Answer key

Recitation #1 Answers
Using and describing data
2. The dataset has 602,833 observations. That means that 602,833 subjects
were surveyed. That is roughly 0.05% of the Indian population.
(602,833/1237000000 x 100)
3. - The command ‘describe’ shows the storage type, display format, and
label of each variable. The variable label allows us to know what the
variable is measures. The storage type tells us whether the variable
contains numbers that are integers, strings (letters), or numbers with
decimals (float). For example, variable year’s type is integer (int) since
years cannot be decimal values (this dataset only has 2004 data). On the
other hand, variable wtper’s type is float (number with decimals) since it
is a variable that measures the respondent’s weight.
- The command ‘codebook’ shows the variable type, label, and range
among other important information (whether there are any missing
values and how many unique values are there). By using ‘codebook’ we
can see which variables are continuous, discrete, categorical, and
- 602,833 individuals are part of this sample.
- India’s population is 1.237 billion. Therefore, roughly 0.05% of the
Indian population was surveyed. (602,833/1237000000 x 100)
- By typing ‘codebook’, we can see that variable serial (Household serial
number) has 124680 unique values. That means that 124680 different
households have been surveyed.
- Continuous variables: wtper
- Discrete: cntry, year, sample, serial, urban, regionw, geolev1, geo1a_in,
geo1b_in, geo2b_in, pernum, age, age2, sex
- Quantitative variables: year, sample, serial, pernum, wtper, age
- Categorical variables: cntry, urban, regionw, geolev1, geo1a_in, geo1b_in,
geo2b_in, age2, sex
Note: For those of you confused about the variable pernum, this variable
is not needed for our exercise today. However, you may be interested in
knowing that pernum numbers all persons within each household
consecutively (starting with "1" for the first person record of each
household). When combined with serial, pernum uniquely identifies each
person in our sample.
294205 π‘€π‘œπ‘šπ‘’π‘›
308627 π‘šπ‘’π‘›
= 0.953 π‘€π‘œπ‘šπ‘’π‘› π‘π‘’π‘Ÿ π‘šπ‘Žπ‘› 𝑖𝑛 πΌπ‘›π‘‘π‘–π‘Ž
6. Our finding is consistent with Amartya Sen’s assertion that “in South Asia,
West Asia, and China, the ratio of women to men can be as low as 0.94”.
We can observe how our 0.953 ratio of women per man is definitely lower
than the ratio of women to men in Europe as reported by Amartya Sen. In
the case of Europe, there are slightly more women than men. Our data
suggests in India it is the other way around.
Punjab’s women per man ratio:
10467 π‘€π‘œπ‘šπ‘’π‘›
11522 π‘šπ‘’π‘›
= 0.908
Haryana’s women per man ratio:
Kerala’s women per man ratio:
6677 π‘€π‘œπ‘šπ‘’π‘›
7507 π‘šπ‘’π‘›
12086 π‘€π‘œπ‘šπ‘’π‘›
10942 π‘šπ‘’π‘›
8. Yes, we can use a similar approach as in question 7.
We type: tabulate age sex
At age less than one year, women to men ratio is 0.921.
At age 1, the ratio is 0.905.
At age 2, the ratio is 0.952.
At age 3, the ratio is 0.977.
At age 4, the ratio is 0.892.
At age 10, the ratio is 0.840.
At age 21, the ratio is 0.934.
At age 31, the ratio is 1.159.
At age 41, the ratio is 0.912.
Professor Sen’s statement says that at birth there are around 106 male
children for every 100 female children everywhere in the world - a female
to men ratio of 0.943 – but that from then on biology seems to side with
women. Even though our results do not show a clearly increasing pattern
in the female to male ratio, and thus seem to contradict Amartya Sen’s
statement, we must note he assumes that women and men receive
similar nutritional and medical attention. Amartya Sen suggests that if
women do not engage in ‘gainful’ employment, then women and men are
less likely to receive similar nutritional and medical attention. In the
article, he also explains Southern Asia ranks one of the lowest in both
‘gainful’ employment for women and female life expectancy. Thus, we
cannot use our results to seek evidence for the increasing biological
advantage of women because there is evidence women and men do not
face the same conditions in India.
Rural female to male ratio: 0.958
Urban female to male ratio: 0.945
There is evidence that the ratio is lower in urban areas. We could
interpret this difference as due to the fact that in urban areas individuals
may have easier access to selective abortion.
This goes in line with Amartya Sen’s assertion that policies or differences
that initially may seem neutral and unrelated to female-male ratios end
up producing a negative effect on female life expectancy due to already
existing societal norms. In this case, the difference in access to abortion
may seem in itself neutral. Nevertheless, the different prospects of men
versus women to find employment may produce a preference for male
babies, which in turn could lead to seeking selective abortion if available.