Research Exam 2 PPT 1 PPT 2

Question 1

Standard Error of the mean

Answer 1

Is used to estimate the true mean of the population from which the sample was drawn

Question 2

CI

Answer 2

a range of values from the sample data that has a given probability of encompasing the true value

Question 3

95% CI is equal to 1-alpha (type I error)

Answer 3

95% CI for the estimated difference between two groups or within the same group over time does not include 0 and the results are significant at the 0.05 level

Question 4

CI for ratio rate

Answer 4

the line of no effect is 1 therefore if 1 is included is passed then it will be no statistical significance

Question 5

CI for differences

Answer 5

the line of no effect is 0 so if the interval and the value cross 0 then it will be no statistical difference but if 0 is not included than is statistical difference

Question 6

Ho

Answer 6

null hypothesis there is no difference between the two groups been compared

Question 7

Ha

Answer 7

alternative hypothesis there is difference between the two groups treatment A does not equal treatment B

Question 8

Reject the null hypothesis

Answer 8

Pvalue<alpha 0.05 and conclude that the alternative hypothesis is true at the 95% Confidence level

Question 9

Failed to reject Ho hypothesis

Answer 9

Pvalue > 0.05 and conclude that there is not enough evidence to say that the null hypothesis is false at the 95% confidence level

Question 10

Type 1 error

Answer 10

reject the Ho when Ho is actually true
is at the 0.05 alpha p-value
there is no true difference between the two groups

Question 11

Type 2 error

Answer 11

failed to reject Ho when Ho is false and there is actually a difference BETA

Question 12

P-value

Answer 12

probability a number between 0 (it wont happen) and 1 (it will defenitly happen) that describes the frequency of an outcome
the probability of our experimental results take into account the H0
-P-value is probability that our observed results are due to chance

Question 13

Independent (unpaired) T-Test

Answer 13

data is parametric is continuous is comparing two independent variables

Question 14

Paired T-test

Answer 14

compares the mean difference of paired matched samples

Question 15

ANOVA

Answer 15

comparing three different groups is parametric and continous data
so ex three different doses of aspirin with the birth weight

Question 16

Simple linear regresion

Answer 16

one continous independent variable and one continous dependendent variable

Question 17

Multiple linear regresion

Answer 17

one continous independent variable and two or more continous dependent variable

Question 18

You are designing a new alert system at the hospital to investigate the impact of several factors on the risk of corrected QTc prolongation. You want to create a model, assessing several variables, to predict which patients are most likely to experience QTc prolongation after the administration of certain drugs or the presence of certain conditions.

Which statistical technique will be most useful in completing such an analysis?

Answer 18

Regresion

Question 19

Multiple logistic regression

Answer 19

1 categorical independent
1 categorical dependent
1 confounding variable

Question 19

Simple logistic regression

Answer 19

for categorical discrete data
1 categorical independent variable
1 categorical dependent variable

Question 19

non parametric statistic

Answer 19

-Not for estimating parameters
-Does not requires a large sample size
-Is for nominal or ordinal data
-No assumption on data distribution
-Can be used for continous data if the sample size is small

Question 19

Investigators would like to determine if a new statin decreases the number of elderly subjects who experience a myocardial infarction compared to standard therapy after considering the influence of a number of confounding variables. Which of the tests below is best?
A. Student t-test
B. Multiple linear regression
C. ANOVA
D. Multiple logistic regression

Answer 19

Multiple logistic regression

Question 20

Parametric statistic

Answer 20

Parametric statistic:
-Requires a large sample size
-For continuous data
-Can make inference of parameters
-Data can be normal or non normal

Question 21

Chi square

Answer 21

non parametric
nominal data
large sample size
Ho assumes no particular relation between outcome and exposure

Question 22

Fisher exact

Answer 22

non parametric nominal data sample doesnt need to be large non paired

Question 23

Researchers planned a study to evaluate the percent- age of subjects who achieved less than a target blood pressure (less than 140/90 mm Hg) when initiated on two different doses of amlodipine. In the study of 100 subjects, the amlodipine 5-mg group (n=50) versus the amlodipine 10-mg group (n=50) were compared. The investigators used blood pressure goal (i.e., the percentage of subjects who successfully achieved the blood pressure goal at 3 months) as their primary end point. Which one of the following is the most appropriate statistical test to answer the question? A. Independent samples t-test.
 B. Chi squared C. Wilcoxon signed rank test.
 D. Independent t-test.

