Professional Documents
Culture Documents
Research Method (Done)
Research Method (Done)
Gather information
Analyzing data participation of workers in total quality management.
Determine factors associated with the problem
Solve problem by taking corrective measures. Applied Research
- Research information is neither intuitive nor haphazardly - Conducted when a decision must be made about a
gathered. specific real-life problem
- Literally, research (re-search) -“search again” - Examples:
Should McDonalds add Italian pasta dinners to its menu?
- Business research must be objective Business research told McDonald’s it should not?
- Detached and impersonal rather than biased Should Procter & Gamble add a high-priced home teeth
- It facilitates the managerial decision process for all bleaching kit to its product line?
aspects of a business. Research showed Fresh and White toothpaste would sell
well at a retail price of RM4.40
Definition of Research Advertisers conduct research to find about the impact of
An organized, systematic, data-based, critical, objective, their promotional campaign.
scientific inquiry or investigation into a specific problem Political parties may conduct research to find out public
undertaken with the purpose of finding solutions. opinion about their policies and the likelihood voters will
(Uma Sekaran) support them in future elections.
6. Data Analysis
- Here, data gathered are statistically analyzed to see if the
hypotheses generated were adequately supported.
Hypothetical-Deductive Method E.g.: Conducting correlational studies to see the
relationship between greetings and satisfaction (i.e two
1. Observation factors).
- First stage of process. E.g.: Conducting a t-test to see the difference between
- Sensing certain changes occurring, new behavior, male and female managers performance.
attitudes and feeling surfacing in one’s environment. - Two types of data analysis:
Eg: sales dropping, frequent production problems, Qualitative analysis – information gathered in narrative
incorrect accounting results, low yielding investment, form through interviews and observations. Data can then
poor participation of workers, absenteeism, increase be categorized.
accident or near-miss incidents. Quantitative analysis – information gathered in value
form of measurement such as numbers, ratios,
percentage, mean, sum etc.
7. Deduction 1.5 Research Design
- The process of arriving at conclusions by interpreting the
meaning of the results of the data analysis. Involves a series of rational decision-making choices on
E.g.: It was found that sales assistants greetings is issues relating to:
positively correlated with customer satisfaction. 1. Purpose of the study (exploratory, descriptive,
Therefore we can deduce that customer satisfaction can hypothesis testing)
be increased by sales assistants constantly greeting 2. The study setting (location)
customer at the store entrance. 3. Type of investigation
E.g.: It was found that there is no difference between 4. Extent of researcher interference (studying or
male and female managerial performance. Therefore we manipulation)
can conclude that men are no better than female in their 5. Time horizon
managerial ability in financial institutions. 6. Unit of analysis (individuals, groups, organizations)
Variables
A variable is anything that can take on differing or
varying values. The values can differ at various times for
the same object/person, or at the same time for different
object/person.
Examples:
- Production units: One worker can varying unit of output
per minute.
- Absenteeism: The number of absent worker can vary
each day.
- Motivation: Level of workers’ motivation varies among
them.
Types of Variables
Four (4) main types:
1. Dependent variable (Criterion)
Variable of primary interest to the researcher.
Researcher wants to understand and describe or explain
it’s variability or predict the dependent variable.
The researcher will be interested in quantifying and
measuring the dependent variable, as well as the other
variables that influence it.
Refer to textbook (pg. 88) for example and exercise.
Hypothesis Development
Definition of Hypothesis – a logically conjectured
(theorized) relationship between two or more variables
expressed in the form of a testable statement.
Formulating testable statement is called hypotheses
development.
Statement in a form of If - then.
- E.g. Student who read more frequently will score a better
grade: If student read more frequently, then they will
score a better grade.
Directional and Non-directional Hypotheses
- Directional : Men are better financial analyst than
women.
- Non-directional: Men financial analysis are different (or
same) the women financial analyst.
Unit of Analysis
- Unit of Analysis is the level of aggregation of the data
collected during subsequent data analysis stage.
- Our research question determines the unit of analysis.
Individual u.a – Data gathered from each individual
and treating each employee’s response as an
individual data source
E.g. Prob. Stat – What is the motivational level of
our employees?
