Statistics in a Modern World 800
Solutions to Exam 1
-
- (i) because premature infants were fed breast milk in the hospital, giving
researchers the opportunity to possibly administer the treatment; it
would be much less likely that they could control breastfeeding in
other infants at home for more than three months
- (i)
- (i) "better on tests of mental development later in life" suggests
a quantitative assessment
- (iii)
- (ii) (bedwetters or not)
-
- "2000 students in Winston-Salen and Forsyth County high schools"
made up (comprised) the sample. Note: researchers do not "make up" a
sample, they take it. Researchers may "make up" a design for producing
data, in the sense of inventing a plan, but they do not invent a sample.
- (i) The researchers were interested in all teenagers, to compare those
who do and do not watch wrestling.
- explanatory, explanatory, categorical (two categorical groups were
mentioned in paragraph 9: those who watched six or more times and those who
didn't watch at all), response
- (i) 630=63% of 1000 (ii) 90=9% of 1000
- 10+170%(10)=10+17=27
- (i) and (ii) (See paragraphs 4 and 12) but not (iii) (See paragraph
13)
- (i)
- Best answer: Teens who are already prone to violence would tend
to be more interested in watching wrestling than non-violent teens. The
majority of students answered along these lines.
-
- (iii) In both studies, researchers took control of the explanatory
variable.
-
- response (see 5th paragraph)
- response (see last paragraph)
- explanatory (see 2nd to last paragraph)
- explanatory (see paragraphs 3-5)
- (i) In principle, the researchers who weighed the filter papers
could have been blind to what stage of the stinging process applied.
But this may have been difficult to orchestrate, so I also accepted
(iv). The researchers in the second experiment could NOT have been
blind to whether a stinger had been scraped or yanked from their own
skin, nor to the length of time before removal of a stinger.
- (i) Stinging paper may be very different from stinging humans.
- (i)
- (iii) There would be no clear bias in pain reporting, but if many
researchers were to report amount of pain, it would be inconsistent;
that is, unreliable.
- (ii) Experiments are almost never conducted on a random sample
of subjects, and we have discussed the fact that this is not a problem,
as long as there is a randomized assignment to treatments.
-
- (v) Since not just one but all of these types of bias are of concern
in this situation, I took 2 points off for those who just answered (i)
and 3 points off for those who just answered (ii), (iii), or (iv).
- 4.4% is stated to be the margin of error. In general, we can be
pretty sure that population proportion is within a margin of error of
sample proportion (discussed in lecture).
- (iv) Sampling frame is the group of individuals from which the sample
is actually taken; it is unknown here, as is often the case (discussed in
lecture).
[ Home
| Calendar
| Assignments
| Handouts
]
|