Statistics Flashcards
(111 cards)
1
Q
95 percent of all scores are within how many Standard deviation ?
A
2 SD of the mean
2
Q
A retrospective study that looks for the odds of a previous exposure on the development of a “ rare disease manifestation “ refers to ????
A
Case control studies
3
Q
If you DO have a negative TEST, what is the likelihood that you DON’T have the disease ?
This question responds to the concept of ??
A
Negative predictive value
4
Q
TP + TN divided by TP + FP + FN + TN ( this formula refers to ? )
A
Accuracy
5
Q
Which number is the mode : 1-2-3-4-8-8-8-20-100
A
8 because is the most freq measurement
6
Q
Gives an indication of how precise a given collection of data is
“ this refers to the concept of ??”
A
Confidence intervals
7
Q
Fill the blank
The lesser the prevalence of a disease, the greater the —
A
NPV ( negative predictive value)
8
Q
In descriptive statistics when given a set of data points, the most frequently appearing measurement is called ??
A
Mode
9
Q
The data point half way between the highest and the lowest in the collection of measurements is called
A
median
10
Q
The smaller the SEM, the more
( blank ) the data is
Fill the blank
A
Precise
11
Q
Mention the type of error
“ Saying a drug or a test helps or makes a difference when it really doesn’t “
A
Type l error ( alpha error ) ( false positive like you know who )
12
Q
Mention the type of error
“ Rejecting the null hypothesis when it really is true “
A
Type l ( alpha error ) ( false positive like you know who )
13
Q
The following example is a type of data gathering called …?
Glucose is in milligrams PER deciliter
Heart rate is beats PER minutes
A
Ratio
14
Q
The following formula refers to ??
a / a+ b divided by c / c+ d
A
RR ( relative risk )
15
Q
(Blank blank ) looks at the risk of a disease based on who was exposed to a potential danger in the past . Belongs to cohort studies. Fill the blank !!
A
Relative risk ( starts with the risks, then looks for the disease ) . Starts with an asymptomatic group and calculates the comparative risk of developing a disease either with the exposure or without it
16
Q
The following set of data points is skewed to the ?
1-2-3-4-8-8-8-20-100
A
Right
Because the single outlier 100 extends the “range” of the data set to the right
17
Q
Mention the type of error
“ Rejecting the alternative hypothesis when it’s actually true “
A
Type ll error or beta error ( false negative ) . Lazy investigator who agrees with the null hypothesis because he is too lazy to work
18
Q
For step 2 CK with a mean of 240 and SD of 18, getting a 254 on the exam gives you a Z- score of ….?
A
+ 0.78 ( because 14/18 = 0.78 ) . Notice that the SD is 18 but you only got 14 points above SD so you must divide
19
Q
All the people with a disease should have a positive test if the test is …..?
A
Sensitive
20
Q
Berkson bias is solved by ??
A
Random selection
21
Q
Mention the type of error
“ Saying there is no statistically significant difference in the data when there really is “
A
Type ll error or beta error ( false negative) . Lazy investigator who agrees with the null hypothesis because he is too lazy to work
22
Q
The 3 studies that assess various groups are called ?
A
1- ANOVA
2- T-test
3- Chi- Square
23
Q
What test answer the question : “Are the means between these groups different ?”
A
T-test does
24
Q
Select the median : 1-2-3-4-8-8-8-20-100 . Explain why ?
A
8 (because is the fifth of the 9 data points which is exactly in the middle )
25
The following statement is an example of ??
| “ there is no greater difference between the new drug and placebo than would occur by random chance “
Null hypothesis
26
Starts with people with the “disease” and then looks backward at other groups that are otherwise matched to assess for risk of exposure
Refers to ???????
Case control studies
27
In diabetic people if you plot : the rate of weight loss , exercise and glucose control against the rate of new end-organ damage . The result will be —
A strong inverse or negative correlation ( -1 )
28
Is the following set of data points normally distributed ?
1-2-3-4-8-8-8-20-100
This is skewed to the ......?
No it is not because 7 out of 9 data points are under 10. It is skewed to the right. It is skewed positively towards 100
REMEMBER that skewed to the right is when the tail of the bell goes to the right
29
Mention the type of error
| “Saying the drug doesn’t work when it really does “
Type ll error ( beta error ) ( false negative )
30
Put an example of “ incidence “ being part of disease frequency as a ratio in the population
Ex : there is one new case of multiple sclerosis for every 1000 people in the population EVERY YEAR
31
The rate and which new disease occur measured in the number of cases PER Unit of time is called ?