Dyad u.a – data gathered from two person in a
group and treating each group’s response as Population, Element, Population Frame, Sample and
separate data source. Subject
E.g. Prob. Stat – how to raise - Population – refers to the group of people, events, or
supervisor-subordinate relationship at work? things of interest that the researcher wishes to
Group u.a - data gathered from different group and investigate.
treating each group’s response as separate data E.g. computer firms in Silicon valley, all blue-collar
source. workers in the mining industry in the country etc.
E.g. Prob. Stat – is there a difference among
children of different age group in their attitude - Elements – a single member of the population.
towards learning? E.g. each blue-collar worker in the mining industry.
- Individuals do not have the same characteristic as groups - Population Frame – a listing of all the elements in the
(e.g. structure, cohesiveness) and group do not have the population from which the sample is drawn.
same characteristics as individual (e.g. IQ, stamina) E.g. payroll list, student list, list of computer firms
etc.
- It is necessary to decide on the unit of analysis when we
formulate the research question because data collection - Sample – a subset of the population, i.e. some but not all
methods, sample size, and variable used may depend on element from the population being studied.
the unit of analysis used. E.g. 200 person from 1000 blue-collar worker in a
firm.
Time horizon - Subject – a single member of the sample.
- Cross-Sectional Studies – a study where data are E.g. each person in the sample of 200 blue-collar
gathered just once, over a period or days/weeks/months worker.
in order to answer a research questions.
E.g. data taken for a period between April - June Sampling
last year to study it behavior during slowdown - Sampling is a process of selecting a sufficient number of
phase. elements from the population, so that a study of the
E.g. a one time survey done to determine the effect sample and an understanding of its properties or
of certain advertising campaign by some characteristics would make us possible to generalize the
companies. population’s elements.
- Longitudinal Studies – studying people or phenomena at - The characteristic of the population such as µ
more than one point in time in order to answer a research (population mean), σ (population standard deviation),
question. and σ2 (population variances) are referred to as its
E.g. data taken before and after a change in top parameters.
management to study its effect on employee loyalty - All conclusions drawn about the sample under study are
towards organization. generalized to the population.
E.g. data taken between April – June from year
2005 – 2009 to study its trend during mid year - Sample statistic X (sample mean), S (standard deviation),
stock evaluation. and S2 (variation in the sample) are used as estimates of
the population parameter.
Probability sampling
Simple Random Sampling (SRS) – every element in the
population has a known and equal chance of being selected
as a subject.
- E.g. lucky draw 1st of 1000, 2nd of 999, 3rd of
998……
- SRS has the least bias and offer the most
generalizability.
- Sometime become cumbersome, expensive and not the
most efficient design.
Complex Random Sampling – similar to SRS but offer a viable
and more efficient approach. Five most common complex
probability sampling:
1. Systematic sampling,
2. Stratified sampling,
3. Cluster sampling,
Cluster sampling Purpose sampling
- Not very common sampling technique in organizational - Process of obtaining sampling from specific target group
research. or from those who have the specific information, or may
- Takes cluster or chunks of elements that have have certain criteria set by the researcher.
heterogeneity (differences) among members within each - Two major types of purposive sampling:
group that are chosen for study. 1. Judgment Sampling
- Cluster samples offer more heterogeneity (differences) 2. Quota Sampling
within group and more homogeneity (similarity) among
group – the reverse of stratified random sampling. Judgement sampling
- Expose to greater biases and the least generalizable of all - Involves the choice of subjects who are most
probability sampling design, because most naturally advantageously placed or in the best position to provide
occurring cluster in the organizational context do not the information required.
contain heterogeneous elements. E.g. what it takes for student to achieve dean’s list
in their academic performance. The sample would
- The only advantage is lower cost and convenience in be those students who have achieved dean’s list in
sampling. their examination results for few semesters.
- It does not offer much efficiency in terms of precision - Judgment sampling demands special effort to locate and
and confidence in the results. gain access to the individuals who have the requisite
information.
Area sampling
- Used when limited number or category of people have
- A form of cluster sampling within an area. the information we want.
- Area sampling constitutes geographical clusters such as - Results are not generalizable to the population being
countries, city blocks, or a particular boundaries within a studied.
locality.
E.g. sampling the needs of people in Dungun before Quota sampling
opening a fast food restaurant to gather information
on taste, attitudes or behavior. - This sampling ensures that certain groups are adequately
represented in the study via assignment of quota.
