代写Math/Stat 511、代做Python，c/c++编程语言、代写Java程序设计、代写RCBD

日期：2019-04-25 10:46

Final Exam Takehome Exam (Due by 3:00 p.m. May 3 Friday) - Math/Stat 511 Spring 2019

DO NOT WORK IN GROUPS FOR THIS EXAM OR DISCUSS THIS EXAM IN ANY

MANNER WITH CLASSMATES

1. A lawn care company was interested in reducing the amount of time that their employees spent on

lawn care activities for individual customers. The company decided to conduct an A/B test to assess

whether trimming the grass near the fence edges had any influence on customer attrition. The company

decided to randomly divide customers into a control group (lawn care with grass trimmed near fence

edges) and a reduced time group (lawn care with grass not trimmed near fence edges). Out of their

238 clients, 119 were assigned to the control group and 119 were assigned to the reduced time group.

One year later, it was assessed whether customers were still using the company’s services. The results

of the study is shown in a table below.

Use one of the methods that we have discussed in class to help provide a recommendation as to whether

you feel: 1) the company should collect more data, 2) the company could reasonably reduce the time

on each client without expecting a substantial increase in attrition rates, 3) the company should stick

with the regular service. To get full credit you must discuss practical significance and whether or

not the study was informative. Use your own judgement concerning what would constitute practical

significance in this case.

Make sure to include any code you use in your answer.

still using stopped_using

regular 87 32

time_reduced 63 56

2. A researcher is interested in comparing soil quality after growing winter wheat, quinoa, chickpeas and

barley. The researcher grows these crops on a single farm, using a randomized complete block design

(RCBD) with 6 blocks and 4 treatments per block. The data can be found in the file quinoa.csv in

the homework data folder. The primary questions of interest are: a) is there any evidence of differences

in soil nitrogen after growing each of the four crops and b) if so, which crops are different from one

another in terms of nitrogen content. Read in the data file quinoa.csv in the homework data folder on

blackboard to prepare for the analysis. Whatever method you choose for analysis, make sure that the

method uses all the data for estimating the error variance.

Include in your answer:

a. A discussion of statistical evidence to answer the researcher’s question concerning any evidence

of difference in soil nitrogen

b. A discussion of statistical evidence to answer the researcher’s question concerning which crops

show evidence of a difference in nitrogen (if relevant)

c. A plot of the average nitrogen levels for each crop with 95% confidence intervals.

d. A discussion of which model assumptions appear plausible for this data. Include plots and describe

what you are assessing with each plot included.

Make sure to include any code you use in your answer.

3. Researchers were interested in determining whether horse survival (yes/no) was influenced by colloid

administration. A retrospective review of medical records of horses with enterocolitis (digestive tract

inflammation) treated with two colloid types, natural plasma or synthetic hetastarch, was conducted.

Data collected included whether or not the horse survived until discharge, along with two important

risk factors PCV (packed cell volume) and TS (total solids). A total of 92 horses were included in

the review. The data can be found in the file hetastarch.csv in the homework data folder. The

variable Outcome indicates whether or not the horse survived until discharge (1=no, 0=yes). With

1

major consideration to the response variable type, use an appropriate model for this data to gather

evidence concerning whether horses that recieve different fluids show a differential in the risk of death

after accounting for important risk factors. Note that the risk factors in this dataset have missing

values. One of the simplest methods for dealing with missing data is to use imputation of the mean.

While this method is not necessarily the best way of dealing with missing data, the amount of missing

data is relatively small and it was thought that it would not have a major effect on the analysis in

this case. Thus, for this analysis, fill in missing values of risk factors with the mean of all non-missing

values.

In your answer, make sure to include the following:

a. State the scope of statistical inference (pop’n/causal) given what you know about the study

b. Include a statement that summarises what is plausibly consistent with the data when considering

the association between survival and colloid administration. This may include the degree of

evidence against the null hypothesis (optional), but should summarize the magnitude of effects

that are consistent with the data. Additionally, you should discuss whether this range covers both

effects that are meaningful and not meaningful in size (use your own judgement here).

Make sure to include any code you use in your answer.

2