Incidence ( 20 new diabetics per week )
32
The likelihood that a test will detect all people with a disease responds to ??
Sensitivity
33
Say yes or no
| “ medical therapies that lower mortality change the incidence of the disease “ ?
NO ! Rather, when there is less disease mortality, patients with a disease live longer and thus increase the prevalence
34
The more centralized ( not spread ) are the data gathered, the ( blank ) is the precision
Greater
35
# Fill the blank
Sensitivity and Specificity start with the ( blank)
Disease ( if you have the disease, what is the likelihood you have a positive or a negative test ?)
36
In descriptive statistics, which is better in the assessment of grouping data points (between the mean and the median )
the median because in the following example : 1-2-3-4-8-8-8-20-100 / the mean is 17 but the median is 8 . This shows that the median is a way of correcting for outliers. In this case 100 is an outlier
37
If we are studying a new medication what would be the null hypothesis ?
What we must do with that null hypothesis?
The null hypothesis is that : there is no benefit of the drug
We must REJECT that null hypothesis
38
Odds ratio is used in the study called ?
Case control studies
39
When a test is described as “gold standard” because it gives the truest answer, this relates to the term .....?
Accuracy ( is equivalent to validity ) is the combination of sensitivity and specificity
40
When the data points cluster around one point which is the opposite to scattered or spread out , of this is known as ....?
Precision
41
If the mean is “ 240 “ and the SD is 18 points , you most score ( blank ) to be better than 86% of your peers
258 ( 240 + 18 ) . That is the mean of 240 plus 18 points of SD . That is 1 SD above the mean
42
The 2 study methods used to assess
1- different groups of data between
2- different sets of data that are in
3- MORE THAN ONE GROUP are called ?
T-test ( T-score ) and ANOVA
43
The most accurate type of study in biostatistics is a prospective trial called ?
RCT ( Randomized controlled trial )
44
The greater the scatter , the less the ( blank ) .
| Fill the blank
Precision
45
If you DON’T have the DISEASE, what is the likelihood you will have a negative test ?
This question responds to the concept of ?
Specificity
46
Accuracy is equivalent to ....?
Validity
47
If the P-value is 0.05 there is a 95 percent chance that the ( blank blank ) is true
Alternate hypothesis
P- value of 0.05 means that if the study were repeated there is a 95 percent chance it would reproduce findings consistent with the current findings
48
What test would you use to analyze if a vaccination status is related to contracting a disease ( like polio ) ?
Chi- square ( answers the question : Are this groups related or not ? ) ........ this means that there is a relation between polio vaccine and polio disease
49
Mention the type of error
| “ Saying there is a statistically significant difference in the data when there really isn’t “
Type l ( alpha error ) ( false positive like you know who )
50
The data set that occurs in a particular order is called ?
Ordinal data ( occurs in a numerical list ) . There are no clear breakpoints or “ intervals “ . Is a secuencias order ( example : temperature has a clear breakpoint at the freezing and boiling points ) that’s why temperature is an example of “interval” data not ordinal
51
Z-score is a way of showing ?
How far above or below your score is compared with the mean
This means that if you are 1 SD above the mean your Z score is 1.0
Example: a score of 238 on step 3 with a mean of 222 and SD of 16 gives you a Z score of +1.0
52
The 2 studies that can analyze more irregular data , that means data that is not distributed in a bell- shaped or normal distribution are ????
T-test and ANOVA
Note : irregular data means for example analyzing or screening the water samples in a random municipality which is “prone” or susceptible to gather data that is not precise
53
Is you are been studied and you know you are being watched, so you change your behavior . This is called ??
Hawthorne effect
54
The formula (a x d divided by c x b ) responds to ?
Odds ratio
55
The likelihood that all people without the disease are correctly identified as disease negative responds to ??
Specificity
56
In 4 tests done , a patient had the following percentage of improvement:
1- cero percent 2- cero percent 3- cero percent 4- 100 percent
A/ Mention what is the final percentage of benefit from the test
B/ Mention what is the confidence interval from the test
A/ 25 percent
| B/ CI ( 0.0 to 1.0 )
57
Mention the type of error
| “Concluding that the drug is ineffective when it actually helps “
Type ll error ( beta error ) ( false negative) Lazy investigator because agrees with the null hypothesis because he is too lazy to work
58
If you DO have a positive test, what is the likelihood you really DO have the disease ?