- Less expensive than most other probability sampling and
not dependent on a population frame. - Quota is fixed based on the total number of each group
E.g. using city/town maps to allow researcher to in the population.
sample areas/blocks in obtaining data about - A form of proportionate stratified sampling based on a
residents. convenience basis (non-random).
- Saves time, cost and effort for researcher to conduct
Double sampling research.
- Undertake when further information is needed from a - Ensures that all subgroups in the population are
subset of the group from which some information has adequately represented in the sample.
already been collected for the same study.
- Results are not generalizable to the population being
- 1st sample taken for preliminary studies then subsample studied, but can be used effectively as a preliminary
of this primary sample is used to examine matters in research before in-depth research being undertaken or
more detail. vice-versa.
Non-probability sampling
- Elements in population do not have any probability
attached to their being chosen as sample subjects.
- Findings from the study of the sample cannot be
confidently generalized to the population.
- Can only be employed only if researcher is less
concerned with generalizability.
- Two categories of non-probability sampling:
- Convenience Sampling
- Purposive Sampling
Convenience sampling
- Collection of information from members of the
population who are conveniently available.
- E.g. distributing questionnaire at a shopping mall asking
shoppers opinion / preferences about product X over
product Y.
- Mostly used during exploratory phase of a research
project to get basic information quickly and efficiently.
Determining sample size - The level of confidence can range from 0 to 100%.
- The second aspect of sampling design is estimating the - A 95% confidence is the conveniently accepted level for
sample size. most business research.
- It is a question of how big is the sample required that we - This is expressed by denoting significance level as p
will be able to precisely (accurately) generalize our < .05
findings to the population with confidence (less doubt).
- A reliable and valid sample should enable us to Sample Data, precision and confidence in esmation
generalize the findings from the sample to the population - When we use sample data to draw inferences about the
under investigation. population, we hope to be fairly “on target” (precision
The sample statistic should be reliable estimates and confidence) and have some idea on the extent of
and reflect the population parameters. possible error.
In reality it is impossible to have a sample x closely
reflecting the population µ. - Because a point estimation provides no measure of
possible error, we do an interval estimation to ensure a
- Therefore to obtain confidence we need to set an interval relatively accurate estimation of the population
estimates within where the µ will lie, with probabilities parameter.
attached to it.
- E.g. sample of 64 customers taken via systematic
Precision sampling. x=105, S=10, x is the sample mean to estimate
µ, we could construct a confidence interval around x to
- Precision refers to how close our estimate is to the true estimate range which µ will fall.
population characteristic.
- The standard error Sx and the level of confidence will be
- Usually, we would estimate the population parameter to used to determine the width of the interval using the
fall within a range, based on sample estimate. following formula: µ=X + KSx where K is the t
- Precision is a function of the range of variability in the statistic for level of confidence required
sampling distribution of the sample mean.
- Different samples taken from a single population will
provide different mean of each sample, but are normally
distributed and have a dispersion attached to it.
- The smaller this dispersion (variability), the greater the
probability that the sample mean (x) will be closer to the
population mean (µ).
- This variability is called the standard error, denoted by
Sx
- The standard error is calculated by the formula:
- The width of the interval has increased and now we are
- S - is the standard deviation
less precise in the estimation of the population, but we
- n - is the sample size are now a lot more confident about our estimation.
- Sx - is the standard error or extent of precision - If we want to maintain our original precision while
offered by the sample. increasing confidence, or maintain confidence while
- The standard error (Sx) varies inversely with the square increasing precision, or to increase both, we need to have
root of the sample size (√n). The larger the n the lesser a larger sample size.
the standard error. Therefore if we want to reduce the
standard error we need to increase the sample size. The sample size , n, is a function of:
- Note that the smaller the variation in the population, the - The variability in the population
smaller is the standard error which implies that the - Precision or accuracy needed
sample size need not be large i.e. low variability in the
- Confidence level desired
population requires a smaller sample size.
- Type sampling plan – SRS vs. Stratified RS
- To summarize, the closer we want our sample results to
the population characteristic, the greater will be the
precision we would aim at. The greater the precision we
require, the larger is the sample size that we must collect.
