WEB PAGE FOR CHAPTER 1 ADDITIONAL QUESTIONS MULTIPLE CHOICE QUESTIONS 1 Statistics are necessary in business research and investigations because: (a) raw data cannot be interpreted (b) of variability in the behaviour of humans, in events and equipment (c) populations are very similar (d) we need to classify data 2 A major benefit of an external consultant is: (a) cost (b) possible greater impartiality (c) ability to put findings into action (d) shorter timescale for research completion 3 Applied research: (a) extends the abstract frontiers of knowledge (b) is simplified research (c) is of no use for business (d) focuses on and may solve everyday business problems 4 In-house researchers have the disadvantage of being: (a) costly (b) perceived as a tool of management by suspicious employees (c) not as competent as outside consultants (d) unable to devote sufficient time to the research 5 Basic research is: (a) employer funded research (b) research conducted by an external consultant (c) research that extends the frontiers of knowledge (d) solves a particular problem in a company
CLASS ACTIVITIES and DISCUSSIONS 1 In groups, examine some recent newspapers and select two claims that are not accompanied by supporting evidence. How might support be obtained for the claims? Share your evidence and arguments. 2 The vice president of a petrol additive company states: ‘Letters from our satisfied customers prove that you can expect to get up to 30 percent better mileage by using our additive containing carbuferate.’ What comments would you make about this claim?
3 ‘If research shows something which I know from my experience to be wrong then the research is faulty. If research proved what I know already then what is the point of research?’ How would you respond to such a comment? Discuss in groups. 4 From the popular press bring into class articles that communicate facts in the form of statistical knowledge. In what ways is this information useful and how might it be used to make decisions and solve problems. Share your examples and ideas. 5 Describe a situation/research project where it would be better to engage an outside consultant rather than undertake it in-house. Explain why. 6 What is the purpose of descriptive statistics? 7 What is the purpose of inferential statistics?
ANSWERS TO MULTIPLE CHOICE QUESTIONS 1 (b), 2 (b), 3 (d), 4 (b), 5 (c)
WEB PAGE FOR CHAPTER 2 MULTIPLE CHOICE QUESTIONS 1 Positivist research aims to: (a) uncover socially constructed meanings (b) examine the positive results of change (c) discover universal laws that predict human behaviour (d) uncover surface illusions so people can change their world 2 If a research starts with a theory from which several hypotheses can be derived this research project can best be described as: (a) deductive (b) basic research (c) applied (d) inductive 3 In quantitative research collected data are represented by: (a) numbers (b) words (c) theories (d) hypotheses 4 The interpretivist paradigm is usually associated with: (a) quantitative data (b) deductive reasoning (c) qualitative data and inductive reasoning (d) qualitative data and deductive reasoning 5 When Human Resource Departments utilize standardized selection tests they are involved in: (a) critical investigations (b) the interpretative approach (c) subjective appraisal (d) the scientific approach 6 An operational definition is one that: (a) provides a definition of work operations (b) defines a variable in a way it can be measured (c) is a critical definition (d) cannot be measured 7 When results from the same study repeated by other researchers are the same we have: (a) an operational effect (b) replication (c) incongruency (d) a positivist paradigm 8 The research cycle is: (a) the annual application for research grants (b) the move from theory to hypothesis (c) the move from data to theory (d) the sequence of inductive and deductive approaches
9 When a subject tries to outwit the researcher we have: (a) an intelligent subject (b) a reactive subject (c) a non-conforming personality (d) a reaction from the researcher
DISCUSSION QUESTIONS 1 Discuss or debate in class the advantages and disadvantages of the scientific method and qualitative approaches to research. 2 Discuss or debate in class in which major ways the method of science differs from non-scientific methods of enquiry? 3 Fred Kerlinger, an American authority on research methods, once said, ‘There is no such thing as qualitative data. Everything is either 1 or 0’. Debate this statement.
CLASS ACTIVITY 1 Select any population of interest and identify quantitative and qualitative variables of the population that could be selected for study.
ANSWERS TO MULTIPLE CHOICE QUESTIONS 1 (c), 2 (a), 3 (a), 4 (c), 5 (d), 6
(b), 7 (b), 8 (d), 9 (b)
ANSWERS TO MULTIPLE CHIOCE QUESTIONS AT END OF CHAPTER 2 QU 2.6. (d), 2.7 (b), 2.8 (c), 2.9 (c), 2.10 (c), 2.11 (a), 2.12 (c).
WEB PAGE FOR CHAPTER 3 MULTIPLE CHOICE QUESTIONS 1 If it is inappropriate due to research design conditions to explain the reasons for the research to the subjects before data collection, which of the following should the researcher do? (a) inform them anyway since cooperation is vital (b) not disclose any information (c) tell the subject they will be informed at the end of the experiment (d) give them a false explanation 2 If investigating the consumer choice reasons of school children with respect to buying junk food or non-junk food at the school snack bar for lunch, which should the researcher do? (a) only use children who consent to answer (b) get the class teacher’s permission (c) get parental permission (d) offer money to those who will take part 3 If a university student drops out of a research study what should the researcher do? (a) require the student to find a substitute (b) amend the student’s personal records to note the fact (c) nothing (d) insist the student continue
CLASS DEBATES 1 Discuss the contention that: ‘The ends never justify the means as far as research with humans is concerned.’ 2 To what extent is the advancement of knowledge and the pursuit of information sufficient justification for overriding ethical values and ignoring the interests of those studied?
GROUP DISCUSSION QUESTIONS AND ACTIVITIES 1 Read through the following case scenarios and answer the queries. (A) A disease is inflicting itself on the whole human population. Morbidity rates are roughly 10% and many who do survive infection can be affected for the rest of their lives with symptoms ranging from being severely crippled, paralysed, and nerve damage. A medical researcher wants to inject an untested vaccine into his colleague’s wife and children to assess its effectiveness and potential side effects. What are the ethical implications related to this study. Should the study be undertaken? (B) A young boy has contracted an as-yet untreatable disease from an animal. Symptoms include encephalitis, flu-like symptoms, delirium, and eventually death. A medical scientist who, as an unlicensed physician, has no legal right to treat the boy, injects an experimental drug with which he has been working on. What ethical problem/issues are raised here? (You may be interested to know that Examples (A) and (B) were derived from real life. In Example (A) Jonas Salk tested a polio vaccine on his laboratory technicians, his wife and children. Salk was later to receive numerous awards for his pioneering work. In Example (B), Louis Pasteur, who also developed the pasteurization process for milk, risked serious legal recourse when he administered the first rabies vaccine to 9-year-old Joseph Meister, on 6 July 1885. The treatment was successful and Pasteur escaped prosecution, instead being lauded a hero.)
