Week 3: Hypothesis Testing and Significance

Question 1

What command generates a new variable?

Answer 1

<gen>
</gen>

Question 2

How do you recode continuous variables into categorical variables?

Answer 2

Use the <recode> or <replace> command in combination with <if> statements to specify ranges and assign categories.
Example:
recode varname (a/b=1) (c/d=2) (e/f=3), gen(newvarname)</if></replace></recode>

Question 3

Which command labels a variable?

Answer 3

<label variable varname “name”>

Question 4

How do you ensure the correct categories after recoding a variable?

Answer 4

Use frequency tables and crosstabs, such as <tab></tab>

Question 5

What is the syntax to assign value labels to a variable?

Answer 5

<label define labelname value1”label1” value2”label2”, replace>
<label></label>

Question 6

What does the <egen> command do?</egen>

Answer 6

Creates derived variables, such as sums or counts, across specified variables.

Question 7

What is the purpose of <foreach> and <forvalues> loops?</forvalues></foreach>

Answer 7

They automate repetitive tasks, such as recoding multiple variables

<foreach> is a more general loop that allows string, numeric, and variables as list (without a pattern).
<forvalues> is a more specific loop. Only numeric is allowed as lists, and lists should have a clear pattern.
</forvalues></foreach>

Question 8

How are missing values assigned in a new variable during coding?

Answer 8

Use <replace varname =. if condition>

Question 9

How do you combine two categorical variables into one?

Answer 9

Use a combination of logical operators (I, &) and <recode> or <gen> to conditionally create a new variable.
Example:
<gen newvar=.>
<replace newvar=0 if var1==a | var2==b></gen></recode>

<replace> b & var2==c>
etc...
<label variable newvar "Variable name">
<label define newvar1 0"abcd" 1"efgh">
<label>
</label></replace>

Question 10

What is the syntax to count specific values across multiple variables?

Answer 10

<egen newvar = anycount(varlist), values(value)>
Example: To count number of those with asthma
<egen asthma = anycount (hedib01-hedib06), values(2)>
Here, we know that (2) denotes the presence of asthma in the initial variable

Question 11

How do you compute the sum of multiple variables into a single variable?

Answer 11

Use <egen newvar = rowtotal(varlist)>
Example:
<egen tot_heart_sum = rowtotal(heart_*)>

Question 12

What command is used to produce a frequency table?

Answer 12

<tab>
</tab>

Question 13

What does reverse coding ensure in survey data?

Answer 13

It aligns differently phrased questions for consistent interpretation in scales or indices.

Question 14

How do you check for errors after creating a new variable?

Answer 14

Cross-tabulate the new variable with its original counterpart and examine discrepancies.

Question 15

What does the <rowtotal> function in <egen> calculate?</egen></rowtotal>

Answer 15

The sum of specified variables across a row for each observation.

Question 16

What is a “derived variable”?

Answer 16

A variable created from existing data, often through transformations or aggregations.

Question 17

How can you handle negative or missing values during recoding?

Answer 17

Use conditional statements like <replace varname =. if varname <0>

Question 18

What does the command <label> do?</label>

Answer 18

It creates a set of value labels that can be assigned to variables

Question 19

Why is it good practice to verify recoding with cross tabs?

Answer 19

Ensure all values have been correctly categorised

Question 20

What is the first step to take in deriving a new variable?

Answer 20

Create a new ‘empty’ variable using the command <gen varname =.>

Question 21

How do you delete a variable if you made a mistake in recoding?

Answer 21

<drop>
</drop>

Question 22

How would you calculate the mean of a continuous variable sorted by a categorical variable?

Answer 22

<mean contvar, over(catvar)>
Note: You can also use logical operators e.g., mean <varname if cond1==2 | cond2==3>

Question 23

What character do you use to capture all variables that start with the same letters?

Answer 23

Example: If you want to capture all variables starting with hedia <hedia*>

Question 24

How can you create a single variable showing the number of heart conditions a respondent has in the ELSA dataset? (hedia0-7)

Answer 24

To count the number of conditions a respondent has, we need to recode the seven hedia variables into seven new variable, each denoting the presence or absence of a heart condition of any type. The scores will be 1 for any heart condition and 0 for no heart condition.
Command:
<recode hedia01 (96=0) (1/8=1), gen(heart01)>
<recode hedia01 (96=0) (1/8=1), gen(heart02)>
etc…

Question 25

How can you create a single variable showing the number of heart conditions a respondent has in the ELSA dataset? (hedia0-7) - using loops

Answer 25

<forvalues j=1/7{
recode hedia0j' (96=0) (1/8-1) (95=1) (min/-1=.), gen heart_j’)