Answer 23

Yes they did achieve/ no they didnt achieve B chi squared

Question 24

A randomized control trial is comparing drug X and drug Y to measure the efficacy for treating cancer-associated nausea and vomiting. The end point of the trial is based on patients’ ranking their nausea and vomiting 2 hours after taking the drug that was assigned to them. The ranking is based on the following scale: 0=no nausea, 1=mild nausea, 2=moderate nausea, 3=severe nausea, 4=vomiting, 5=severe nausea and severe vomiting Which of the following statistical tests is the most appropriate to perform to assess the difference in nausea and vomiting severity between the two drugs? a) Wilcoxson signed rank test B) Paired student t-test C) Wilcoxson rank sum test D) Independent student t-tes

Answer 24

C) Wilcoxon rank sum test

Question 25

Kruskal Wallis

Answer 25

Similar to one way ANOVA Compared 3 groups with one dependent variable but is categorical , discrete non parametric, non normal distributed -Is used to measure the sample means if they are identical -Requires 5 observations

Question 26

Suppose weights of poplar trees are different based on treatments (none treatment, fertilizer, irrigation, or fertilizer & irrigation). Each weight sample determined by the treatments is independent and random, and each sample size is 5; yet, the weight samples are not normally distributed.

Answer 26

Kruskal Wallis

Question 27

Kaplan Meir method

Answer 27

Uses the survival time to estimate the proportion of people that survive under the same circumstances in a given length of time uses a table and graph

Question 28

Cox proportional hazards

Answer 28

Incidence rate vary over time are commly referred as hazard rates IR based on follow up time contibuted by each person in the study

Question 29

Population

Answer 29

a group of individual with the characteristics of interest of the researcher All people with characteristics All people in defined setting

Question 30

Target population

Answer 30

population to which one would like to make an inference

Question 31

Source population

Answer 31

population to which the study subject are drawn from

Question 32

Sampling frame

Answer 32

subset of the population from which to sample from at a specific time frame

Question 33

Study sample

Answer 33

participants actually in the study

Question 34

Simple random sampling

Answer 34

randomly select individuals from the sampling frame is useful when the target population is small and the sampling frame is done -Assign numbers to pts -Computer randomly select the pt -Lottery method

Question 35

Systemic sampling

Answer 35

Random -Use specialized intervals from which the subject are drawn from into the sample -Disadvantage is the natural periodicities in the population -Allows investigators to predict and manipulate the selection of the sample

Question 36

Stratified random sampling

Answer 36

The population is dividided into subgroups and via randomly selection individuals from each subgroup -Ensures higher precision compared to simple random sampling -Allows fro more control and precision

Question 37

Cluster sampling

Answer 37

is a random sampling of natural grouping individuals in the population - is based on geographic location -Can be divided into single stage, double stage or multistage -Sampling frame may not be possible in a large population

Question 38

researcher is interested in studying the use of melatonin among adults with hypertension in a primary care clinic. The researcher divides the sampling frame into 4 subgroups based on specific age groups (i.e., 30 to 39 years old, 40 to 49 years old, 50 to 59 years old, and 60 years or older). The researcher then randomly samples 100 patients with hypertension from each subgroup. Which of the following best describes this sampling technique? A) Convenience sampling B) Quota sampling C) Stratified random sampling D) Cluster sampling

Answer 38

Stratified random sampling

Question 39

Non random sampling

Answer 39

Individuals does not have equal chance of being selected -Unable to make inferences of the population from the sample

Question 40

Convenience sampling

Answer 40

Non random sample The pt is there is accessible No time consuming and easy to perform

Question 41

Quota sampling

Answer 41

Non random sample but is based guided in specific characteristics like race,sex,economic class\ -Ensures inclution of a particular segmented of the population into the sample -Researcher set stratas within the sample

Question 42

Consecutive sampling

Answer 42

Non random sample the subject that statisfie the characteristics of the study they wll offer them to it -Is useful for the prediction of the number of patients from which the sample size has been selected -concern of seasonal trends

Question 43

Power analysis

Answer 43

Using predetermined alpha a, investigators will try to determine and achieve the desired B by choose a sample size to detect the clinically/significally meaningful effect

Question 44

Calculating sample size using power analysis (determining specific formula to use)

