Chapter 1 asddf asdff sadf asf asdf asdf asff aasdf asdfs dad PDF

Preview

Summary

asdfsad asddf asdff sadf asf asdf asdf asff aasdf asdfs dad...

Description

Part I Data Exploration 1. Data Collection

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 32

Learning Outcomes At the end of this chapter you should be able to: 1. 2. 3. 4. 5. 6.

state the sources of data; identify and discuss aspects of sampling; explain the sources and consequences of bias in sampling; explain and identify response bias; explain and identify selection bias; explain and identify non response bias;

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 33

Contents 1.0 1.1 1.2 1.3 1.4

Introduction Sources of data Surveys Survey instrument Selecting the sample

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 34

1.0 Introduction      

Decisions should be based on evidence. Evidence in quantitative disciplines is in the form of data. Obtaining data is therefore the first step in this process. The research question or question of interest should be clearly and unambiguously formulated. This is an aspect that is often ignored, leading to meaningless inquiry. This then leads to determining what data is required. Not obtaining the “correct” data to answer the question of interest is a common problem. The next step is obtaining the data.

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 35

1.1 Sources of data      

Primary data—collected directly by the researcher for a particular purpose. Secondary data—collected by someone else, but still useful for answering question of interest. Obtaining primary data is an expensive and time consuming exercise. Always first see if the required data exists and is available. Sometimes secondary data will be free, and other times it will attract a cost. Many agencies collect data and then sell them to interested parties. If secondary data is not available then primary data must be obtained.

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 36

Data repositories     

Reserve banks Bureau of statistics UN WHO World Bank

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 37

1.2 Surveys Most data in business and social sciences is obtained by a survey.  We design a set of questions to ask people and the select a sample of people to ask these questions.  Thus there are two important aspects to surveys: selecting the sample and designing the questions or survey instrument.  If we do these well then the data from the survey will give us an insight into population at large.



STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 38

1.3 Survey Instrument Questionnaire design is a large and complex area. The way a question is written and the overall questionnaire design can affect the responses.  Generally most surveys start with demographic questions, but sometimes these are at the end of the survey.  Response bias refers to issues with the way questions are asked (or the way people answer questions).  Note that response bias does not refer to people being ”biased” and deliberately lying on a survey. That is a whole different issue! 

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 39

Some examples of Response Bias Leading questions - ”Do you support the government’s decision to turn back the boats?” versus ”Do you support preventing the deaths at sea of refugees seeking asylum in this country?”  Social desirability - people want to present themselves in a favourable light, whether in reference to personal habits or views on contentious issues (perhaps even fooling themselves). How do we ask questions about time spent exercising, overeating, or views about abortion or same-sex marriage?  Yea saying - some people have a tendency to say yes or answer in the positive. This can especially be a problem when we ask things like ”were you satisfied with ....” 

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 40

1.4 Selecting the sample The sample must be selected in such a way that it is representative of the population. If not then the results may be compromised.  We call it selection bias where a subset of the population is systematically selected according to some attribute. Equally, a subset of the population is excluded based on some attribute. Consequently, proper randomisation is not achieved.



STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 41

Some examples of Selection Bias Spatial or Location bias: only certain locations are sampled from. This can be a physical location, or more general, such as where we sample people of one occupation.  Temporal or time bias: Samples are taken at fixed times or in a narrow window of time. Again only a certain part of the population is sampled from.  Incorrect population is sampled from. For example, the question of interest is the attitude of Perth residents to hours prices in Perth, but the sample includes residents from other cities.  Self-selection bias occurs when respondents select themselves into the sample. That is, the surveyor does not select the sample—people choose whether to respond or not. Examples are radio and TV station polls and online polls.



STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 42

Non response bias Related to selection bias is the problem of non-response. Non-response occurs when a respondent does not answer one (or more than one) item.  This can lead to bias if the people who respond are different to the people who do not respond.  This could be because he variable being measured is related to the non-response. For example, questions on salaries can lead to non-response if people on higher (or lower) incomes are less likely to respond.  Non-response bias can also come about in another way, for example in an area of high illiteracy sending out questionnaires will give a high non-response rate amongst the illiterate, or forcing a survey to be filled out electronically may loose a lot of response from older people. 

Non-response is not a problem if the population of people who do not respond has similar characteristics as the population of those people who do. The only way to definitely know this is to survey the non-respondents! STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 43

Example 1.1 The 1936 US Presidential Election      

In the 1932 US presidential election, Democrat candidate Roosevelt won a sweeping election victory over the sitting Republican president, Hoover. In his re-election campaign in 1936, the Literary Digest conducted a survey. The survey was conducted by sending out 10 million ballots to voters selected from their own subscribers, telephone books and motor vehicle registrations. Over 2.3 million ballots were returned, and this is comparatively a very large sample. The results were published and indicated a landslide victory for the Republican challenger Alf Landon. The election was actually won by Roosevelt with a large margin, winning 46 of the then 48 states.

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 44

What happened? 

The magazine had run a poll for every election since 1920 and correctly predicted the election outcome in each case.



The magazine tried to determine what had gone wrong.



Polls of this nature suffer from the problem of non-response bias, that is, people choose if they will respond or not, and this can lead to a very biased sample.



For example, only people who feel strongly about an issue will respond to phone-in or on-line surveys conducted by television and radio stations.



The results of such surveys are not reliable, even if the sample size is large.



This poll probably also suffered from bias in that during the depression years those who had magazine subscriptions, telephones or cars were quite wealthy by the standards of the time.

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 45

Elections affect businesses! Election results affect business, as government corporate policy may depend on the ruling party’s philosophy. For example, a party whose policy is to reduce tax, if in government, will change the way business operates and invests. Reference: http://www.scribd.com/doc/259298/Why-the-1936-Literary-Digest-Poll-Failed

STAT1520: Economic and Business Statistics

R Nazim Khan and Marty Firth – 46...