1.
When conducting research, Type 1 errors (a-error)
occur for all of the following reasons except:
A.
Small sample size
B.
Too many analyses
C.
Analyzing data too often and stopping the study when a significant
difference is found
D.
Too few subgroups
E.
Too many variables
Click
here for answer
2.
A study was conducted to evaluate the efficacy of a new medication in
lowering the pulmonary hypertension that is seen post-operatively after repair
of congenital heart defects compared to the “traditional approach”.
The a-error
was set at 0.05, b-error
at 0.2, and the authors were looking at a 25% improvement.
At the conclusion of the study, no difference was found between the new
drug and the traditional therapy. True
statement(s) regarding this study include:
A.
There is a 50% chance that the null hypothesis was accepted, when there
was a difference
B.
The power of the study is 20%
C.
The P-value was set at 0.5 in this study, and there is an 80% chance that
an improvement in response to the new drug was missed
D.
There is an 80% chance that the authors did not miss a 25% difference
between the two treatments
Click here for answer
3.
The number of groups is not a factor in the selection of a statistical
test in which of the following type of data:
A.
Nominal data
B.
Ordinal data
C.
Continuous data
D.
None of the above
E.
All of the above
Click here for answer
4.
True statements regarding correlation include all of the following
except:
A.
How strong the correlation is, depends on the slope of the line
B.
With a larger sample size, one is more likely to get statistical
significance with the same correlation co-efficient (r)
C.
For ordinal (ranked) data, Spearman’s rho or Kendall’s tau are used
instead of Pearson r for continuous
data
D.
When one is interested in correlations using nominal data, such
correlations are obtained with odds ratio or relative risk determination
E.
Correlation does not make judgments as to whether one variable affects or
predicts another
Click here for answer
5.
A study is being conducted to evaluate the effect of the presence of an
attending physician looking over the shoulder of the fellow while he is
performing a procedure, in terms of complications related to the procedure.
At the end of the study, relative risks are calculated.
True statements include:
A.
A relative risk value of 0.35 indicates that the presence of an attending
is associated with lower complications
B.
A relative risk value of 1 means
that it doesn’t make any difference
C.
A relative risk value of 5 indicates that the presence of the attending
increased the risk of complication five times
D.
None of the above
E.
A, B and C
Click here for answer
6.
A study is being conducted to evaluate the following:
Does dressing change by the nurse in the PICU increase the risk of wound
infection? 1,000 children are
enrolled. Of the dressings changed
by the nurse, 75 did develop wound infection while 25 did not.
Of the dressings changed by the primary surgical attending, 125 developed
wound infection while 775 did not. The
relative risk value of the nurse changing the surgical dressing in this case is
closest to:
A.
1
B.
5
C.
10
D.
.035
E.
7
Click here for answer
7.
True statements pertaining to standard error of the mean include all of
the following except:
A.
Standard error of the mean measures the dispersion of a number of sample
means around the true population mean
B.
Standard error of the mean represents the precision with which a sample
mean estimates the true population mean
C.
Standard error of the mean is usually larger than standard deviation
D.
There is a 95% chance that the true mean of the population from which the
samples were obtained lies within ±
1.96 standard error of the mean of the sample mean
E.
When you say “with 95% confidence, 86% ±
5% of the people find this journal useful,” the ±
5% represents ±
1.96 standard error of the mean
Click here for answer
A.
Standard deviation
B.
Standard error of the mean
C.
Mean
D.
a-error
E.
b-error
8.
____ Quantifies
the variability (dispension) of the individual values from the mean within a
sample
____ Provides the
confidence interval of the population mean
____ Should not be
used as a measure of central tendency for ranked data
____ Wide
variability of the data is associated with this
____ When
conducting a research project, the more variables we look at, the greater the
chance of producing the item listed above
Click here for answer
9.
Lidocaine is being infused at 20 mg/kg/min into a patient who has a
volume of distribution for lidocaine of 1.1 L/kg.
If the half-life of lidocaine is 100 minutes at a steady state, the
lidocaine concentration would be:
A.
1.6 mg/ml
B.
2.6 mg/ml
C.
0.6 mg/ml
D.
8.6 mg/ml
E.
None of the above
Click here for answer
10.