2 Is it ethical for a researcher to precode questionnaires in invisible ink so that they know who submitted each one? Is it ethical to use different coloured paper for questionnaires for different departments of a company? Are these two situations similar or different? 3 A researcher plans to use a one-way mirror to video test users’ reactions to a new alcoholic party drink in a simulated party environment. Are there any ethical issues involved here? 4 When you visit your favourite popstar’s website you are requested to complete a questionnaire. Unknown to you the popstar management plan to sell your personal information to a marketing company who are selling the popstar’s memorabilia via mail catalogue and email. What do you think the ethical issue is here? How could it be mitigated? 5 A researcher encourages subjects to take part in a survey by telling them it will only take 5 minutes when in fact it takes 20. Is this unethical? 6 A researcher becomes a participant observer in a project group to analyse inter-personal group relations and group decision making. None of the group know that the researcher is actually observing them and think he is a genuine member of the group. Does a researcher have the right to deceive others? 7 A health products company launched a new herbal product claimed to increase weight loss. A threemonth long research study reported that people using the product lost 5% more weight than those on a placebo with no reported side effects. Since then it has been criticized by those who have taken it that the weight losses are not maintained after four months and long-term side effects like migraine and heartburn occur after six months. What ethical issues are salient here? 8 The manager of a research company tells a new client that they should have a survey on their customers’ perceptions done before any survey on how their customers would react to new products which is what they originally wanted. The manager’s assistant knows that the company already has some data on their customer perceptions from a previous survey just done for a competitor. Why not present these results as an appendix to the second study and save the client money? The manager feels that repeating the study is not unethical even though they know what the likely results of the repeat study will be and that it would be more unethical to disclose work done for someone else. Who do you agree with? 9 The new CEO feels that the changes he has introduced have improved employees’ feelings about their work environment. To support his feeling he decides to carry out an organizational climate survey. He finds one used in a previous survey in the company before he took charge and decides to use it without consulting the firm that produced it and conducted the original survey. Is this unethical? 10 The CEO wants to survey the organizational effectiveness of his company. The questionnaire contains some quite sensitive items which the researcher is sure many employees will not answer or will lie to and suggests the items be rewritten and in some cases removed. The CEO instead requests that conversations of employees be recorded by hidden microphones in the departmental tea rooms and company dining room followed by content analysis. Should the researcher accept this change? 11 Suppose you are a salesperson for a new type of burger dressing. Your boss requests that you gather groups of friends together and visit all the test outlet locations and ask for that dressing in order to make it appear it is popular The company will pay for the purchases over the next two weeks to ensure large orders are placed and rival brands no longer bought by the outlets. Do you agree to this strategy? 12 Mary had just started a job with a business research firm. Her first job was a survey of customer perceptions of a supermarket chain. She attends a meeting with her boss when the results are presented to the client. She notices that some of the data has been altered in a very minor way to make it look more positive. When she mentioned this to her boss afterwards he argued that while objectivity is fine in the academic world, in the real world of business it is important to keep the client happy. Do you agree?
13 Create a checklist in question form for assessing whether a piece of proposed research is ethical. Use such questions as: 1 2 3
Will any person or definable minority group be able to be identified in the published research? Will the research involve persons who themselves are unable to give informed consent? NOW ADD TO THIS LIST
Compare your lists around the class and create a composite list that you could all use as a guide towards ethical research conduct when you next do some research. Remember that a ‘yes’ to any question does not mean abandon the research. It acts as a prompt for you to think about the research design and possible modifications to it, consider ways to avoid the problem and be ready to argue the case for what you want to do when facing your organization’s research ethics committee 14 (a) State two responsibilities of researchers. (b) State two rights of respondents/participants. (c) State two responsibilities of sponsoring clients. For each explain the reasons and issues involved.
A list of some Professional Associations and their Ethical Guidelines, with useful online sources: Academy of Management Australian Research Council British Computer Society (BCS) British Educational Research Association (BERA) British Market Research Association Centre for Business Ethics. Bentley College, Waltham, MA. Centre For Research Ethics, Goteburg University, Sweden Council of American Survey Research Organisations, Port Jefferson, NY. Economic and Social Research Council
http://www.aomonline.org/aom.asp?ID=185 http://www.arc.gov.au/grant_programs/research_ethics.htm http://www.bcs.org.uk/ethics/freedom.htm http://www.bera.ac.uk/guidelines.html
European Business Ethics Network. HIS Inc. Japan Marketing Research Association MRS National Health and Medical Research Council Qualidata.
www.eben.org
http://www.bmra.org.uk http://ecampus.bentley.edu/dept/cbe/ethicscentres/domestic.html www.cre.gu.se www.carso.org http://www.esrc.ac.uk/ESRCInfoCentre/Images/ESRC_Re_Ethics_F rame_tcm6-11291.pdf
http://www.ihs.com/Investor-Relations/ethics-policy.htm http://www.jmra-net.or.jp/guideline/kouryou-e.html http://www.mrs.org.uk/standards/codeconduct.htm http://www.nhmrc.gov.au/ethics/human/index.htm http://www.qualidata.essex.ac.uk/creatingData/confidentiality.asp
Additional information on ethical issues Institutional ethics committees Even when clear ethical standards and principles exist, there will be times when the need to do accurate research runs up against potential ethical issues. No set of standards can possibly anticipate every ethical circumstance. In order to ensure that researchers consider all relevant ethical issues in formulating research plans most institutions and organizations have set up Institutional Ethics Committees, a panel of persons who review grant and research proposals with respect to ethical implications and decides whether additional actions need to be taken to assure in particular the safety and rights of participants. These committees provide increased protection for both the organization and the researcher against potential legal implications of neglecting to address important ethical issues of participants. For professionals not located within institutions many professional organizations have issued ethical research principles or codes which members are expected to abide by or else suffer sanctions such as loss of professional registration. The American Association of Public Opinion Research code of ethics can be found at http://www. aapor.org/ethics/code.html Code of Professional Behaviour of the Australian Market and Social research Society (AMSRS). While based on marketing this code applies across all business areas. Access this code through the following website: http://www.mrsa.com.au/index.cfm?a=detail$id=115$eid= The Australian Psychological Society Code of Ethics can be found at: http://www.psychology.org.au/ aps/ethics/default.asp There is a degree of similarity between the ethical codes of all organizations, and clearly research with human participants needs to consider the implications and well-being for participants, clients and society.
Impact of technology on research ethics The masses of information on the Internet and ease of information transfer has made plagiarism far easier. New technology has also increased the likelihood that participants’ privacy can be violated. Data must not be sold or even given to other persons or organizations. This is particularly applicable to employee data and company financial data. Appropriate firewalls must be in place. Research in cyberspace should provide no special dispensation from the general ethical obligations already noted. The use of the World Wide Web is a relatively new research tool that provides a unique research environment because: z
the distinction between private and public space is unclear;
z
data can be easily collected without consent;
z
the participants and the researcher may never meet or speak to each other and the identities they may choose to assume may be 'virtual', bearing little resemblance to their 'real' self.
Not only does this create ethical implications for the collection of data in terms of privacy, anonymity and confidentiality, but it also points to the need to raise awareness about concerns regarding the mistaken inclusion of vulnerable or unsuitable populations (e.g. children) in a research project, who are not identifiable because of pseudonyms. The perceptions people have of what constitutes public and private domains on the Web may not correspond with their actions when they come to log on. People often use public domains in cyberspace for private conversations. The use of the terms 'private' and 'public' refer to the accessibility of information, not the individual's own perception of the privacy of their actions. The individual's perception of privacy may well be determined by who they believe to be looking at their work and what they believe is being done with it. As with all social research, it is recommended that the participant supplying the data, whether an individual author or a site owner, is consulted personally.
Researchers must NOT assume that where access to the Web is ‘public’, that the information available in such domains is also ‘public’ and 'up for grabs'. Extracting data in such a manner from the Web contravenes some basic ethical principles: z
the author's privacy has been invaded;
z
informed consent has not been obtained from the author;
z
the anonymity of the author is at risk.