Answer 44

-Knowledge of how the primary end point is measured (type of data) -Type of hypothesis test

Question 45

Calculating sample size using power analysis (information input in the formula)

Answer 45

1-Measure variability and precision 2-Specification of the magnitude effect 3-Stated level of significance (a) 4-Target level of power (1-b)

Question 46

Power analysis (measure of variance)

Answer 46

More precise less variation smaller sample size More variation larger sample size Variability represented based of the type of data Continous data outcome--> population sd Categorical dichotomous outcome-->depends on the proportion of the control group compared with the treatment group

Question 47

Power analysis (magnitude of effect)

Answer 47

Clinically relevant treatment effect that the study should be able to detect Difference depends on the type of data Continous outcome-->Difference between the two population means Dichotomous outcome-->difference between the proportions -Smaller effect size larger sample size needed to detect the difference -Researchers should be able to choose effect size based on judgment, experience and expertise -Need to be both clinically and statistical significant

Question 48

Power analysis (significance level)

Answer 48

alpha a should be set priori to the study during the study planning Conventionally is 0.05 other values is 0.10 and 0.01 As the alpha increases the sample size needs to decrease to detect the difference

Question 49

Power analysis (target level of power)

Answer 49

Increase the power will increase the sample size to detect difference -1-B is detect prior to the study during the planing -Generally is 0.8-0.9 (90%)

Question 50

Suppose you are an investigator interested in evaluating the difference in glucose levels with a new diabetes medication compared to placebo. Based on past literature, you determine a clinically meaningful difference in the glucose level as 20 mg/dL and an assumed standard deviation of 75 mg/dL. The required total sample size for this study with α of 0.05 and power of 80% is calculated to be 442 patients. Which of the following changes to the assumptions would increase the required total sample size (assuming the all other assumptions will remain the same)? A)Reduce standard deviation from 75 mg/dL to 60 mg/dL B)Decrease power from 80% to 75% C) Increase alpha from 0.05 to 0.10 D)Reduce effect size from 20 mg/dL to 15 mg/dL

Answer 50

D) reduce the effect size

Question 51

Sample size increases

Answer 51

1-Smaller effect size 2-Decreasing a 3-Increasing Power 4-Increase variability

Question 52

Sample size decreases

Answer 52

1-Larger effect size 2-Increasing a 3-Decreasing power 4-Decreasing variability

Question 53

Association

Answer 53

If there is a statistical link between the health outcome in patients with a specific exposure

Question 54

Causation

Answer 54

The exposure causes the health outcome it can be to presence of adverse exposure or absence of preventative exposure

Question 55

Artificial (spurios) association

Answer 55

Arises bias in the study -Biases in method selecting the case/control Bias in recording the information Bias in design the study - non representative study group Bias in the conduct of the study- loss of follow up, observe bias, measurment errors

Question 56

Non causal association

Answer 56

The confounding factor causes both but there is no causal relationship between both Ex cigarette smoking causes increase in drinking coffe and increase risk of CHD mortality

Question 57

Causal association

Answer 57

Causality is inferred -One factor causes the disease -The removal is shown to decrease the frequency of the disease -Not all exposure lead to the outcome -Multiple combination of exposure is likely to lead to the outcome

Question 58

Counterfactual definition of Cause in Pharmacoepidemiology

Answer 58

Ideal world: cause and effect of drug on a outcome -Same pt -Same time -Same place -Conditions are the same and see if they got the disease when they took the drug vs when they didnt

Question 59

Conterfactual Ideal in Pharmacoepidemiology

Answer 59

Real World: The fact (exposed group take the drug) vs the Proxy for counterfactual (not take the drug) and evaluate the associations between the drug on the outcome -The proxy needs to have similar characteristics as the exposed group

Question 60

In pharmacoepidemiological studies, which of the following scenarios best illustrates the counterfactual ideal? A) Patients taking Medication W are compared to a group of patients with similar characteristics who are also taking the same medication. B) Patients taking Medication Z are compared to a group of patients with different characteristics who are also taking the same medication. C) Patients taking Medication X are compared to a group of patients with similar characteristics who are not taking the medication. D) Patients taking Medication Y are compared to a group of patients with different characteristics who are not taking the medication.

Answer 60

C) Pt taking medication X are compared to patient with similar characteristics that are not taking the medication