Created
September 4, 2018 19:33
-
-
Save chrishanretty/5977fcbe18f1180c51e8b9ccc0806cea to your computer and use it in GitHub Desktop.
Calculate the 95% confidence interval for S v SD
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
## Let's suppose that polls work as they are supposed to | |
## Let's take the average of the last six polls | |
## Let's call p_1 the poll average proportion voting for the Social Democrats | |
## and p_2 the poll average proportion voting for the Sweden Democrats | |
p_1 <- 0.246 | |
p_2 <- 0.189 | |
## In this case, the standard error of the difference between the two | |
## proportions is equal to the square root of the sum of the squares | |
## of the standard error attaching to each proportion. | |
## The standard error for each proportion is equal to the sqare root | |
## of the proportion times one minus the proportion, divided by the | |
## sample size. Let's say that we have an "as-if" sample size of | |
## 1,000. Let's call the standard errors se_1 and se_2. | |
n <- 1000 | |
se_1 <- sqrt((p_1 * (1 - p_1)) / n) | |
se_2 <- sqrt((p_2 * (1 - p_2)) / n) | |
## We can the calculate the combined se. | |
comb_se <- sqrt(se_1 ^ 2 + se_2 ^ 2) | |
## This is slightly higher than the individual standard errors | |
comb_se > se_1 | |
comb_se > se_2 | |
## Now let's work out the 95% confidence interval for the difference. This is approximately two standard deviations in either direction | |
lower <- (p_1 - p_2) - comb_se * 2 | |
upper <- (p_1 - p_2) + comb_se * 2 | |
## In fact, we can work out the probability of a result so extreme as | |
## to put the Sweden Democrats ahead by taking the difference in | |
## proportions, converting it to a z-value, and working out the | |
## probability of z-values that extreme or more, according to a normal | |
## distribution table. | |
zscore <- (p_1 - p_2) / comb_se | |
## pnorm = cumulative density function | |
## 1 - pnorm = prob of results more extreme than | |
1 - pnorm(zscore) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment