring Variable for all individuals within groups

Time to event/death only for all individuals within groups

Censoring Variable Only for all individuals within groups

Mean survival for everyone in all groups

Time to event/death, Censoring Variable for all individuals within groups

The underlying statistical test in a Log-rank test is a chi-square test that uses observed and expected values. (T/F)

true

Large population studies (like the strong heart study) follow very large populations of people over a long period of time to detect different outcomes.  Suppose a researcher is working on one of these studies.  They are looking to tell if there is a difference in a preventative treatment to help prevent death from a certain disease.  After a 30 year follow up, the following contingency table represents their data.

	Dead	Alive	Totals
Treatment	8	9,992	10,000
Control	9	24,991	25,000
Totals	17	34,983	35,000

What test is appropriate for this data?

Fisher’s Exact

A researcher is planning an analysis for survival data.  Assume they have the appropriate data and all assumptions are met.  People who die earlier matter more to this research question.  Which statistical test should they use?

Gehan-Breslow-Wilcoxon Test

A Fisher's exact test can be used for all sample sizes, where as Chi-square test can only be used under certain assumptions. (T/F)

true

A study on a new prevention treatment for coronary artery disease has participants randomly assigned to receive either the prevention or to receive a placebo.  The research team follows all these people over 10 years and determines if coronary artery disease is present after the 10 year follow up.  The following contingency table is the result: 

 

	CAD	No CAD	Totals
Treatment	20	93	113
Control	47	66	113
Totals	67	159	226

 

What is the probability of CAD given Treatment, and the probability of CAD given Control?  P(CAD|Treatment) and P(CAD|Control)

I want the answers in that order.

0.18, 0.42

The null hypothesis for a Chi-square test of independence is that the two categories being tested are dependent on one another. (T/F)

false

A researcher has a study where they have 4 novel drug dose groups and a control (no drug).  They are trying to determine which group has less cell death.  The data look like this:

Links to an external site.

	Cell Death	Cell Alive	Totals
Control	25	75	100
Dose Group 1	20	80	100
Dose Group 2	18	82	100
Dose Group 3	14	86	100
Dose Group 4	15	85	100
Totals	92	408	500

Assume that the requirements for a chi-square test are met.  If they run a chi-square test only (no other testing or residuals), they will know exactly which group is different from the others. (Also note:  No censoring variables this is not survival analysis.) (T/F)

false

This question uses the data that is on D2L for Quiz 9:  There are two questions related to this data, questions 19 and 20.

The data represent a sample from a study which examined number of days until dropping out of cardiac rehab.  Cardiac rehab is a nutrition and exercise regimen given to individuals who have had cardiac health problems like heart attacks.  The control group received the common cardiac rehab regimen.  The "treatment" group received an intense cardiac rehab regimen, which involved more intense exercise and highly monitored diet.  Before testing the effect the intense rehab had on cardiac issues, the researchers were interested in the feasibility of the new approach.  They felt like those under the intense rehab would drop out at a higher rate.  The research team was not concerned with early dropouts.  

Perform the appropriate test on this data to determine if the dropout is different among the two groups.  Which of the following is the test statistic and p-value related to that test?

X²=4.39, p= 0.0361

This question uses the data that is on D2L for Quiz 9:  There are two questions related to this data, questions 19 and 20.

The data represent a sample from a study which examined number of days until dropping out of cardiac rehab.  Cardiac rehab is a nutrition and exercise regimen given to individuals who have had cardiac health problems like heart attacks.  The control group received the common cardiac rehab regimen.  The "treatment" group received an intense cardiac rehab regimen, which involved more intense exercise and highly monitored diet.  Before testing the effect the intense rehab had on cardiac issues, the researchers were interested in the feasibility of the new approach.  They felt like those under the intense rehab would drop out at a higher rate.  The research team was not concerned with early dropouts. 

Which of the following represent the median dropout time for each group?

Control:  41 days, Intense Rehab: 26 days