Second Doctoral Summer School Sozopol June 7 11

Second Doctoral Summer School Sozopol, June 7 -11, 2007 QUALITATIVE and QUANTITATIVE RESEARCH METHODS Assoc. Prof. Dr. Zhelyu Vladimirov Faculty of Economics and Business Sofia University St. Kliment Ohridski

METHODS and TECHNIQUES METHODS TECHNIQUES (What to do? Why to do? ) (Haw to do? ) Historical review and analysis Focus group Case studies Observation Field experiments Interviews (personal, by mail, by telephone) Surveys Step by step procedures to gather data and analyse them

QUALITATIVE OR QUANTITATIVE DATA COLLECTION? ¡ ¡ The choice depends on which type of data is needed for a particular research problem The main difference between qualitative and quantitative research is not “quality” but procedure It is quite possible to quantify qualitative data or to exercise qualitative analysis of the quantified data Qualitative and quantitative methods are not mutually exclusive

THE DIFFERENCE IN EMPHASIS IN QUALITATIVE VERSUS QUANTITATIVE METHODS (1) QUALITATIVE METHODS ¡Emphasis on understanding ¡Focus on understanding from respondent’s point of view ¡Interpretation and rational approach ¡Observations and measurements in natural settings ¡Subjective “insider view” and closeness to data ¡Explorative orientation ¡Process oriented ¡Holistic perspective ¡Generalization by comparison of properties and contexts of individual organism QUANTITATIVE METHODS ¡Emphasis on testing and verification ¡Focus on facts or reasons for social events ¡Logical and critical approach ¡Controlled measurement ¡Objective “outsider view” distant from data ¡ Hypothetical deductive; focus on hypothesis/theory testing ¡Result oriented ¡Particularistic and analytical ¡Generalization by population membership

THE DIFFERENCE IN EMPHASIS IN QUALITATIVE VERSUS QUANTITATIVE METHODS (2) Qualitative methods Research problems focusing on: Person’s experience or behaviour; Uncovering and understand a phenomenon about which little is known Employ a limited number of observations Want to do in depth studies Common in business studies Quantitative methods Cut reality into discrete pieces, which are then combined into statistical clusters Common in economic studies

NO METHOD IS ENTIRELY QUALITATIVE OR QUANTITATIVE, BUT TECHNIQUES CAN BE EITHER QUANTITATIVE OR QUALITATIVE TECHNIQUES Qualitative ¡Conversation ¡Structured ¡Unstructured and semi structured interviews, etc. Quantitative observation ¡Structured interview ¡Structured surveys ¡Attitude scaling ¡Field equipment Historical review; Group discussion; Case study; Survey; Experiment

(1) HISTORICAL REVIEW Question Techniques Problem Requirement What Go through Trust to happened existing human in the memory records and past? reports; Review the archives; Talk to different people To be critical and compare different explanations for the situation or event

(2) FOCUS GROUPS Question Techniques Problem Research question Discussion on a certain topic with several respondents at the same time Discussion is influenced by: the size of the group, personalities of people involved, the physical and geographical arrangement of the meeting, the “chemistry” between the interviewer and the group Requirement Skilful coordination of the group’s interactions

(3) CASE STUDIES Question Research question Techniques In depth interviews Purpose The area of research is relatively less known; For theory building type of research Spread up Most frequently used for thesis and dissertation research in business studies

(4) OBSERVATIONS Question Research question Techniques Advantage Listening and watching other people’s behaviour in a way that allows some type of learning and analytical interpretation Collect first hand information in a natural setting Disadvantage Difficult to translate the events or happenings into scientifically useful information