This question responds to the concept of ??
Positive predictive value
59
What test answer the question
| “ Are these groups related or not ?”
Chi-square
60
Blood groups ( A-B-AB ) which are name only, belong to the type of data called ....?
Nominal data ( name only ) Ex : hepatitis types A , B, C
61
The te following statement refers to the concept of ?
“Those with no disease will have a negative test “
Specificity
62
SEM ( standard error of the mean ) . When calculating it you reach the conclusion that the more samples are added to the data set, the grouping becomes A ( narrower ) or B ( wider ) .....? That means the grouping becomes A ( more precise ) or B ( less precise )
Answer between A and B
Both answers are A because it becomes narrower and more precise if you add more samples to the data set . “ The smaller the SEM, the more precise the data is “
63
A negative result of a sensitive test will ( blank ) the disease
Exclude
64
TP / TP + FP
Refers to ??
PPV ( positive predictive value)
65
Study that is used to assess the risk of a disease by calculating the “ relative risk “ is called ?
Cohort study
66
What type of data is represented ?
| 1- HIV positive or HIV negative
Nominal data ( only name )
67
The correlation coefficient or “r” between Diabetes Mellitus and the development of 1- stroke 2- blindness 3- myocardial infarction is +1 what means they have a ( blank ) correlation
Strong
68
If the CI is narrower, this means the test is ( blank blank )
More precise
| Ex: a CI ( 5 to 7 ) is narrower than a CI ( 4 to 8 ) . Thus is more precise
69
If the test is perfectly specific, the “ false positive rate “ will be
A/ high
B/ none
Choose between A or B
B/ none
70
To calculate the NNT ( number needed to treat ) the denominator in the formula is ??
ARR ( absolute risk reduction )
1 X 1 divided by ARR
See page 16
71
The following statement is an example of ?
| “ the new drug works !!!!”
Alternate hypothesis
72
Positively skewed is skewed to the ?
1- right
2- left
Select the correct between 1 and 2
1- to the right
73
Choose between A and B
When an outcome has a CI that crosses 1, the results are:
A/ significant
B/ no significant
B/ no significant
74
# Fill the blank
Negative predictive value and Positive predictive value start with the ( blank )
Test ( if you have a negative/ positive “test” what is the likelihood you really don’t/do have the disease ? )
75
# Fill the blank
The greater the prevalence of a disease, the greater the ( blank )
PPV ( positive predictive value)
76
If you wanna cut CI in half, that means make the Confidence Interval narrower, you must ......?
Increase the sample four times
77
Mention the type of error
| “ Accepting the null hypothesis when you should reject it “
Type ll error ( beta error ) false negative
| Lazy investigator that agrees with the null hypothesis because he is too lazy to work
78
Answer the following questions by filling the blank using A or B:
A- sensitivity / specificity
B- negative predictive value / positive predictive value
(Blank ) start with the disease
( Blank ) start with the test
A ( start with the disease )
| B ( start with the test )
79
To assess 2 groups of data we must use ( blank )
T-test ( note that ANOVA is to assess 3 or more groups )
80
When you have a clear breakpoint in a set of data, this is called ....?
Interval data Ex : CD4 count is monitored and stop MAI prophylaxis at 50 CD4 cells and we stop PCP prophylaxis when it rises above 200 CD4 cells
81
If you DO have the DISEASE, what is the likelihood you will have a positive test ??
This question responds to the concept of ??
Sensitivity
82
The average of all data points is called
the mean
83
Mention the type of error
| “ accepting the alternate hypothesis when it isn’t really true“
Type l error ( alpha error ) ( false positive like you know who )
84
When a test can be reproduced . Every time the test is done is going to be the same result .... this relates to ....?
Reliability ( a test can be reliable and no necessarily accurate ) ...... you can have a wrong result over and over again
85
To observe “prospectively “ over time what happens to groups of patients with certain exposures or underlying diseases we must use the study called ?
Cohort ( are observational and prospective studies )
86
40 000 students take USMLE step 3 each year. The mean score is 222 with a SD of 16 . How many students scored above 254 ?
1000 ( because 254 is 2 SD above the mean what means that 2.5 percent where above 254 so 2.5 percent of 40 000 is 1000 students
87
If the mean is 222 and the SD is 16 . A score of 254 is how many SD above the mean ??