Confidence
- Confidence denotes how certain are we that our
estimates will really hold true for the population.
- In the previous example we have more confidence on the
How to Determine The Sample Size
40 – 60 range than the 45 – 55 range.
Example : A manager wants to be 95% confident the expected
- The narrower the range the lower the confidence. There monthly withdrawals in a bank will be within a confident
is a trade-off between precision and confidence. interval of +RM500. S = RM3,500. What is the sample size he
- Confidence reflects the level of certainty with which we need?
can state that our estimates of the population parameters,
based on our sample statistics, will hold true.
Topic 8 : Measurement of variable
Nominal scales
- A measurement that allows researcher to assign certain
categories or groups which are mutually exclusive (i.e.
no 3rd category which respondent would normally fall Topic 9 : Measurement scales
E.g. Gender: (1) Male/ (2)Female.
E.g. Race: (1) Malay (2) Chinese (3) Indian Rating scales
- Normally displayed in frequencies and measured in Rating scales often used in organizational research:
percentages. - Dichotomous scale
- Category scale
Ordinal scale
- Likert scale
- Categories of variable in such a way to denote
differences among the various categories, it also - Numerical scales
rank-orders the categories in some meaningful way. - Semantic differential scale
E.g. ranking preference from best to worst or first - Itemized rating scale
to last by assigning numbers 1,2,3…to each level.
- Fixed or constant rating scale
- Ordinal scale provides more information than nominal
scale by distinguishing categories by rank-ordering them. - Stapel rating scale
- However it does not give any indication about the - Graphic rating scale
magnitude of the differences among the ranks…i.e. - Consensus scale
difference between ranking 1-2 is not the same as
difference between rank 3-4. Dichotomous scale
- Is used to elicit a Yes or No answer.
Interval scale
- A nominal scale is used to elicit a response.
- Interval scale allows certain arithmetical operation to be
performed on data collected from respondents. - E.g. Do you own a car?
Yes ____ No _____
- It let us measure the distance between any two points on
a scale.
Category Scale
- Consists of categories of a variable (nominal) arranged in
order (ordinal) and measures the magnitude of the - Uses multiple items to elicit a single response.
differences among individual. - Also nominal scale is used to elicit a response.
- It provide an indication about the magnitude of the - E.g. What is your race?
differences among the scale. i.e. difference between Malay ____
point 1-2 is the same as difference between point 3-4. Chinese ____
- A more powerful scale than the nominal, and ordinal Indian ____
scale and can measure central tendency (mean), the African ____
dispersion (range), standard deviation and variance. European ____
Others ____
Ratio scale
Likert scale
- Ratio scale not only measures the magnitude of the
difference between point on a scale but also taps the - Designed to examine how strongly subject agree or
proportions in the differences. disagree with a statement on a 5-point scale.
- Most powerful of the four scales because it has a unique - This is an interval scale and the differences between any
zero origin and combine all the properties of the other two points on the scale remain the same.
three scales. - Conducted over a summated of several items measuring
E.g. a person weighing 200 kg is twice as heavy as a particular concept or
one who weigh 100 kg. (Note multiplying or response
dividing both this number by any number given
number will still preserve the ratio 2:1)
E.g. age, income, height, distance etc
Semantic Differential Scale
- Several bipolar attributes are identified at the extremes of
the scale and a respondent are asked to indicate their
attitude.
- Bipolar adjectives used: Good-Bad, Strong-Weak,
Hot-Cold etc.
Responsive ___ ___ ___ ___ ___ ___ Unresponsive Graphic Rating Scale
Beautiful ___ ___ ___ ___ ___ ___ Ugly - A graphical representation helps the respondents to
Courageous ___ ___ ___ ___ ___ ___ Timid indicate on this scale their answer to a particular question
by placing a mark at the appropriate point on the line.
Numerical scale - This is an ordinal scale.
Similar to semantic differential scale, with a difference that
number on a 5-point or 7-point scale are provided, with
bipolar adjectives at both ends.
Forced choice
Fixed or constant sum scale - Paired comparison scale is used when, among a small
- Here, respondents are required to distribute a given number of objects, respondents are asked to choose
number of points across various items. between two objects at a time.
- Very much like an ordinal scale - Paired sample comparison is good method if the number
of stimuli is small.