There are currently very few guidelines for ethical codes of conduct regarding web-based social research. Getting consent from participants in an online location can be very difficult. This is largely because of the 'faceless' nature of Web participants and the possibility that they may also be assuming a pseudo-identity.
Cross-national research Given the global environment in which business operates, more research is being conducted across national and cultural boundaries. This raises special ethical and political issues relating to personal, cultural and national sensitivities and disparities in wealth, power, the legal status of the researcher, political interest and national political systems. Researchers should note that important regulations governing human subjects' research, Privacy, freedom of information and copyright may be different from their own country. In some developing countries, local people may be unaware of their ‘human rights’. Indeed, individuals can be subject to reprisal for raising these issues. It is important that researchers have an awareness of the local situation before embarking on their research. A potential misuse of power is insensitivity to different cultural perspectives. However, sensitivity to the values inherent in local practices does not require uncritical acceptance of them. What is required is a willingness to explore differences without prejudice and to seek, as far as possible, to understand them, informed by knowledge of local traditions and material circumstances. Researchers should also be aware of ethical issues that can occur when working within a repressive political climate. They should ensure that their research activities do not jeopardize the security of local participants, guaranteeing complete confidentiality where participants have taken serious personal risks to assist them. In some societies it would be culturally inappropriate for researchers to ask individuals to participate in research without consulting the community or gaining permission from community leaders.
Copyright issues Researchers making audio or video recordings should obtain 'copyright clearance' from interviewees if recordings are to be publicly broadcast or deposited in public archives. This is best done at the time of interview, using a signed form. Different countries have their own rules on data protection and copyright, so check up on what your country’s laws require. There are at least two different copyrights present within an interview. z
z
The words spoken: the copyright owner is the speaker, i.e. research participant. Included in this case are transcripts made either verbatim or later from recordings. The recording: the copyright owner is the person or organization who arranged for the recording to be made, i.e. the researcher or their funding body only has copyright over the physical recording.
Scope of copyright and 'fair dealing' Recorded speech which is 'in copyright' cannot be copied, passed on to others, played in public or broadcast without the copyright owner's permission. However, the spoken words in a recording, whether in sound or transcribed form, can be copied for private study, research, criticism or review. Short extracts can be used for illustration purposes in publications provided they do not form a 'substantial' part of the recording transcript in question, although what can be considered to be 'substantial' is difficult to quantify.
Transfer of copyright Copyright is a form of property and can be 'assigned' to another person or organization. These assignments should be made ideally in writing and signed by the copyright owner. Transfer of copyright from the participant to researcher will be needed if the researcher ever wants to publish large extracts of the participant's words, and can be carried out as part of the consent process with the participant assigning copyright of their spoken words to the researcher.
Useful material on some copyright issues Qualidata (2002a) Consent to use information given in the interview [Internet], Qualidata. available from http://www.qualidata.essex.ac.uk/creatingData/qualidataexampleconsentform.doc (Example of a form which could be used to assign copyright of the interview participants’ words to the researcher.) Thomas, B. (2001) Ethical Guidelines [Internet], Available from http://www.nmgw.ac.uk/~ohs/ ohs/ethics.html (Details of the responsibilities of researchers before, during and after an interview including the handling of copyright issues.)
Example of consent letter and information sheet These examples only contain the information and would in the real world be printed on headed paper with address and contact numbers at the top.
Example of Information Sheet Dear …. I am conducting a study of the way in which investment advisors make investment decisions and I am inviting you to participate in a structured survey which will be administered individually. Your employer BettaInvest Group have recognised the importance of this research and support your participation in it. Participation is voluntary and there is no penalty or loss of any benefits whatsoever if you choose not to participate. The survey will take approximately 25 minutes. If you are interested in participating you should know that: z
z z
You can cease participation at any time and ask to have your responses removed from the study without need to provide an explanation. You will not be personally identified. Your name will not be included on the data file. The results will form the basis of a research article and report to the Association of Investment Advisors. The results will be reported in general terms.
If you decide to participate, please contact me at the phone or email address above. Thank you for taking time to consider being involved in this project. Please feel free to contact me with any questions or concerns. Yours sincerely
Consent letter Dear ……. Thank you for offering to be involved in the investment decision project. Please note the following: z
z z
You can cease participation at any time and ask to have your responses removed from the study without need to provide an explanation. You will not be personally identified. Your name will not be included on the data file. The results will form the basis of a research article and report to the Association of Investment Advisors. The results will be reported in general terms.
Please read the following carefully: I understand the contents of the Research Project Information Sheet and this Consent to Participate letter. I agree to participate and give my consent freely. I understand how the study will be carried out as described. I understand that whether or not I participate it will not affect my employment conditions, and that I can withdraw at any time without providing a reason. I am satisfied that I understand my role in the project and how the results will be used and presented. Signed…………………………………………………(Participant) Date…………………………………
ANSWERS TO MULTIPLE CHOICE 1 (c), 2 (c), 3 (c)
WEB PAGE FOR CHAPTER 4 Below are some lists that you may find useful in conducting your literature search. Some useful journals Academy of Management Executive Academy of Management Journal Academy of Management Review Administrative Science Quarterly Annual Review of Psychology Asia Pacific Journal of Management Asia Pacific Journal of Human Resources Australian Journal of Management British Journal of Management British Journal of Administrative Management Business Europe California management Review European Business Review European management Journal Far Eastern Economic Review Group and Organization Management Harvard Business Review Industrial Relations Journal International Journal of Contemporary Hospitality management Journal of Applied Psychology Journal of Business Communication Journal of Business research Journal of Consumer Affairs Journal of Consumer Research Journal of Management Studies Journal of Managerial Psychology Journal of Occupational and Organisational psychology Journal of Organisational Behaviour Journal of Product and Brand management Leadership Quarterly Management Accounting Research Organisational Behaviour and Human Decision Processes Personnel Journal Personnel psychology Project management Journal Public Administration Review Training and Development Of course there are many others!!
Some useful databases Database name
More details
Subject area
ABI/Inform (via Proquest)
Full-Text
Academic Research Library (via Proquest) Academic Search Elite (via EBSCO) Accounting and Tax (via Proquest)
Full-Text Full-Text Full-Text
Asian Business (via Proquest) AusStats : Australian Bureau of Statistics (ABS) Banking information source (via Proquest) Business source premier (via EBSCO)
Full-Text Full-Text, Australian Full-Text Full-Text
Business, includes ABI/Inform Dateline, ABI/Inform Global and ABI/Inform Trade and Industry databases Arts, business, science Multidisciplinary Business and accounting newspaper articles. Includes Accounting and Tax newspapers and Accounting and Tax periodicals databases Asian business Australian statistics Business Business
CBCA Business (via Proquest) Connect 4
Full-Text Full-Text, Australian
Current Contents Connect (via Web of Knowledge)
Citations
Digital Dissertations (via Proquest)
Citations and abstracts
EconLit (via EBSCO) Emerald
Full-Text Full-Text
Europa World (via Routledge Reference)
Data
European Business (via Proquest) Gartner Core Research
Full-Text Full-Text
Informit
Full-Text, Australian
JSTOR
Full-Text
Canadian business Annual reports, company information Citations, table of contents and current alerting service for sciences and arts Dissertations and theses citations Economics Business, science, library and information management World politics, world economics, international organisations Business Information technology, business Multidisciplinary; 6 full-text and over 50 index/abstract databases; core Australian database Arts, business, science and education archives
Legal (via Proquest)
Full-Text
MasterFILE Premier (via EBSCO) Newspaper Source (via EBSCO) Oxford University Press (OUP) Journals ProductScan (via Datamonitor) Proquest 5000
Full-Text Full-Text Full-Text Database Full-Text
PsycINFO (via American Psychological Association)
Citations and abstracts
Regional Business News (via EBSCO) Safari Business Books online (via Proquest) SpringerLink
Full-Text Full-Text, e-books Full-Text
US National newspaper abstracts (via Proquest)
Full-Text
Web of Science (via Web of Knowledge)
Citations
Wiley InterScience
Full-Text
Law, business law, civil law, criminal law Multidisciplinary International newspapers Arts, business, science Business Multidisciplinary collection of over 30 full-text and index/abstract databases Psychology, psychiatry, medicine, nursing, sociology, education, pharmacology, physiology Business Business Arts, business, science, education, technology, medicine Newspaper articles from USA newspapers Science, arts and social science citation indexes, includes cited reference searching Science, business, education
WEB PAGE FOR CHAPTER 5 MULTIPLE CHOICE QUESTIONS 1 Customer satisfaction is an example of: (a) an independent variable (b) a dependent variable (c) an operational variable (d) an abstract concept 2 In a conceptual framework abstract concepts are represented by: (a) rectangles (b) ellipses (c) triangles (d) arrows 3 The process of developing concepts into variables is called: (a) describing the theoretical framework (b) describing the conceptual framework (c) operationalizing the variables (d) describing relationships 4 The repeated measures design uses: (a) two non-randomized groups tested twice (b) two randomized groups tested twice (c) one non-randomized groups tested twice (d) one randomized group tested twice 5 In a between groups design we have: (a) allocation to same sized groups (a) influence of individual differences on the treatment (a) non-randomized allocation to groups (a) control of individual differences 6 The problem with an intact group is that: (a) Sample size is too big (b) Random selection to the group has not occurred (c) There is no equivalent control group (d) Both (b) and (c) 7 A theory can be expressed graphically as a: (a) conceptual framework (b) theoretical framework (c) elliptical framework (d) propositional framework (e) hypothetical framework 8 A measurable concept is called a /an: (a) non-abstract concept (b) abstract concept (c) conceptual framework (d) variable (e) dependent
9
When there is no established conceptual framework for some research project and little previous published research on that issue exists, then a series of specific ………………… may need to be posed: (a) abstract concepts (b) research objectives (c) research questions (d) hypotheses (e) research hypotheses
10 A research problem can be narrowed down and focused in the form of hypotheses when: (a) previous research with detailed findings has been published (b) exploratory research needs to be undertaken (c) hypotheses need to be tested (d) little previous research has been done on that issue (e) there is no strong theoretical base to the research 11 A variable: (a) is something which is measurable (b) can be used to represent an abstract concept (c) forms part of the theoretical framework (d) is represented by a rectangle in a schematic diagram (e) all of the above 12 A statement of a predicted relationship among variables is called: (a) an educated guess (b) an hypothesis (c) a conceptual prediction (d) a theoretical framework
ADDITIONAL QUESTIONS, ACTIVITIES AND DISCUSSION ITEMS 1 Complete the table below to emphasize differences between types of research. Print the table out or copy to Word doc. Exploratory
Descriptive
Problem Definition Conceptual Framework Theoretical Framework Operationalization of Variables Hypotheses Precision required Common data collection techniques Measurement of causation Time and cost resources Advantages Disadvantages 2 Taking it in turn in pairs, explain to a peer member of your class: (a)
why we randomize allocation to groups;
(b) the purpose of counterbalancing; (c)
the purpose of operationalizing variables;
(d) the benefits of ‘blind’ and ‘double blind’ techniques; (e)
the role of a moderating variable.
Correlational
Experimental
3 Specify potential IV, DV and moderating variables in a study to investigate employee dis-satisfaction. Drawing frameworks will help. Work in groups and share your answers. 4 Specify potential IV, DV and moderating variables in a study to investigate choosing between two brands of the same item by adult customers. Drawing frameworks will help. Work in groups and share your answers. 5 Outline the design of an experiment that examines the research question: ‘does temperature affect employee productivity?’ Work in groups and share your answers. 6 Outline the design of an experiment that explores how the number of packets of an item sold is affected by the packet size (1kg or 2kg) and by gender of customer. Work in groups and share your answers. 7 What are the criteria for a true experiment? Provide a rationale for your answer. Work in groups and share your answers. 8 In groups devise operational definitions of (a) job involvement, (b) trade union activist, (c) downsizing, (d) brand awareness, (e) accounting principles, and (f) unemployment. Share them in discussion. 9 Read the following article to gain a grasp of the relative costs of various types of research: www.mrsa.com.au/index.cfm?a=detail&id1209&eid=91 10 A researcher would like to compare the number of defective items produced in a week by employees who have attended an intensive training programme on the production of these items with employees who have not attended such a course. Compare your anwers with those of your group. (a) What is the dependent variable of this study? (b) What level of measurement will be used? (c) What is the independent variable? 11 A researcher studies the factors that determine how many cars a family possesses. Compare your answers with those of your group. (a) Name some potential IV‘s (b) Name the DV (c) Name a potential moderating variable 12 In groups, devise conceptual frameworks and theoretical frameworks for the following: (a) Punctuality is associated with job satisfaction. (b) Hours of study per week affects final grade, particularly for those students who attend tutorials regularly.
ANSWERS TO MULTIPLE CHOICE QUESTIONS 1 (d), 2 (b), 3 (c), 4 (d), 5 (b), 6 (d), 7 (a), 8 (d), 9 (c), 10 (a), 11 (e), 12 (b)
ANSWERS TO ACTIVITIES IN CHAPTER 5 Qu. 5.2 Compare your frameworks with the ones below. They do not have to be exactly the same but close enough to permit relevant hypotheses to be drawn.
More meaningful work (DV)
Job enrichment (IV)
Conceptual Framework
Extent of autonomy
Extent to which Job is challenging
Perception of doing something worthwhile
Perceptions of more meaningful work
Using range of skills
Feedback availability
Possible theoretical framework (You may have other similar IV’s) Qu. 5.4 Problem statement : How can brand awareness be brought about to increase revenues for Emblem Hotels? Conceptual framework
Brand awareness
Increased revenue
Theoretical framework
Nature of building facilities
Brand awareness of users
Building maintenance level
Increased revenue
Service quality level Degree of franchisers’ cooperation
IVs
MV
DV/IV Or IVV
DV
WEB PAGE FOR CHAPTER 6 SPSS ACTIVITY Load Chapter 7 SPSS Data File A and undertake the following tasks. (a) Create a total score, mean score and percentage for variables q13 – q18 inclusive for all 80 cases. (b) Using the frequencies menu, check for input errors on the variables gender, age, the workload is too heavy and I am not given clear instructions. Obtain appropriate tables and graphs, etc. Print out your results and compare your results with those of other class members.
QUESTIONS 1 What type of scale is used (a) in a list of the order of finish in a marathon, and (b) when describing the sex of the runners in that marathon. 2 Is calendar time counted in months equal interval data? 3 What scale of measurement is most appropriate for each of the following? (a) attitudes towards currency exchange controls (b) numbering of houses along a road (c) scores on an accountancy test (d) annual income in dollars (e) government departments (f) Janice is the most popular member of the group (g) results of a beauty contest (h) time taken to complete a task (i) nationality 4 In a study investigating the role of practice in improving performance the IV is which variable? 5 What level of measurement is each of the following DV’s: (a) Employees are divided into those who have blue eyes and those who have brown eyes. (b) These two groups are then measured on the time required for their eyes to adjust to a sudden change in light intensity. (c) Each employee is also classified as a heavy smoker, moderate smoker, light smoker or nonsmoker. (d) Each is also classified as having either normal vision or better, or below normal vision.
Additional material – VARIABLES AND LEVELS OF MEASUREMENT The term variable refers to a characteristic that varies from person to person or case to case. If we are interested in the time commuters in Vancouver must spend on their drive to work each morning, the variable is time spent. In a study concerning the income of wage earners, the variable is income. These measurements need not correspond very well with everyday notions of measurement such as weight, distance, and temperature. For example, eye colour is a variable (because a set of people will include some with brown, some with blue and others with green eyes), as are types of vehicle and political parties. Thus measurement of a variable can involve merely categorization. Companies may use different categories such as department manager, supervisor, and administrative assistant, for classifying staff. These categories constitute a variable since they are different positions to which people can be allocated. Such categorization techniques are an important type of measurement in statistics. Variables may be classified as quantitative or qualitative.
Variable. A variable is the characteristic of the population that can take on more than one value and demonstrates variation from one observation, event or person to another.
Qualitative and quantitative variables Qualitative variables Qualitative variables are classified into categories according to the characteristics by which they differ rather than by ‘how much’. The data obtained with qualitative variables are limited to counting and classifying. For example, the marital status of credit applicants, the gender of students in the statistics class, ethnicity, political affiliation, the makes of car used by the Melbourne police force are all examples of qualitative variables. In every case, the observations are measured non-numerically as categories. Qualitative variables such as sex, ethnic group, branches of a company, and so on, are usually recorded through the use of code numbers, e.g. male = 1; female = 2, rather than by verbal labels. Of course these numbers have no mathematical significance!
Quantitative variables These are those which vary in quantity and are thus recorded in numerical form, for example, age, price, items sold per day. Quantitative variables differ in terms of how much individuals, objects, or events possess a given characteristic. Thus, the numbers we use to deal with quantitative variables can be added, subtracted, multiplied and divided. The incomes of all wage earners is an example of a quantitative variable. Other examples include the heights of all people we might be interested in if we are in the business of manufacturing clothes, scores students receive on the final examination in statistics, or the time spent in minutes in the queue at the supermarket. In each case, the observations are measured numerically.
Discrete and continuous variables Discrete variables A discrete variable is one that is restricted to whole numbers and can therefore take on only a specific set of values. They are characterized by gaps in which no real values exist. For example, the number of foreign born workers in a factory is a discrete variable. There might be 14 or 15 of these individuals but never 14.7. A person is either a male or female and cannot be assigned any value between the two. The number of companies in the country making bicycles and the number of computers in stock at a dealership are all discrete too. Typically, the data accumulated with discrete variables result from the process of counting. There may be 15 companies making bicycles, 12 computers in stock, and 34 immigrant employees. Discrete variables may, of course, be either qualitative (sex, marital status, country of birth) or quantitative (number of books in a library, number of items repaired this year under warranty) A discrete variable. Consists of separate indivisible categories so that no values can exist between two neighbouring categories.
Continuous variables A continuous variable is one that can assume any value, including fractional ones, within a range of values. Weight of content of tins of beans, temperature for baking different types of bread, tread wear on car tyres, distances on various delivery routes, scores on a stress test, daily calorie intake, and so on, are examples of continuous variables. An individual can be described as 25.5 years old and weighing 75.2 kilos.
To go from one unit to the next on a continuous variable, one must pass, theoretically at least, through a large number of fractional parts. For example, as the temperature increases it does not suddenly go from 24C to 25C or even from 24C to 24.1C but rather passes gradually through a succession of infinitesimally small changes as it increases. Therefore the measurement of continuous variables is always an approximation of the true value. No matter how accurately one measures, it is not possible to measure and record all the possible values of a continuous variable. As a result, continuous variables are measured to the nearest convenient unit and the recorded values of these variables are, in practice, discrete with definite predetermined points along the continuum chosen to report the variable values. A continuous variable. Is one that can take on an infinite number of intermediate values between any two identified values. Because a continuous variable is in real terms measured in specified intervals like millimetres, or seconds, we consider each interval to have a midpoint that represents the score or measure and bounded by upper and lower limits (often termed the real limits) which link it to the next interval. Thus as Figure 2.1 illustrates, the upper boundary or limit of one interval on a continuous scale falls at the lower limit of the next defined interval.
19.5
20.5 20
21.5 21
22.5 22
23.5 23
24.5 Real limits 24
mid-points
Lower and upper limits of each score are indicated by the longer lines. Figure 1 Illustration of real limits and midpoints of a continuous variable. Because of this limitation in accuracy with continuous data, the distinction between continuous and discrete variables is not always clear-cut and unambiguous. Take currency as an example. The money we carry on our persons and with which we pay our bills is discrete. However, in the world of finance, in which actual money may never change hands, money is frequently expressed in fractional terms. For example, exchange rates are expressed to four decimal places and, in the field of cost accounting, the cost per unit may be stated to several decimal places. In this context then, money is continuous. Alternatively, in some research studies, we often find variables that are inherently continuous being treated as though they were discrete. For example, age is a continuous variable, but subjects in a study may be placed into length of work experience categories, such as 0–5 years, 6–10 years, 11–15 years and so on, rather than using actual years and months. Salary is often placed into categories too.
ANSWERS TO QUESTIONS 1 ordinal; nominal 2 no; as months vary in days 3 a = interval; b = nominal; c = interval; d = ratio ; e = nominal; i = nominal 4 practice 5 a = nominal; b = ratio; c = ordinal; d = nominal
f = ordinal; g = ordinal; h = ratio;
WEB PAGE FOR CHAPTER 7 SPSS ACTIVITY Using the SPSS Chapter 7 Data File B 1 Obtain frequencies, descriptive statistics, and a variety of graphs and charts for the data for the variables age, hours worked per week and main method of transport. Choose the descriptives, graphs and charts you deem most appropriate. 2 Make new variables called ‘qTotal’ and qMean which will provide the total and mean for each of the two questionnaire items Qu1 and Qu2. 3 When you have done this, play around with the data. The best way to learn how to use SPSS is simply play around and see what it will do.
MULTIPLE CHOICE QUESTIONS 1 Statistics that describe typical scores are called: (a) models (b) standard deviations (c) Z scores (d) measures of central tendency 2 Statistics which describe the dispersion of scores around a typical score are measures of: (a) deviation (b) variability (c) dispersibility (d) skew 3 The term average is synonymous with the statistic called: (a) the mode (b) the mean (c) the median (d) the variance 4 The sum of deviations around the mean must equal: (a) 1.0 (b) –1.0 (c) 0 (d) a value that varies with the number of scores 5
The difference between the highest and lowest values is called: (a) interquartile range (b) the range (c) the quartile range (d) the semi interquartile range
6 When all scores have the same value: (a) the distribution is abnormal (b) the mean, median and mode are unequal (c) the standard deviation is equal to zero (d) the distribution is unimodal
7
For which of the sets of values is the variance largest? (a) 2,2,3,4,4 (b) 3,3,3,3,3 (c) 1,1,3,5,5 (d) 1,2,3,4,5
8
If the variance is 16 then the SD is: (a) 2 (b) 4 (c) it depends on sample size (d) 8
9
For which of the following sets of scores would the value of SD be the smallest: (a) 1114777 (b) 1234567 (c) 7777777 (d) 1112333
10 The disadvantage of using the range as a measure of dispersion is: (a) It is difficult to calculate (b) It is heavily influenced by extreme values (c) It is determined by only two points in the data set (d) It changes from one sample to the next 11 A mean can be calculated on data from which of the following scales? (a) nominal and ordinal (b) interval and ordinal (c) interval only (d) ratio and interval 12 In checking for input errors, which of the following would you use? (a) a box plot (b) a stem and leaf plot (c) a histogram (d) all the above 13 If you have ordinal data, which measure of central tendency would you use? (a) the mode (b) the standard deviation (c) the median (d) the mean
ADDITIONAL QUESTIONS AND ACTIVITIES 1 On what type of graph would it be appropriate to show the number of gold, silver and bronze medals won by a country’s athletes at the Asian Games. 2 Explain how a stem and leaf display contains more information than a grouped frequency distribution. 3 What kind of chart can be used to display negative quantities? quantities that arise in business.
Give two examples of negative
4 Show the data below in a pie chart after calculating the percentages. Household expenditure per week Food Housing Clothing Travel
$130.00 $536.00 $ 34.00 $100.00
5 A survey of where overseas tourists like to spend their vacations in your country was taken to determine what attractions should be advertised. The results were national parks beaches cities golf courses mountain climbing
15% 48% 8% 25% 4%
Construct a bar chart to display the results. 6 A study of the number of ‘bugs’ in software programmes analysed over the last eight years were: 8.5%, 8%, 7.5%, 7%, 6.4%, 6%, 5.6%, 5%. Construct a line graph and comment on the chart. 7 A sample of N = 20 scores has a mean of 5. What is the ∑X for this sample.? 8 Find the median of this distribution: 3, 10, 8, 4, 10, 7, 6 9 In a recent survey comparing picture quality of three brands of colour television, 63 people preferred brand A, 29 preferred B and 58 preferred brand C. What is the mode for this distribution? 10 Find the mode of the following distribution: 5, 6, 9, 11, 5, 11, 8, 14, 2, 11 11 Taking a Traffic Census. Which is the most commonly used method of personal transport in your area? Which are the busiest times of day? Develop a traffic census form for your own area. Produce sufficient copies to monitor every 30 minute period from 8 am to 5 pm. and arrange a rota to cover each period at the same point on a main road with separate recorders on each side of the road. Produce a report with appropriate tables, graphs ands charts on: (i) the popularity of each particular type of transport (ii) the pattern of traffic flow during the day Stages in this project are: (a) Preparation of the census form (b) Selection of an appropriate census point. Bear in mind the safety of the census takers during rush hour (c) Testing the form out on a half-hour period to discover any weakness (d) Drawing up a rota of recorders to cover the time period (e) The taking of the actual census (f) Analysis of data to find the usage of various types of transport (g) Analysis of data to reveal hour-by-hour movement of traffic (h) Production of the report
12 Survey of the Price of Houses. Obtain a copy of the local paper which contains house sales. Draw up a list of housing categories based on (a) the number of bedrooms, e.g. 1 bedroom, 2 bedroom.... over 5 bedrooms, and (b) types of housing, e.g. apartment, single storey house, double storey house….. Record the asking price for every house in each category. Calculate the mean price for each category. Display your data using appropriate tables, graphs and charts. 13 Examine some recent newspapers and magazines. Find some graphs, histograms etc. and consider whether they are providing information in a clear and simple way and that they are not conveying an erroneous impression. If you find any that do not appear to be clear, simple and intend to deceive, work out better ways to display the information. 14 Measure some variable that every person in your class possesses, e.g. height, weight, shoe size. Form a histogram or bar chart (depending on the type of data) and comment on any features of interest in the distribution. 15 Survey 40 people (20 males and 20 females) on the way they feel about the economic policies of the government. Use the following categories: excellent, above average, average, below average, poor. Construct a histogram showing the differences in opinion between males and females. 16 Explain in your groups what way the standard deviation as a measure of dispersal is different from the measure of range. 17 The representative of the employees' union in a small company complains that the average salary there is only £18,000. The owner of the company counters with a statement that the average salary is a whopping £25,000. The salaries are as follows: Discuss in your groups who is lying? X
f
£70,000
1
39,000
1
21,000
1
19,000
1
18,000
5
11,000
1
ADDITIONAL MATERIAL WAYS OF MISREPRESENTING NUMBERS – Misuse of descriptive statistics and graphical representations in reports Numbers presented in chart or graph form often cause people to suspend judgement and common sense, Homo sapiens becomes homo credens... ready to believe anything. 'A picture is worth a thousand words' it is often said. But what if it is a wrong or misleading picture? A good piece of advice for all users of other people's numbers, charts and graphs is: 'Do not believe everything you see, especially the parts you cannot see!'
Advocacy statistics What has to be acknowledged at the outset is that a great many descriptive statistics are used not in order to reach a valid conclusion but to prove the case, however invalid, the presenter wants to make. The real difficulty with much of the statistics presented for business, political, and social purposes lies in the difference between technical truth and advocacy truth. Technical truth should always include qualifying statements of uncertainties. The providers of much of our daily intake of statistical data are usually advocates of causes or points of view, either as protagonists or antagonists. In a selling situation (and it does not really matter whether you are selling products, policies, presidents, or prime ministers) one is advocating a case, where not only are the qualifications and uncertainties excluded but aspects of the technical truth which might be damaging to the presentation of the case are withheld and only the positive points of importance to the case are made. Emotive words, expressive phrases, are employed to create a more desirable impression than the bare facts warrant. Users of numbers are in the same position, and businesses concerned with the sale of products and services rely heavily on the art of advocacy. This is not to say that every statistic is intended to deceive, but there are many ways of presenting the same set of data. As a future manager or business person, the best safeguards you have is to possess sufficient knowledge and understanding of the basic techniques of collecting, processing and presenting data, including their uses and limitations, to have the confidence to question them and challenge them so that your business will not suffer through wrong decisions being made on the basis of deceptive statistics. Here is a basic checklist of questions that a user of statistical information should employ to assess the material being displayed to them by others.
The data z z z z z z z z z z z z z
What do they tell us? Who said so? Is the source stated? With what degree of confidence? Relating to what time period? Is there a short written or verbal summary? How precise are the figures? What degree of accuracy is acceptable? Were there sufficient data to justify the conclusion? Do they answer the question we want to answer? What is the range of the data? Which average is used – mean, median or mode? Percentage of what? Compared with what?
Charts and graphs z z z z z z z z z z z
What does it say? Does it say it clearly enough? Does it make sense? Why this particular form of chart or graph? Is there a better way of displaying the data? Is anything missing? Do comparative charts use the same scale? Would they be better on log scale? Is the comparison valid? Is the conclusion valid? Is there a brief written or verbal summary? Is the source stated?
About samples z z z z z z z z z z
What kind of sample is it (e.g., random, quota)? How big is it? When did the sampling take place? What degree of confidence can we have in the results? Are the results generally applicable? Are the sample details included in the report? Who is included in the sample and why? Who is excluded from it and why? How many people responded? How many did not reply?
It is not enough just to ask the questions. You have to know why you are asking them, sound as though you know why you are asking them, and understand the response and its adequacy. When putting the questions, you must give a firm impression that you know what you are about. As a general rule no set of data, either in tabular or graphic form, should ever be presented without comment, on the basis 'Here it is – sort it out for yourself’. Researchers should 'sort it out’' and provide a brief written or verbal summary of what the data show, why the data have been summarized the way they have been – i.e., reasons for choosing the particular representative figures – and what the main conclusions are that can be drawn from them. Increasing demands are placed on the creativity of those presenting raw data, central tendencies and percentages to come up with new and innovatory displays that can mislead the unwary. But remember that in every complex chart there is a simple statistic trying to get out. Following are some of the most common traps to be looked for.
Spurious precision – misleading means and plausible percentages There is something very impressive and reassuring about a number taken to several places of decimals. For example, 'In a recent test, Loo brand antiseptic disinfectant killed 99.25 per cent of all known germs– dead!' Really!! What about the other 0.75 per cent? But, in any case, does ‘really matter? Or what about an advertising claim that 'Independent laboratory tests' (how reassuring, especially when you see the whitecoated doctor-like figures in a clinical setting) show that children using Den Clean toothpaste have suffered 21.75 per cent fewer cavities over a trial period as compared with a control group using an 'ordinary' brand of toothpaste'? Does it really matter whether the figure is 21 per cent or 22 per cent? The statement that the average cinema-goer (whoever he or she may be) visits the cinema 12.48 times a year is not only saying something which is physically impossible but implies a degree of precision which is meaningless. It may be quite wrong to infer that the average cinema-goer visits on average a cinema once a month, since the pattern of visits may be four times a month over a three-month dark, wet and cold winter period and no visits during the rest of the year.
These are examples of spurious statistics which purport to say something terribly meaningful but offer a degree of precision totally unnecessary for the purpose in hand The pricing of many products provides another example of spurious precision. 'Shoes for only $59.99 per pair!' 'Drive away this car for $24,995 all-in!' What do you suppose is behind this pricing psychology'? Could it be intended to make us feel that we are getting a bargain - and so much better value than for $60 or $25,000'? Precision should not be confused with accuracy. – Indeed, precise figures tell us nothing about their accuracy. If a survey shows that the per capital consumption of chocolate last year was 2.423812 kg, this is very precise indeed – but, is it accurate? Many percentages are quoted out of context or with important additional or qualifying background information omitted. Here are some examples: 'Aircraft passenger deaths up by 10 per cent over last year.' Yes, but what was the increase in the number of aircraft passenger miles flown during the two periods being compared. Were there any unusual or special factors operating such as one major crash which can distort figures? 'Consumer advertising up by more than 300 per cent between 1996 and 2006.’ So what? There must be very few things that have not trebled in value over this 10 year period. The percentage is undoubtedly the most widely used statistic. Whenever we want to compare different sizes or different shares or different rates of change, we call on percentages. Index numbers (e.g., cost of living, retail prices, wage rates, production, imports and exports), markups, discounts are all variants of the percentage. The only defence against the phoney percentage is to ask questions: ‘percentage of what?’ or ‘of whom?’; ‘Compared with what or whom?’; ‘How many make up the whole 100 per cent?’; ‘Who are they?’
Playing fast and loose Loose comparisons are fairly common in marketing and advertising circles. If a dog food advertisement proclaims 'Now twice as meaty’ what does it really mean? – 'twice as meaty as what'?; twice as meatier as before?; twice as meatier in terms of' the measured meat equivalent in the average dog food, in all other dog foods, or the dog food with no meat in it at all? What does ‘50% more coffee flavour’ mean? What about 'Puffin cigarettes give you 10 per cent less tar'? Product claims like these always beg the question: better than, more than, less than . . . what really is the comparison? Unless the standard or yardstick against which the comparison is made is provided, then such statements have no real statistical meaning. Such comparative statements, of course, are intended to make a good and positive impression.
Scaling the heights – keeping histograms and polygons honest It is possible to draw graphs for the same group of data which give entirely different impressions – some are deliberately misleading. Graphs of the same data can convey entirely different impressions, as shown in Figure 1 (a) and (b), which report crime statistics for three similar suburbs. In A, cruising police patrol cars were eliminated during a 3 month trial period; B had 5 cruising cars during the period; while C was flooded with 15 cars. Your conclusions about the effects of patrol cars would probably depend on which graph you saw. Figure 1(a) gives the impression that the presence or absence of patrol cars is associated with dramatic difference in crime rate. Note, however, that the largest difference – 1,000 versus 970 – is only 3%. Such a small difference could just as well be attributed to chance factors or to differences in crime reporting procedures. The graph is misleading because it violates the height-width rule and because the Y axis begins with a frequency of 950 crimes instead of 0 crimes.
1000
1000 900 800 700 600 500 400 300 200 100 0
990 980 970 960 950
(a)
A A
B
C
A No patrol cars
B 5 patrol cars
C 15 patrol cars
(b)
A No patrol cars
B 5 patrol cars
C 15 patrol cars
Figure 1(a) and (b) Number of reported crimes in three similar neighbourhoods during a 3 month test period. Note how graph (a) falsely gives the impression of a great difference in crime rate across the three conditions. In another example, a government Finance Minister seeking a further period of office, might show the state expenditures during his term as illustrated in Figure 2. His opponent might show the same expenditures as illustrated in Figure 3 800 600 Millions $ 400 200
1994
1995
Figure 2 Finance Minister’s version
1996
1997
1998
800
750 Millions $
700
650
600
1994 1995
1996 1997
1998
Figure 3 Opponent’s version. Both histograms convey exactly the same information, but the Minister’s histogram provides an impression of minor and gradual increases in expenditure while the opponent’s emphasizes steeply rising expenditure. Essentially, the impression conveyed by a graph depends largely on the proportions chosen for the abscissa and ordinate. This illustrates why it is suggested that as a general rule a histogram or polygon should be about three-fourths as tall as it is wide, because this gives relatively undistorted proportions. It is also recommended that the units on the abscissa and ordinate begin at the zero point. However, this is not always possible especially with score values on the abscissa. If a part of the abscissa or ordinate has been omitted, then the reader's attention should be called to this fact by means of slash marks drawn to cut the axis. Most people in most countries of the world read horizontally but when it comes to numbers in tables and graphs they tend to think vertically. When we look at a graph we think in terms of 'up'– i.e., the vertical height of a point or line above the horizontal axis indicating the higher value or the bigger change in the value of, by convention, the dependent variable. Here is an example of distortion with a simple line graph.
Figure 4 shows three simple line graphs relating to the level of advertising in an organization’s expenditure over the period from 1996 to 2006.
250
100
100
200
80
90
150
60
80
100
40
70
50
20
60 50
1996 98 00 02 04 06
1996 98 00 02 04 06
1996 98 00 02 04 06
Figure 4 Distortion with scales. All three graphs are based on exactly the same data but visually they tell a different story. Which graph would you expect a large shareholder who is critical of the small amount of advertising to use, and which would you expect the responsible advertising manager to use in his response? Taken in isolation, there is nothing wrong with any of the graphs, but the impression given by each is markedly different. This is done by progressively (from left to right) cutting off the extreme ends of the vertical scale, which immediately causes the line to rise more steeply. By compressing the horizontal year-line, it could be made to rise even more sharply. Another trick which can be used is to begin the horizontal scale above the zero line. It probably would not matter too much provided any comparative graphs were lined up exactly the same way. The point is, however, that only one of the graphs would be used by the pro- and anti- advertising lobbies.
Changing the scale When comparing graphs, they should all should use the same vertical scale, which is only practicable when the values are roughly within the same range, or they should be reduced to a uniform scale which provides a fair comparison (i.e. ratio or log scale). While every sales and marketing manager looks for the silver lining in the company's sales charts, they would be well advised to look for the scalar lining as well. In other words, beware of the 'up' line and, equally, of the 'down' line that has been doctored to appear worse, by comparison, than the facts warrant. To take an example, suppose that two product managers are each putting forward a request for bigger marketing allocations for their separate products, based on past performance and future prospects. If each product manager makes his individual presentation to senior management, what sort of impression will they each make? The chances are of course, that the sales of each product will not be within the same range, in which case distortion will have a serious effect on making sound comparisons.
Figure 5 shows two alternative presentations to marketing management by two product managers, each responsible for products X and Y respectively. They each decide to show the other's figures on the same graph for comparison purposes when making their pitch for a bigger marketing budget. Graph A
Graph
500
500
100 400
400
B
80 300
X
300
Y X
200
60
200 Y
100
40 100 20
0
0 1996
97 98 99 00 01 02
1996 97
98
99
00 01
02
Figure 5 Sales of two products compared (arithmetic scale). The manager of product X produces graph A, based on a common arithmetic scale. The manager of product Y produces graph B, again using an arithmetic scale, only this time the vertical scales are different. He tried several alternatives but settled for this one because it put his product in the most favourable light in terms of relative growth. How can a reasonable decision be made? Only by placing them both on the same log scale. Clearly, the way in which information is presented can depend very much on the provider's or commentator's point of view and the case that they wish to advocate. There is nothing wrong in this, provided that the receiver or user of the information knows what the data represent and what has been done to them in tabular or graphic terms.
Big is beautiful When does two times two equal eight? When it's a pictogram of the type illustrated in Figure 6. a type much used by advertisers in particular.. What has happened in Figure 6 is that the product manager has doubled all the dimensions of the washing machine symbol. The volume of the washing machine on the right-hand side of the pictogram is, in fact, eight times that of the one on the left-hand side. In other words, the pictorial ratio is 1:8, whereas the sales ratio is 1:2. Since most people will make the comparison on a volume basis, the drawing gives a totally false impression of growth. Leaving out the third dimension (depth) does not provide the answer, since the areas of the front faces of the washing machine symbols are still in the ratio of' 1:4. Even putting volume figures above each of the symbols (e.g., 10,000 and 20,000 respectively) does not overcome the deception. If the manager wishes to show that sales have doubled then the correct way is to use a simple bar chart.
1996 Figure 6 Sales of Swirly washing machines has doubled.
2006
All the foregoing examples of' the statistical traps that lie in wait for the unsuspecting user of numerical data illustrate one fundamental principle. In business affairs, it is wise to assume that if anyone can draw the wrong conclusions from a given set of data, then they probably will (a variant of' Murphy's Law).
ANSWERS TO QUESTIONS IN CHAPTER 7 Qu. 7.1 (a)
5.
Qu. 7.1 (b)
bimodal
Qu. 7.2 18
ANSWERS TO MULTIPLE CHOICE QUESTIONS IN CHAPTER 7 1 (b), 2 (c), 3 (c), 4 (c), 5 (c), 6 (b), 7 (b), 8 (c), 9 (c), 10 (d), 11 (b), 12 (c), 13 (b), 14 (d).
ANSWERS TO WEBSITE QUESTIONS 1 Bar chart or pie chart. 2 You can see the individual values and determine the mode. 3 Two-directional horizontal bar chart. Financial losses; customer decrease compared to previous year. 4 Percentages are 16.25%; 67%; 4.25%; and 12.5%. 7 100 8 7 9 Brand A 10 11
ANSWERS TO MULTIPLE CHOICE QUESTIONS ON WEBSITE 1 (d), 2 (b), 3 (b), 4 (c), 5 (b), 6 (c), 7 (c), 8 (b), 9 (c), 10 (c). 11 (d), 12 (d), 13 (c)
WEB PAGE FOR CHAPTER 8 MULTIPLE CHOICE QUESTIONS – SET A 1 If the scores are clustered to the low end of the frequency distribution the distribution is: (a) unimodal (b) multi-modal (c) negatively skewed (d) positively skewed 2 If a distribution is unimodal and symmetrical then: (a) the mean is greater than the mode and the mode is greater than the median (b) the mode and median are equal but both less than the mean (c) the mean is less than the mode and the mode is less than the median (d) the mean, mode and median are equal 3 Skewness in a distribution can be determined by comparing: (a) the mode and the median (b) the median and the mean (c) the mean and the mode (d) any two measures of central tendency 4 For any distribution of raw scores the mean and standard deviation of Z scores are: (a) 1.0 (b) 0.1 (c) 0.0 (d) 1.1 5 A normal distribution always has: (a) a mean of 100 and an SD of 15 (b) a mean of 50 and an SD of 10 (c) a mean of 0 and an SD of 1 (d) a mean of 1 and an SD of 0 6 A probability of .05 means: (a) the event lies within the middle 95% of the distribution (b) the event lies outside the middle 95% of the distribution (c) the event will occur 50% of the time (d) the event will occur once every five times 7 The frequency distribution of a normally distributed set of values can be completely described by: (a) the mean and median (b) mean and SD (c) median and variance (d) median and SD 8 To calculate a Z score we need to know: (a) the mean and standard deviation (b) the raw score and the mean (c) the raw score and the standard deviation (d) the raw score, the mean and the standard deviation
9 A Z score of + 1.0 implies a value better than what percentage of the values? (a) 50% (b) 68% (c) 75% (d) 84% 10 The area under the normal curve is considered to be equal to: (a) 1.0 (b) 10 (c) 100 (d) varies depending on the size of sample 11 If a group of subjects have a mean score of 20 and a standard deviation of 4 on a test, approximately 95% of the scores lie between: (a) 16 and 28 (b) 18 and 22 (c) 16 and 24 (d) 12 and 28 12 In a normal distribution, what proportion of scores fall within the interval between the mean and one standard deviation above the mean? (a) 25% approx (b) 34% approx (c) 64% approx (d) 84% approx 13 A set of values has M = 75 and SD = 25. What is the Z score for a raw score of 100? (a) +1.0 (b) +2.0 (c) +3.0 (d) –1.0 14 In a normal distribution the area to the left of Z = 0 contains what percentage of the distribution? (a) 0 (b) 25% (c) 50% (d) 100% 15 What does p
