There is no word count, however the assignment MUST NOT exceed 14 A$ size papers (excluding appendices) as anything beyond page 14 will NOT be marked
SWINBURNE UNIVERSITY OF TECHNOLOGY Assessment 2: Comparing Sampling Methods Page 1 of 3 SURVEY SAMPLING Assignment 2: Exercise in Comparing Sampling Methods General • Assignment 2 is a report worth 25% of the marks for this unit. • Marks will be allocated according to the quality of the written report which must be written up using the answer template provided: • Your report must be type written in Word format and should be no longer than 14 A4 sides (excluding Appendices). • Your report must be submitted via Blackboard. 1. Introduction The aim of this assignment is to make population estimates about several characteristics of road deaths in Australia using different sampling methods and to compare the estimates. 2. Aim The aim of this assignment is to make population estimates about several characteristics of road deaths in Australia from 1989 to 2010 using different sampling methods and to compare the estimates. 3. Sampling Frame Data files have been made available for this assignment as Excel files. a. Individual accident data called: Motor_Fatalities_A2_SP3_2018.xlsx b. Motor Fatality Count data called: Motor_Fatalities_counts_SP3_2018.xlsx You can assume each frame is clean, but you should quickly check them in case there are any obvious problems. The data have not been deliberately ‘doctored’ with duplicates or other incorrect data. If any suspicious values are found, say from Frequency tables, set those values as missing or if a duplicate found delete the observation. Note: Use the relevant datasets provided as your sampling frames. Although we do have a complete census in this case, assume that you only have the observations selected in your samples to do your calculations. This data was chosen to illustrate the sampling methods covered in the unit as you are dealing with actual data with a reasonable number of variables to examine and don’t have to collect Assessment 2: Comparing Sampling Methods Page 2 of 3 any data yourself. In practice when the population is available, you would work with the full population. 4. Select samples from the Motor Fatalities data This section uses the file Motor_Fatalities_A2_SP3_2018.xlsx. For each of the following sampling schemes assume you have enough funds to survey a sample of about 2000 of the population, regardless of the sampling method. Select the following samples using SAS, and in your report, explain your approach in sufficient detail for a peer to replicate how you selected each sample. The names of all files submitted should start with SurnameX (your surname followed by your first initial), and the datasets submitted should be saved in an .xlsx or .csv file format. a. Obtain a simple random sample of 2000 fatalities saved as: SurnameX_SIT90005_A2_SRS.csv Obtain two stratified samples, each of about 2000 deaths, stratified by the same strata variable, and using a variable that is either already in the frame, or one you have chosen to create from the dataset given. The variable chosen should have between 6 and 16 strata and is not one of the two existing variables Age or Speed. If you choose to create a new variable, you should provide some statistically sound reasoning why you chose to create the variable for stratification. b. The first of the two samples is to be obtained using proportional stratified sampling and saved as: SurnameX_SIT90005_A2_STRAT1.csv c. The second of the two samples should be obtained using stratified sample of your choice and saved as: SurnameX_SIT90005_A2_STRAT2.csv. d. Obtain a Probability Proportional to Size (PPS) cluster sample of the deaths using a suitable number of clusters from the set of clusters of your choosing. Choose or create a variable to use as the cluster variable. It should have a minimum of 30 clusters. For example, you may create a variable called Time_Gender which is the product of time and gender and has 48 groups. Then choose an appropriate number of cases from each cluster. Then if you wanted 2000 in you sample you could use 20 clusters with 100 cases from each cluster. Determine the clusters you want to include as in Topic 12 notes section 12.9 and save as: SurnameX_SIT90005_A2_CLUSTER.csv. e. Obtain one systematic sample of 8 years from the Fatality Count file save as: SurnameX_SIT90005_A2_Count.csv TJ Highlight TJ Highlight TJ Highlight TJ Highlight TJ Highlight Assessment 2: Comparing Sampling Methods Page 3 of 3 a. Using each of your four Motor Fatality samples generated in 4a, 4b, 4c, and 4d, determine estimates and their corresponding 95% confidence intervals, using the Finite Population Correction (FPC) throughout of: 1) the average age of the victim in your population; and 2) the proportion of deaths where the speed limit was 80 km/h or above. Apart from the two-stage cluster sample, obtain estimates using SAS, and verify these in Excel in all samples. b. Compare the standard errors of the stratified and cluster samples with that obtained using the simple random sample obtained in 2c. c. Display your results in a table which makes it easy to compare the results. d. Treating the Motor Fatality Count data as a sample, estimate for the period 1996 to 2015: 1) the total number of fatalities in this period; and 2) the ratio of fatalities to crashes. e. Your own question Using one of the samples that you obtained in 4, create and answer a question of your choosing which involves a comparison. Briefly report your results, providing some reasoning for the question and the sample selected to answer it. 6. Discussion Discuss any problems with the study and possible improvements that may be made. 7. Overview/Executive Summary. People who read only the executive summary of the report should be given sufficient information in the summary to understand the essence of the document without having to go into the finer details. References The data was obtained for this assignment came from the Australian Road Deaths Database. Questions 5. https://bitre.gov.au/statistics/safety/fatal_road_crash_database.aspx STA70005: SURVEY SAMPLING Assignment 2: Exercise in Comparing Sampling Methods General