COLLECTING PRIMARY DATA THROUGH OBSERVATIONS HUMAN NON PARTICIPANT (MECHANICAL) Human field observation Human laboratory observation Mechanical field (laboratory) observation ADVANTAGE In field observation the observer is a natural part of the situation or event People who are being observed know that they are being observed and by whom Danger: The observers can be influenced by the event, situation or culture and everyday lives of the subjects Reactions are observed in a controlled setting in a laboratory or in other virtual reality Researcher observes a natural setting but is not part of the situation (Using video camera in supermarket) Record the hotline statistics (questions asked; problems customers have, etc. ) Data collected are more objective and accurate Issue: Behaviour of people is influenced because of a non participatory observer, but only Issue: Ethical aspects of in the beginning; people get used to it this method of data in a very short time collection Respondents are often careful in replying to sensitive or embarrassing questions

INTERVIEW AND QUESTIONNAIRE TECHNIQUES Question Interviews Asking those who have Personal experienced a particular phenomeno Telephone n so that they can explain Mail it Questionnaires Structured questionnaires – the questions and the answers to be given are predetermined (multiple choice) Unstructured questionnaires – the questions are only roughly predetermined and there are no predetermined answers Semi structured questionnaires – the questions are predetermined, but respondents can use their own words and ways to answer Differences The most obvious difference between a questionnaire and an interview is the cost Interviewing is a much more flexible method than the questionnaire Interviews are considered more appropriate for qualitative studies, while questionnaires are con sidered more suitable for quantitative types of research

(5) SURVEYS ¡ Surveys refer to a method of data collection that utilizes questionnaires or interview techniques. The survey is an effective tool to get opinions, attitudes and descriptions as well as for getting cause and effect relationships ¡ Surveys deals with reconstruction of processes that occurred prior to the investigations

PLANNING A SURVEY Conceptualize and structure the research problem 1. Consider the aims of the research 2. Review the current state of knowledge 3. Assess the various resources available Analytic survey? Test a theory by identifying the independent, dependent and extraneous variables, and their relations, and Controlling variables through statistical techniques such as multiple regression Descriptive survey? Identify the phenomena whose variance you wish to describe The focus is more on a representative sample Establish a priori assumptions/ hypo theses Determine the sampling strategy by defining the research population and designing a means of accessing a representative (random) sample Are data to be collected through one approach? Or does the research problem require the repeated contact of a single sample or several equivalent samples? Interviewer administered questionnaire/schedule More expensive Risk of interviewer bias Respondent completed/ postal administered questionnaire Less expensive High rates of “non response”

CONSTRUCTING QUESTIONNAIRES ¡ What type of information is required? ¡ How it is to be administered through mail, personal interview, telephone interview or a combination? ¡ Individual questions: Is it necessary to ask a certain question? What are the benefits of dummy variables? Is it necessary to have several questions on one issue? Can questions be interpreted differently? Would respondents be willing to give answers to the questions? ¡ Open ended or close ended questions? ¡ Should we have “Don’t know” alternative, providing an escape route? The responses received for questionnaires with or without an escape route differ by up to 20 25% ¡ Length of the questionnaire no standards available ¡ The precise wording of questions is crucial (example) ¡ What type of scale we should use?

GUIDELINES FOR CONSTRUCTING QUESTIONNAIRES ¡ ¡ ¡ The questions must be asked in a very simple and concise language The alternative answers (close ended questions) should use clear and unambiguous language Checking and ensuring that everybody understands the question in the same manner Each question should deal with only one dimension or aspect We should not offer an alternative such as “Don't know” or “No comment” The questions should not be of a suggestive nature Questions should be formulated in a polite and soft language (by answering questions, the respondent is doing us a favour) Questions should be placed in a “right” order (easy to answer questions and positive types of questions should be placed first) There should also be a logical order from general to specific questions The layout of the questionnaire is also important Pre testing the questionnaire on several real companies or respond ents

INTERVIEWS (BY MAIL, BY PHONE, PERSONAL ¡ ¡ ¡ ¡ Interviews demand real interaction between the researcher and the respondent and that is why the researcher needs to know the respondent Interviews are often considered the best data collection methods There are two types of interviews structured and unstructured interviews Semi structured and unstructured interviews demand greater skills from the interviewer Unstructured interviews are considered advantageous in the context of discovery. Interviews also are difficult to interpret and analyse Coding of in depth inter views is a difficult task

