Ritchie Lectures 3, 4, 5 Flashcards
Statistical significance implies that a result is biologically significant.
False
When testing significance…
a. Always expect the alternative hypothesis to be true
b. Only use Z statistics for published data
c. You are measuring differences between your data and what is expected under the null hypothesis
d. Ensure that the sample size does not exceed N>20 because the SEM will be too large
c. You are measuring differences between your data and what is expected under the null hypothesis
In terms of experimental hypotheses…
a. The null hypothesis only applies to data that forms a Poisson distribution
b. The alternative hypothesis states that observed differences reflect chance variation
c. Both the null and alternative hypotheses must be defined to test significance
d. The null hypothesis states that observed differences are real
c. Both the null and alternative hypotheses must be defined to test significance
What would be considered a large sample size?
a. N ≥ 3
b. N ≥ 20
c. N ≥ 10
d. N ≥ 5
b. N ≥ 20
The study design can be readily checked by running a test of significance.
False
A student’s T test requires small degrees of freedom.
True
The P value does not depend on sample size
False
The central limit theorem only applies to random samples
True
When testing significance, random sampling is assumed
True
What is the Mean and Standard Deviation of the following data set: 5, 4, 8, 3.
a. Mean= 5, SD= 2
b. Mean= 20, SD= 3.46
c. Mean= 4.5, SD= 2
d. Mean= 5, SD= 1.87
d. Mean= 5, SD= 1.87
Mean = (5+4+8+3) / 4 = 5
SD = √[((5-5)2 + (4-5)2 + (8-5)2 + (3-5)2) / 4]
= √[(0 + 1 + 9 + 4) / 4]
= √[14 / 4]
= √[3.5]
= 1.87
How is Standard Error calculated?
a. √SD
b. √SD – 2 X √N
c. SD / √N
d. √(SD + N)
c. SD / √N
What is the Standard Error of the following data set: 5, 4, 8, 3.
a. 0.94
b. 5
c. 0.05
d. 2
a. 0.94
Mean = 5
SD = 1.87
N = 4 (4 samples)
SEM = SD / √N
= 1.87/√4
= 0.94
In practice, most sampling is WITH replacement.
False
What is INCORRECT about the central limit theorem?
a. The sample from a population must follow a binomial distribution for the probability histogram to follow a normal curve
b. N must be relatively small
c. The probability histogram must be put into standard units
d. The area of a probability histogram = “chances”
b. N must be relatively small
In regards to confidence intervals:
a. Intervals are random and a result of sampling
b. Intervals always contain the unknown population mean
c. Intervals are independent of sampling
d. Intervals are only relevant in epidemiological research where N = 1
a. Intervals are random and a result of sampling
What does a test statistic measure?
a. The Standard Error of a small population sample
b. The likelihood that the null hypothesis and alternative hypothesis are both wrong
c. The difference between the data and what is expected of the null hypothesis
d. The likelihood that the alternative hypothesis does not correlate with z
c. The difference between the data and what is expected of the null hypothesis
What is the test statistic (z) given the following information: Mean = 4, Result = 5.2, SE = 0.4
a. 1.0
b. 3.0
c. 4.0
d. 5.2
b. 3.0
Test statistic = (observed quantity – expected value) / SE
When calculating the P value, the basis is that the null hypothesis is right.
True
What is the correct way to calculate the SE of a difference of means from two independent samples? Answer using the following data set as an example: Sample 1 SE, 0.59. Sample 2 SE, 1.02.
a. SE (difference or sum) = [(1.02) + (0.59)]^2 = 2.59
b. SE (difference or sum) = √(1.02)^2 + √ (0.59)^2 ] = 1.61
c. SE (difference or sum) = √[ (1.02)^2 + (0.59)^2 ] = 1.18
d. SE (difference or sum) = [(1.02) - (0.59)]^2 = 0.18
c. SE (difference or sum) = √[ (1.02)^2 + (0.59)^2 ] = 1.18
