In: Statistics and Probability
It is specified that a limited amount of federal funds have been allocated to assist jurisdictions whose percent of low-income working families exceeds a threshold set at 4/5 of the upper limit of a confidence interval.
Let P = population proportion of low-income working families
p = sample proportion of low-income working families
Upper limit of a confidence interval for P = p+zxSE(p)
Therefore it is specified that a limited amount of federal funds have been allocated to assist jurisdictions whose
P>4/5(p+zxSE(p))
If the confidence level is fixed, suppose it is 90% then z = 1.645
then P>4/5(p+1.645xSE(p)) and if it is compared with original P>(p+1.645xSE(p)) at 90% level wich leads to create benefit to the additional 1/5, that is 20% of the low income people and it creates additional burden allocating funds at a fixed confidence level say 90%.
On the other hand, if the confidence level is increased more than 90% and at which z>1.645 which results in increase in upper limit there by less number of people of low income will get the benefit and less allocation of funds which may become un-ethical and funds also may be miused.
To control the chances of un-ethical practice and misuse of funds and providing additional benefit to the 20% of the lower income group and not also getting additional burden for the funds, some what taking P>4/5(p+zxSE(p)) is better than taking P>(p+zxSE(p)) for confidence levels more than 90%.