Set up workspace
Point estimate of the mean: 171.1 Point estimate of the median: 170.3
Point estimate of sd: 9.4 Point estimate of IQR: 177.8 - 163.8 = 14
(180 - 171.1)/9.4## [1] 0.9468085
(155 - 171.1)/9.4## [1] -1.712766
Both are within 2 SD of the mean, so neither would be considered “unusual” by that statistical measure.
We would expect them to be “similiar” but not identical. For instance, mean would be expected to be within about 2 SD of the mean given above.
We would use the Standard Error: SD/sqrt(n) = 0.417
9.4/(sqrt(507))## [1] 0.4174687
FALSE - We are 100% confident about the sample mean is between these two numbers.
FALSE - since the number of observations is 436, even though the data is slightly skewed, we can still assume an approximately normal distribution for sample means from this population.
FALSE - CI is not about the sample mean
TRUE
TRUE that it would be narrower
FALSE based on a 1/sqrt(n)… so would need to be 9 times larger
TRUE - margin of error is 1/2 of the CI
Yes: independence - less than 10% of the total population skew - from sample, doesn’t appear there are significant outliers indicating underlying pop. skewed sample > 30
H(null): Mean of 32 is true average H(alt): Mean of 32 is not the true average
30.69 - 1.96*(4.31/sqrt(36))## [1] 29.28207
30.69 + 1.96*(4.31/sqrt(36))## [1] 32.09793
We would expect 95% of the sample CI’s to include the true population mean. This one does. So we would not reject the null at this level
2*(1 - pnorm(-(30.69 - 32)/(4.31/sqrt(36))))## [1] 0.0682026
Chance of seeing this mean greater than 5% limit we had established, so do not reject the mean.
pnorm(1.65)## [1] 0.9505285
30.69 - 1.65*(4.31/sqrt(36))## [1] 29.50475
30.69 + 1.65*(4.31/sqrt(36))## [1] 31.87525
Would reject the null hyopothesis at this lower confidence level/narrower CI.
z <- (118.2 - 100)/(6.5/sqrt(36))
1 - pnorm(z)## [1] 0
Z score huge (error in approach?)… Would reject null hypothesis of no difference from the general population.
118.2 - 1.65*(6.5/sqrt(36))## [1] 116.4125
118.2 + 1.65*(6.5/sqrt(36))## [1] 119.9875
Yes. 90% CI does not include the mean of 100 by a long shot. Expect 90% of these 90% CI’s to include the true mean of the population.
Sampling distribution is the distribution of the means of repeated samples taken from a population. Expect this distribution to be approximately normal given independent observations and not highly skewed underlying population distribution for small sample sizes. As sample size increase, standard error of the distribution decreases (bell narrows).
1 - pnorm((10500 - 9000)/1000)## [1] 0.0668072
1000/sqrt(10)## [1] 316.2278
N(9000, 316.2278)
pnorm(10500 - 9000)/(1000/sqrt(15))## [1] 0.003872983
About 0.3%
No. Has to be approximately normal. Also, sample < 30.
P-value will decrease. Because it is inversely dependent on the square root of the sample size.