- E.g. In choosing a car, indicate the importance you attach
to each of the following five aspects by allotting points - Forced choice enables respondents to rank objects
for each to total 100 in all. relative to one another.
Safety ___ - Easier for respondents, if the number of choice to be
Power ___ ranked is limited in number. E.g. Rank 1 most preferred,
Brand name ___ 5 for least preferred.
Design ___ - Utusan Melayu ___
Color ___
- New Straits Times ___
Total points 100
- Berita Harian ___
Stapel scale - Harian Metro ___
- This scale simultaneously measures both the direction - Malay Mail ___
and intensity of the attitude toward the items under
study. Comparative scale
- The characteristic of interest to the study is placed at the - Comparative scale provides a benchmark or a point of
center with a numerical scale ranging, say +3 to -3, on reference to assess attitudes toward the current object,
either side of the item. event, or situation under study.
- Does not have a zero point and is an interval scale. - E.g. In a volatile financial environment, compared to
- E.g. Assessing leadership in organizations stocks, how wise or useful is it to invest in Treasury
bond? Please circle the appropriate response.
Goodness of measures Parrallel Form Reliability
- It is important or make sure that the instrument the we - Occurs when responses on two comparable sets of
develop to measure a particular concept is accurately measures tapping the same construct are highly
measuring the variable, and that in fact, we are actually correlated.
measuring the concept we really want measure. - Both forms have similar items and the same response
- Scale develop could often be imperfect, and errors are format, the only changes being the wordings and the
prone to occur in the measurement of attitudinal order or sequence of the questions.
variables. - If the forms are highly correlated (say .8 and above) we
- We need to assess the “goodness” of measures we may fairly certain that the measure are reasonable
develop so we can be reasonably sure that the instrument reliable.
we use in our research is valid (logical) and reliable
(dependable). Consistency of measures
- To ensure measures developed are reasonably good we - The internal consistency of measures is indicative of the
need to begin with item analysis first, and subsequently homogeneity of the items in the measure that tap the
establishing the reliability and validity of these measures. construct.
Items should “hang together as a set” and be
Item analysis capable of independently measuring the same
- Item analysis is done to see if the items in the concept so that the respondents attach the same
instruments belong there or not. overall meaning to each of the items.
Each item is examined for its ability to discriminate - Two test of reliability:
(separate) between those subjects whose total 1. Interitem Consistency Reliability
scores are high, and those with low scores. 2. Split-Half Reliability
- Reliability tests how consistently a measuring instrument
measures whatever concept it is measuring. Interitem reliability
Concern with stability and consistency of - A test of the consistency of respondents’ answers to all
measurement. the items in a measure.
- Validity tests how well an instrument that is developed - The most popular test of interitem consistency reliability
measures the particular concept it is intended to measure. is the:
Concern with whether we measure the right 1. Cronbach’s coefficient alpha (Cronbach’s alpha), which
concept. is used for multipoint-scaled items, and
2. Kuder-Richardson formulas, used for dichotomous
Reliability items.
- Reliability of a measure indicate the extent to which it is - The higher the coefficient, the better the measuring
without a bias (error free) and hence ensures consistent instrument.
measures across time and across the various item in the
instrument. Split Half reliability
- Reliability of a measure is an indication of the stability - Split-half reliability reflects the correlations between two
and consistency with which the instrument measures the halves of an instrument.
concept and helps to assess the “goodness” of a measure. - The estimates would vary depending on how the items in
the measure are split into two halves
Stability of measures
- The ability of a measure to remain the same over time – Validity
despite uncontrollable testing condition or the state of the - Reliability is necessary but not sufficient condition of the
respondents themselves – is indicative of its stability and test of goodness of a measure.
vulnerability to change in the situation.
- Validity ensures the ability of a scale to measure the
- Two test of stability: intended concept.
1. Test-Retest Reliability
2. Parallel-Form Reliability - When we ask a set of questions (i.e., develop a
measuring instrument) with the hope that we are tapping
Test-Retest Reliability the concept, how can we be reasonably certain that we
are measuring the concept we intended to measure and
- The reliability coefficient obtained with a repetition of something else?
the same measure on a second occasion.
- Several types of validity tests:
- A set of similar survey questionnaire were distributed 1. Content Validity
now, and then the second time (several weeks to 6 2. Criterion-Related Validity
months apart) on to the same group of respondents, to 3. Construct Validity
assess its correlation scores.
- If the coefficient is high, thus the better the test-retest Content validity
reliability, and consequently, the stability of the - Content validity ensures that the measures includes an
measures across time. adequate and representative set of items that tap the
concept.
The more the scale items represent the domain or Topic 10 : Questionnaire
universe of the concept being measured, the greater
the content validity. Questionnaire
- A panel of judges (i.e. experts, professionals) can attest - Definition: A questionnaire is a preformulated written
(testify) to the content validity of the instruments. sets of questions to which respondents record their
answers, usually within rather closely defined
Criterion - related validity alternatives.
- Criterion-related validity is established when the When to use a questionnaire:
measure differentiate (distinguish) individuals on a - When researcher knows precisely what information is
criterion it is expected to predict. needed.
- This is done by establishing concurrent validity or - When large numbers of people are to be reached in
predictive validity. different geographical region.
Concurrent validity - is established when the scale - When groups of people can be assembled in conference
discriminating individuals who are known to be room to whom questionnaire can be administered, and
different; that is, they should score differently on collected immediately.
the instrument. Advantage of questionnaires
Predictive validity - indicates the ability of the
measuring instrument to differentiate among - Helps researcher obtain data fairly easy.
individuals with reference to a future criterion. - Information from questionnaires is easily coded.
- See example 9.12 in textbook. - Benefits the scientific community if the measures are
well validated and are reliable.
Constrict validity - Often is a catharsis for respondents.
- Construct validity testifies (shows) to how well the
results obtained from the use of the measure fit the Administering Questionnaire
theories around which the test is designed. Questionnaire can be administered:
- This is assessed through convergent and discriminant - Personally face-to-face individually or in group.
validity. - Mailed to respondents
Convergent validity – is established when the
- Electronically distributed
scores obtained with two different instrument
measuring the same concept are highly correlated.
Personally Administered Questionnaire
Discriminant validity – is established when, based
Questionnaire can be personally distributed when:
on theory, two variable are predicted to be
uncorrelated, and the scores obtained by measuring - The survey is confined to a local area,
them are indeed empirically found to be so. - The organization is willing and able to assemble groups
of employees to respond to the questionnaires at the
Testing Goodness of Measures: Forms of Reliability and work place.
Validity Advantage:
- Completed questionnaire can be collected quickly form
respondents,
- Any doubts on any questions in the questionnaire can be
clarified on the spot.
- Researcher can introduce the research topic and motivate
respondent to offer frank response.
- Less expensive and save time when conducted in large
group of respondent at the same time.
- Does not require much skill to administer the
questionnaire compared to interviews.
Mail Questionnaire
Advantage:
- Main advantage is a wide geographical area can be
covered in the survey.
- Respondents can complete the questionnaire at their
convenience, at their homes, and at their own pace.
Disadvantage:
- Low return rate of questionnaires.
- Any doubts cannot be clarified.
- Very low return rate make it difficult to establish
representative ness of the sample because those
responding to the survey may not at all represent the
population they are supposed to.
Ways to improve response rate: Open-Ended vs. Closed Questions
- Send follow-up letters. Open ended question allows respondents to answer them in
any way they choose.
- Enclosing some token or monetary incentives with the
questionnaire. - E.g. What are the things that you like about your
supervisor?
- Provide the respondent with self-address, stamped return
envelope. - E.g. Comment on the performance of this portfolio
investment.
- Keep the questionnaire as brief as possible.
- Closed ended question ask respondent to make choices
- Notify respondent in advance, introductory letter using a among a set of alternatives given by the researcher .
reputed research organization
- Any item in a questionnaire using nominal, ordinal, or
Principle of wording refers to: Likert or ratio scale are considered closed ended.
- The appropriateness of the content of the questions, - E.g. Which age group do you belong to.
1) 21 to 30
- How questions are worded and the level of sophistication 2) 31 to 40
of the language, 3) 41 – 50
- The type and form of questions asked, 4) 51 – 60
- The sequencing of the questions and, 5) above 61
- The personal data sought from the respondents. - E.g. Female are more hardworking than male student.
1-Strongly Agree, 2-Agree, 3-Undecided, 4-Disagree,
Content and purpose of the questions 5-Strongly Disagree
- The nature of the variable tapped – subjective feelings or
Positively and Negatively Worded Questions
objective fact will determine what kind of questions will
be asked. - A good questionnaire should include both positively
Subjective nature (feelings) - such as satisfaction, and negatively worded questions
involvement where respondent’s beliefs, perception - Phrasing questions positively and negatively will
and attitudes are to be measured, the questions minimize respondent’s tendency from mechanically
should tap the dimension and elements of the circle the points towards one end of the scale.
concept - E.g. - Measuring Achievement
E.g. 5 point Likert scale for 4 – 6 item 1. I normally accomplishes completing many task on time
(questions) given to me by my lecturer. (positive)
Objective measure – such as age, and education 2. I have occasionally failed to complete my assignments
level of respondents are tapped, a single direct on time. (negative)
question preferably one that has ordinal scaled set
of categories would be appropriate
E.g. 1- PMR, 2 – SPM, 3-STPM, 4-Diploma,
5-Bachelor, 6-Masters
Loaded Questions
- Avoid phrasing question in an emotionally charged
manner.
- E.g. To what extent do you think that management is
likely to be vindictive if the Union decides to go on
strike?
Social Desirability
- Avoid wording question that elicit socially desirable
responses.
- E.g. Do you think that HIV infected person should be
separated from the society?
- E.g. To what extent do think that HIV infected person be
given fair treatment anywhere?
Length of Questions
- Ask simple questions than a too long ones.
- Rule of thumb, a question or a statement in the
questionnaire should not exceed 20 words, or exceed one
full line in print.
Sequencing pf Questions
- The funnel approach.
- Sequence from general to specific question.
Ask organizational questions and then to
departmental and job related questions.
- From relatively easy to progressively more difficult
questions.
Ask routine question and then gradually proceed to
more analytical and problem solving questions.
- Do not place contiguously (side by side) a positively
worded and a negatively worded question tapping the
same element or dimension or concept.
E.g. Learning is the most difficult thing to do.
E.g. Learning is not the most difficult thing to do.
- Reword them differently and place them distant apart in
the questionnaire.
Topic 11 : Data Collection: Fieldwork
Unstructured interviews
Data Collection: Fieldwork - Interviewer enter the interview setting with an a
unplanned sequence of questions to be asked of the
Sources of Data respondent.
- Data can be obtained from primary or secondary sources. - Used to identify broad problem area (explore and probe)
- Primary data – refers to information obtained firsthand by interviewing all levels of employee (respondent).
by the researcher on the variables of interest for the - Objective is to bring some preliminary issues to the
specific purpose of the study. surface so that the researcher can determine what
- E.g. individuals, focus groups, panel of respondents, variables need further in-depth investigation.
internet questionnaire. - The type and nature of the question asked might vary
- Secondary data – refers to information gathered from according to the job level and type of work done by
sources already processed by other means or existing in them.
records and reports. Broad and open-ended question were asked to get initial
- E.g. company records or archives, government views about the situation and narrowed down after
publication, industry analysis offered by the media, web getting good responses or commitment by the
sites, Internet etc. respondent.
- Researcher must politely encourage /motivate respondent
Data collection method to answer questions.
- Data can be collected in a variety of ways, in different - Researcher listens carefully and take note or detect
settings, field or lab and from different sources. important messages respondent might convey.
- The choice of data collection methods depend on the
facilities available, the degree of accuracy required, the Structured Interviews
expertise of the researcher, the time span of the study, - Interviews conducted when it is known at the outset
and other associated costs (start) what information is needed.
- Three (3) main data collection in survey research: - Largely used in qualitative research.
Interviews: (Structured and Has the advantage of - Questions developed were obtained from information
Unstructured) flexibility in terms of collected during unstructured interviews.
- Face to face adapting, adopting and
- Question are likely to focus on factor that have surfaced
changing questions as the
- Telephone researcher proceed with the
during unstructured interview which are relevant to the
- Computer assisted problem.
interview.
- Electronic media - The interviewer has a list of predetermined questions to
Questionnaires: Has the advantage of be asked of the respondents either personally, through
obtaining data more telephone, or internet PC.
- Personally administered
efficiently in term of the - Based on respondent’s answers, an experienced
- Sent by mail researcher time, energy and researcher may ask other relevant questions not on the
- Electronically administered costs. interview protocol to identify or reinforce new factors
Observation; Has the advantage of and get deeper understanding of the research variables.
- Individual and events with accuracy of data.
or without videotaping
/audio recording.
Bias in Interviews
- Information gathered during interviews should be as free Topic 12 : Data Processing and Analysis
as possible of bias (errors, inaccuracies in the data).
Preliminary Step Before Data Analysis
- Bias can come from interviewer, interviewee, or the To ensure that we get reasonably good and quality data, four
situation. (4) preliminary steps need to taken before data analysis:
From Interviewer: 1. Getting data ready for analysis.
- When no trust or rapport not establish with the 2. Getting a feel for the data.
interviewee 3. Testing the goodness of the data.
- When responses either misinterpreted or distorted. 4. Testing the hypotheses.
- When interviewer unintentionally encourage/discourage
certain types of response through gestures and facial
expressions.
From Interviewee:
- Didn’t give true opinion but provide information they
think what the interviewer like to hear.
- Didn’t understand the question and feel difficult/hesitant
to seek clarification.
- Personal likes and dislikes (dress or interviewer,
manners etc)
- Answer in socially acceptable manner, rather than true
sentiments
Getting Data Ready for Analysis
From Situation:
- Nonparticipants – unwillingness or inability of the Editing Data
interviewee to participate in the study.
- Data need to be edited after they were collected through,
- Trust levels and rapport established – no sincerity, questionnaires, interview and observation (primary
unsuitable location/place of interview. source) or through secondary sources.
- Insufficient / unavailability of time - Editing should be done preferable the very same day the
- Inconsistency of question/wording among respondent to data are collected so that the respondents can be
respondent. contacted for further information or clarifications to
avoid confusion.
Ways to Minimize Bias - Incoming mail responses should be checked for
- Establish credibility and rapport, and motivating incompleteness and inconsistencies.
individuals to respond. - By scanning through certain question the researcher can
- Question technique - Funneling approach help to complete some related unanswered question.
- Ask unbiased questions – leading, loaded, expression Handling Blank Responses
etc. - Not all respondents answer every item in the
- Clarifying issues – restate or rephrase important questionnaire, probably, because respondent fail to
information given by respondent understand the question, did not know the answer, not
willing to answer, or simply indifferent to the need to
- Help respondent to think through the issues. – ask respond to the entire questionnaire.
question in simpler way, rephrase, respond, remind, or
give examples as guidance. - If 25% and above blank responses, advisable to exclude
this questionnaire and report the number of questionnaire
- Take notes – make written notes while interviewing and being rejected due to excessive missing data in the final
immediately after the interview. Record interview on report submitted to sponsor.
tape if the respondent allows, however this can cause
bias in response. - If few (2-3 items) out of 30 items not answered in an
interval-scaled item with a mid-point would be:
1. Assign mid-point to that particular item missing (most
common), or
2. Allow computer to ignore the blank responses during
data analysis (most common), or
3. Assign to the item the mean value of the responses of all
those who have responded to that particular item, or
4. Give the item the mean of the responses of this particular
respondent to all other questions measuring this variable,
or
5. Give a random number within the range for that scale.
Coding, Categorization and Entering Data Types of Data Measurement
- Please proceed to SPSS in class data analysis workshop. - Please refer to the questionnaire developed by you.
- Have your laptop in ready mode with SPSS software - Identify the measurement in each question of the
installed. questionnaire.
- Make sure you bring along your filled questionnaire Nominal
form with data collected for your survey project. Ordinal
Interval
Data analysis Scale / ratio
Reliability Test
- Reason: To identify the stability and consistency of
variables (questions) and its measurement used in the
questionnaire.
- Applicable for Ordinal and Scale/Interval/Ratio data of
identical measurement only.
- To perform data recoding go to (Variable/Data View)
Analyze
Scale
Reliability Analysis...>> (scroll all variables with
identical measurement on the left side and place in the
Items: box on the right).
….follow lecturer’s guide ………. Click OK
- Try testing other variables
- 10 mins