2 SD ( 222+ 16+ 16 = 254) . Remember that being 2 SD above the mean puts you in the top 2.5 percent of all test takers because 95 percent of all scores are within 2 SD and if you can see : from 95 to 100 there are five points divided into : 2.5 + 2.5 because the pyramid is symmetrical so this means that there is 2.5 points at each side of the pyramid
88
The following is good or bad ?
| Narrower CI + small SEM + huge sample
Very good !!!!! Because = increased validity or accuracy of the test
89
Mark A or B or C ( or as many as are true )
“The new drug works with a P-value of < 0.05 “
A/ the null hypothesis is rejected
B/ the null hypothesis is accepted
C/ the alternate hypothesis is accepted
A and C because if the null is rejected is the same that the “ alternate “ is accepted
Remember that P-value < 0.05 is GOOD !!!!
90
Mention the type of error
| “ saying that the new drug works when it doesn’t “
Type l ( alpha error ) ( false positive like you know who )
91
TN / TN + FN
Refers to ??
Negative predictive value
92
If the test is perfectly sensitive then the “ false negative rate “ will be:
A- none
B- high
A- none ( the most sensitive the test is the lower the “ false negative rate “ will be )
93
In positively skewed distribution which is greater : the mean or the median
the mean ( in this case the mean is greater than the median which is always greater than the mode )
94
The following formula refers to ? (a/c divided by b/d )
Odds ratio
95
You can assess “recall bias “ when assessing the study called ????
Case control
96
To assess 3 or more groups of data we must use ( blank )
ANOVA
97
The type of bias where hospitalized patients are used as trial subjects instead of the general population is called ??
Berkson Bias ( is solved by random selection of trial subjects )
98
TP / TP + FN refers to .....?
Sensitivity
99
To improve the precision of a test you must increase the ( blank )
Sample size
100
Early detection of a disease is confused with increased survival based on treatment
This refers to ????
Lead time Bias
101
Fill the blank
| With a very specific test, having a positive result will rule the disease ( blank )
In ( rules in, which is opposed to rule out )
102
TN/ TN + FP refers to ......?
Specificity
103
68 percent of all scores are within how many standard deviations ?
1 SD of the mean
104
The following statement is an example of ??
| “ the drug is not better at effecting a cure than random chance “
Null hypothesis
105
To calculate the NNH ( number needed to harm ) the denominator of the formula is ......?
AR ( attributable risk )
| Example: page16 ( 1 divided by AR )
106
The percentage “decrease “ of risk of death or disease from a treatment compared with 100 percent of the people in a population refers to the concept of ??
ARR ( Absolute risk reduction )
Example : without treatment for anterior wall myocardial infarction the percent dying is 40 ( out of 100) . Now with treatment ( angioplasty) that percent is 20 , so the ARR is 20 %
107
With a very specific test, having a positive result means ......?
You have the disease
108
Hawthorne effect is solved by ??
Using a placebo control and blinding both the investigator and the subjects
109
A drug to prevent stroke from atrial fibrillation has a mean benefit of 30 percent of RRR ( relative risk reduction) with a value of 0.7. If the study has a CI: ( 0.5 to 1.5 ) . What is the :
A/ validity
B/ Accuracy
Answer A and B
A and B are both low because accuracy and validity are the same , so with that CI that crosses 1, those two are LOW . Precision is also low
NOTE : When CI crosses 1 means that this results are not precise enough to be useful
110
If the risk of dying from Myocardial infarction in treated patients is 20 percent opposite to untreated patients ( 40 percent). What is the RRR( relative risk reduction) ?
```
50 percent ( always seems to be a much larger number, that is why it can be used to exaggerate the effectiveness of medications : see the example below )
Ex: in patients without heart disease but high LDL levels there is a mortality rate of 3% ( from heart disease in those otherwise healthy people ) OVER 5 years . Now with the use of statins for 5 years that risk is decreased to 2% ( that difference of 1 % is the ARR of 1 % ) which means you must treat 100 people for 5 years to save 1 life. The RRR is 33% in that scenario (from 3% to 2% ) which you can use to say that there is a Risk Reduction of 33% by using statins ( referring to RRR, not ARR of 1% ) . Meanwhile the risk of serious liver toxicity with statins is at least 3% so the NNH someone from statins is 33. The number you need to treat to help one person is 100 ( for every person you help with statins in a generally healthy hyperlipidemic population, you harm 3 people.
```
111
A score of 238 on step 3 with a mean of 222 and SD of 16 gives you a Z score of .......?
+ 1.0 ( one SD above the mean )
