Question

5. A simple random sample of n300 full-time employees has been selected from a company list containing the names of all N-5000 full-time emplovees in order to evaluate job satisfaction. This list represents the entire workforce. There are no part-time emplovees. For each of the following possible survey errors, provide a brief definition of the error and a possible example in each case. a. Coverage Error b. Non-response Error

Answer 1

Answer)Before going into specific parts of the question, let us see what is the overall theme of this question, the question is looking at sampling errors which could happen in a survey, in the given case company is doing a job satisfaction survey and has taken 300 samples from 5000 employees, hoping that it will rightfully reflect the mood of entire workforce and can help them rectify their fallacies.

Now let us move to specific parts of the question and see the definition of survey errors, general example of that error and possible example in this particular case

a) Coverage error-As the name suggests when there is not proper coverage of the entire population in the sample it is known as coverage error, in other words, if the ratio between the entire population 's various groups and groups in the sample is not 1:1, it is known as coverage error, there could be over coverage of certain groups in sample and under-coverage of other groups in sample which will lead to incorrect representation of entire population.In coverage error, error is due to wrong methodology followed.Let us see one general example below

General example-There are elections in a country, five candidates are standing for President 's post, a survey agency wants to know which age group is favoring which candidate, this agency has divided age groups into four parts 18-25,25-40,40-60 and 60+. Total population of that country is 10 million,total voting population is 8 million(rest 2 million are not aged enough to vote), out of this 8 million 3 million are in the 18-25 age group,2 million in 25-40,2 million in 40-60 and 1 million in 60+. Now the agency decides to sample 0.1% or 8000 of people to see the mood of the country, the survey team called people in their offices and asked for their opinion, since 60+ aged people are retired, this agency did not get any coverage of 60+ age group and its survey suffered from coverage error.

In the given case, let us say there are 3000 marketing employees,1000 finance employees,500 management people and 500 human resources people totaling 5000 people, but in survey of 300, survey team decided to survey people from 9 am to 11 am only but marketing department starts working from 12 noon so marketing department will not be covered in the survey and this survey will suffer from coverage error.

b) Non-response error occurs when respondents selected for the survey do not give responses because they are unwilling to or are not available which leads to less representation of that group in which they belong. Continuing with our general survey of the previous example, this time surveyor went to homes of people and was able to find all aged groups people but people in 25-40 aged groups were not interested in these surveys and thus survey will suffer from non-response error. In the given case let us say marketing people fear that human resources will have access to survey data and thus declines to answer questions which will lead to a distorted survey result.

c) Sampling error-Sampling error occurs when the random sample taken for the survey is not true representation of the total population, major difference between coverage error and sample error is that in coverage error methodology is wrong so even if you increase the size of the sample you will still get wrong coverage but in sample error it is the randomness which is causing error,in sampling error you can increase the size to make it closer to true representation or select a new random sample,continuing our general example let us say our surveyor selects random 2000 houses and visits them,he gets 8000 responses but out of 8000 7000 were people in age group of 25-40 or 40-60,so this survey will suffer from sampling error but if he decides to survey another 2000 houses he may get a better representation,in given case let us say 300 names pulled out consists names of management and human resources only ,this will lead to sampling error in survey.

d) Measurement error-Measurement error(also known as observational error) is difference between actual value and measured value of an observation,it is divided into two components- systematic and random,systematic meaurement error occurs when there is a flaw in system observing opinion/value, that is methodology of sample is incorrect,it happens owing to inaccuracy and can be eliminated if system is corrected, random error is present owing to chance and could be absent or present in next observation,random error is unpredictable and unlike systematic error cannot be eliminated.In our general example let us say surveyor wants to know which candidate will respondent vote for but he s/he asks respondent who do respondent think will win,now this is a systematic error as most of the repondents will think strongest candidate will win but some will vote for other weak candidate who they like for his/her promises,so survey will show 90% people voting for candidate A but on results day it is 70% as remaining 20% people only mentioned winnable candidate but not the candidate they are going to vote for.Random error for same case could be when surveyor asks whom they will vote for and respondent responds with candidate A but surveyor mistakenely notes candidate B,this is random observational error.In our given case let us say that our surveyor wants to know if our respondents are dissatisfied with their jobs but asks question are they planning to switch their jobs,most of respondents may not be satisfied with their jobs but are not looking to switch so they will say no and surveyor will do a systematic measurement error of saying most people are satisfied with their jobs.Random error could be when respondents are asked if they are happy with their work and few respondents misunderstood it as if they are happy with their performance and answers yes even if they dont like what they do at organization.