Week 71 (24.8.08)

Learn

The credibility of research findings (Chapter 5, Pg 149, Research Methods for Business Students)

Reducing the possibility of getting the answer wrong means that attention has to be paid to two particular emphases on research design: reliability and validity.

That is why good research design is important.

Reliability

Refers to data collection techniques or analysis procedures will yield consistent findings.
4 threats to reliability:

1. Subject or participant error: You should choose a more "neutral" time when employees may be expected to be neither on a "high", looking forward to the weekend, nor on a "low" with the working week in front of them.
2. Subject or participant bias: Elaborate steps can be taken to "ensure the anonymity"of respondents to questionnaires. Care should also be taken when analysing the data to ensure that your data are telling you what you think they are telling you.
3. Observer error: Three of us conducting interviews with potential for at least three different ways of asking questions to elicit answers.
4. Observer bias: There may have been three different ways of interpreting the replies!

Validity

Refers to whether the findings are really about what they appear to be about. This potential lack of validity in the conclusions was minimised by a research design that built in the opportunity for focus groups after the questionnaire results had been analysed.

6 threats to validity:

1. History: If the research is conducted shortly after a major product recall this may well have a dramatic, and quite misleading, effect on the findings.
2. Testing: If the operators believe that the results of the research may disadvantage them in some way, then this is likely to affect the results.
3. Instrumentation: In the above example, if the telesales operators may have received an instruction that they are to take every opportunity to sell new policies between the times you tested the first and second batches of operators. Consequently, the calls are likely to last longer.
4. Mortality: This refers to participants dropping out of studies, ie a year-long management development programme.
5. Maturation: In the earlier management development example above, it could be that other events happening during the year have an effect on their management style.
6. Ambiguity about causal direction: This is particularly difficult issue. Eg: What was not clear about was whether the poor performance ratings were causing the negative attitude to appraisal or whether the negative attitude to appraisal was causing the poor performance ratings.

Unlearn

NIL.

Relearn

Able to criticise study design more effectively. Cautious about the threats of study reliability and validity for my own research design in BAP.

Week 70 (17.8.08)

Learn

Chapter 12 - Analysing Quantitative Data

• Data for quantative analysis can be collected and subsequently coded at different levels of numerical measurement.
• Data are entered for computer analysis as a data matrix in which each column usually represents a variable and each row a case.
• All data should, with few exceptions, be recorded using numerical codes to facilitate analyses.
• Where possible, you should use existing coding schemes to enable comparisons.
• For primary data you should include pre-set codes on the data collection form to minimise coding after collection. For variables where responses are not known, you will need to develop a codebook after data have been collected for the first 50 to 100 cases.
• You should enter codes for all data values, including missing data.
• Your data matrix must be checked for errors.
• Your initial analysis should explore data using both tables and diagrams. Your choice of table or diagram will be influenced by your research question(s) and objectives, the aspects of the data you wish to emphasise, and the level of measurement at which the data were recorded. This may involve using:

1. tables to show specific values
2. bar charts, multiple bar charts, histograms and, occasionally, pictograms to show highest and lowest values
3. line graphs to show trends
4. pie charts and percentage component bar charts to show proportions
5. box plots to show distributions
6. scatter graphs to show relationships between variables

• Subsequent analyses will involve describing your data and exploring relationships using statistics. As before, your choice of statistics will be influenced by your research question(s) and objectives and the level of measurement at which the data were recorded. Your analysis may involve using statistics such as:

1. the mean, median and mode to describe the central tendency
2. the inter-quartile range and the standard deviation to describe the dispersion
3. chi square, Cramer's V and phi to test whether two variables are significantly associated
4. Kolmogorov-Smirnov to test whether the values differ significantly from a specified population
5. t-tests and ANOVA to test whether groups are significantly different
6. correlation and regression to assess the strength of relationships between variables
7. regression analysis to predict values

Unlearn

NIL. Mostly new learning

Relearn

As for the BAP assignment, the survey i undertook are only quantitative questionnaires to understand the perceptions of SEs on leadership across supervisory levels in Pfizer. With these results, i am going to analyse these quantitative data using tables and diagrams (bar charts) to show specific values in answering my research question(s) and objectives. As this research is not looking at any relationship, therefore, no correlation or regresssion analysis is required.

Week 69 (10.8.08)

Learn

Chapter 7 : Selecting samples

For many research questions and objectives, it will be impossible to either collect or to analyse all the data available owing to restrictions of time, money and ofter access.

Sampling techniques provide a range of methods that enable to reduce the amount of data needed to collect by considering only data from a subgroup rather than all possible cases or elements.

Sampling provides a valid alternative to a census when:

• it would be impracticable for you to survey the entire population
• your budget constraints prevent you from surveying the entire population
• your time constraints prevent you from surveying the entire population
• you all collected all the data but need the results quickly

Sampling techniques available can be divided into 2 groups:

• probability or representative sampling - the chance, or probability of each case being selected from the population is known and usually equall for all cases.
• non-probability samples - the probability of each case being selected from the total population is not known and it is impossible to answer research questions or to address objectives that require you to make statistical inferences about the characteristics of the population. You may still be able to generalise from non-probability samples about the population, but not on statistical grounds.

Unlearn

NIL

Relearn

Learning from the above, for my BAP assisgnment, I chose to use non-probability sampling because it can increase the accuracy of the results by not having any influence on the sample population. The questionnaire sent to them does not focus on whom i select but instead base on the convenience of the SEs to provide the information of the survey through "free post".

Unlike intranet survey for probability sampling, respondent may felt bias in responding because of the sensitivity in sending back the results via intranet (researcher know who send back the results of the survey).

Week 68 (3.8.08)

Learn



Chapter 11: Collecting Primary Data using Questionnaires

The design of questionnaire will affect the response rate and the reliability and validity of the data you collect. Response rates, validity and reliability can be maximised by:

• careful design of individual questions
• clear layout of the questionnaire form
• lucid explanation of the purpose of the questionnaire
• pilot testing
• carefully planned and executed administration

Self-administered questionnaires are usually completed by the respondents. Such questionnaires are administered electronically using:

1. Internet
2. Intranet
3. Posted to respondents who return them by post after completion
4. Delivered by hand to each respondent and collected later
5. Telephone questionnaires

Your choice of questionnaire will be influenced by a variety of factors related to your research question(s) and objectives(s):

• characteristics of the respondents from whom you wish to collect data
• importance of reaching a particular person as respondent
• importance of respondents' answers not being contaminated or distorted
• size of sample you require for your analysis, taking into account the likely response rate
• types of question you need to ask to collect your data
• number of questions you need to ask to collect your data

Unlearn

NIL

Relearn

Learning from the above choice of questionnaire, the design of questionnaire differs according to how it is administered and, in particular, the amount of contact i have with the respondents.

Weighing the advantages and disadvantages in administering the questionnaires, as the topic of my business research is on "leadership", it may pose high sensitivity and distort the respondent result if i chose answering through intranet (even though anomynity and confidentiality is assured) because intranet-mediated questionnaires administered in conjunction with email has direct answers from the user which may result them not comfortable to share the accurate information with me.

Therefore, delivered by hand is the technique that I will choose weighing the benefits over the risk in achieving an accurate results for a better survey outcome. Closed questions can be designed so that responses may be entered using optical mark readers after questionnaire has been returned to ensure a smooth data input and easier data analysis.