In the above question, without a bolus, the steady state will be reached
in:
A.
6.7 hours
B.
5.7 hours
C.
8.7 hours
D.
3.7 hours
E.
None of the above
Click here for answer
11.
It is necessary to compare data from two groups by non-parametric tests
when:
A.
There are fewer than 10 data points in each group
B.
The two groups are not of equal size
C.
The data are expressed as ranks
D.
The data are expressed as ratios
E.
The two groups have different standard error of the mean
Click here for answer
Answers
1.
D
Type I or a-error
is more likely to occur when you have too many variables rather than too few
subgroups. (Fuhrman BP, et al.
Pediatric Critical Care, 2nd Edition; pp 172-174)
2.
D
The power of a study is equal to 1 – b-error
which was set in this study at 0.2, and therefore the power of this study would
0.8 or 80%. The P-value of this study was set at 0.05, and there is only
a 20% chance that the authors actually missed an improvement due to the new
drug. There is an 80% chance that
an improvement due to the new drug was not missed. (Fuhrman BP, et al. Pediatric
Critical Care, 2nd Edition; pp 172-174)
3.
A
Groups are not a factor for nominal data.
(Fuhrman BP, et al. Pediatric
Critical Care, 2nd Edition; pp 169-170)
4.
A
In evaluating how strong the correlation
is, it is more dependent on how tightly all the points are scattered around the
slope line. Correlation does not
make judgments as to how one variable affects or predicts another.
A regression co-efficient permits prediction.
(Hermansen M. Biostatistics: Some Basic Concepts.
Caducens Medical Publishers, Inc.; p 56)
5.
E
Relative Risk (RR) refers to the risk of
having complications in the presence of a risk factor compared to the absence of
the factor. The range of RR is from
0 to infinity. A RR of 1-0
indicates that there is no difference in risk.
Calculation of RR is explained in the answer to Question #6 below.
(Hermansen M. Biostatistics: Some
Basic Concepts. Caducens
Medical Publishers, Inc.; pp 164-166)
6.
B
Infection
Present Absent
Total
Risk Factor
Nurse
Present A = 75 B = 25 100
Absent C = 125 D = 775 900
Total 200 800
1,000
Relative Risk
=
Risk of infection when risk factor present
Risk of infection when risk factor absent
=
A/A+B
=
75/100
C/C+D 125/900
= 5.4
(Hermansen M.
Biostatistics: Some Basic
Concepts. Caducens Medical
Publishers, Inc.; pp 164-166)
7.
C
Standard error of the mean is a descriptive statistic, and standard error
of the mean helps determine the range in which the population mean exists.
Standard error of the mean equals standard deviation divided by the
square root of the number of variables, and therefore, standard error of the
mean is always smaller than standard deviation.
SEM =
S.D.
Ö
n
(Hermansen M. Biostatistics:
Some Basic Concepts. Caducens
Medical Publishers, Inc.; pp 38-41)
8.
A, B, C, E, D
Self-explanatory. (Hermansen M.
Biostatistics: Some Basic Concepts.
Caducens Medical Publishers, Inc.; pp 38-41)
9-10.
B, A Clearance (CL)
= Dose at Steady State (DSS)
Concentration at Steady State (CSS)
= Kel ´
Volume of Distribution (Vd)
Kel
= 0.693
¸ Half life
CSS
= 20 ´
100
0.693 ´
1.1
= 2.6 mg/ml
It takes four times the half-life to reach a steady state concentration.
=
4 ´
100
=
6.7 hours
(Kearns GL. Clinical
pharmakokinetics in infants and children. Clin
Pharmacol, 1988; Vol. 7:198)
11.
C
Parametric methods of statistical
analysis use distribution assumptions (i.e., normal distribution) of data, and
the distribution is described by mean and standard deviation.
Non-parametric methods are also called distribution-free.
These methods are based on analysis of ranks rather than actual data, and
therefore, they are sometimes called rank methods.
Skewed data are commonly analyzed by non-parametric methods. Methods using ranks are especially suitable for data which
are scores rather than measurements. Examples
include: Apgar scores and stages of
disease. (Altman DG.
Practical Statistics For Medical Research.
Chapman & Hall 1991, pp 171-173)