PREPARING FOR AN INTERVIEW (1) Analyse the research problem (2) Understand what information you need to have from an interviewee (3) See that who would be able to provide you with that informa tion. (4) Draft an interview guide or interview questions (5) Pre test the first draft of the interview questions as a pilot study (6) Decide how much time the interview should take (no more than 1. 30 hours) (7) Create a situation where the respondent willingly offers time (8) How you are going to record the information (type recording? ) (9) Ask if the interview is to be treated confidentially (10) Create a reason or a reward for the respondent (why should they answer your questions? ) (11) Send a confirmation letter about the appointment (12) Consider all the costs (travelling costs, hotel, etc. ) (13) Plan your time if you have more than one interview per day

THE INTERVIEW ¡ Introduce the study and its purpose to orient the respondents ¡ Use simple and understandable language ¡ Leave it entirely to the informant to provide answers to questions ¡ Show interest and enthusiasm in the respondents and their “story” ¡ Control the situation and the time with care so as to get the relevant information ¡ Develop a relationship with the interviewee ¡ Be careful about sensitive questions ¡ Ensure perfect functioning of the equipment at the time of the interview

POST-INTERVIEW ¡ Write down the important points from the interview as well as notes on the practical details ¡ Write a “Thank you” letter to the respondent ¡ Write down all the information on the tape in the same order ¡ Later develop a descriptive report of the interview relevant for the study ¡ Sometimes it is useful to send this descriptive report to the interviewee for comments

FOCUS GROUPS ¡ Focus group a small group of people (around 10 people) interacting with each other to seek information on a small (focused) number of issues, and the discussion may last from half an hour to around two hours ¡ There should be some homogeneity among the individuals in one focus group, which will encourage more in depth and open discussion ¡ The observer can observe the group, sometimes without disturbing the discussion ¡ The moderator plays an important role in keeping the discussion on the focus issue and also in ensuring that it goes smoothly

ADVANTAGES AND DISADVANTAGES OF THE FOCUS GROUPS Advantages Very rich and in depth data expressed in respondents’ own words and reactions It is a quick, flexible and inexpensive method Allows the researcher to interact directly with respond ents Allow the collection of data from people who are not literate or from children Disadvantages This type of data collection makes it very difficult to summarize and categorize the information It can be difficult to gather people at a location The small numbers who are willing might not be representative of our population The responses of the group members are not independent of one another (mutually influenced)

USEFULNESS OF THE FOCUS GROUPS IN BUSINESS STUDIES ¡ Obtaining general background about a topic ¡ Generating research hypotheses ¡ Stimulating new ideas and creative concepts ¡ ¡ ¡ Diagnosing problems/success factors for a new product, service or program Generating impressions of products, programmes, services or institutions/firm Learning how respondents talk about the phenomenon, which may help designing questionnaires or other instruments Interpreting previously obtained quantitative data The representativeness is most important as we observe only a few individuals

Conclusions (from the case-study) By economic sectors ¡ ¡ ¡ In the retail and tourist sector there almost no elements of the French Social Model (FSM) These sectors are of great labour intensity, and the recent high unemployment in the country gave an advantage to employers not to invest in human capital Surprisingly there are no traces of the FSM in the bank sector, where only the training of newly employed is better developed Probably the foreign banks’ filial use already proved products (and do not develop new ones) and do not see a great advantage to invest in local employees Only in sectors of energy and in the manufacturing as a whole there are some elements of labour politics of the parent companies We assume that the specificity of the manufacturing and key sectors like energy require stronger the application of similar social politics by the multinational filial

Conclusions (from the case-study) By the way of acquisition In the franchises there almost no elements of the FSM. Obviously the local companies accept only the standards of work/service from the parent company, but not the standards of the social policy ¡ In the proper filial some elements of the FSM are present in combination with the national model of the receiving country. As a whole the filial are better transmitters of the social policy of the parent company ¡

General Conclusions (from the case-study) ¡ ¡ ¡ Commonly all investigated companies have almost no trade unions, in single cases there are representatives of workers and employees, but they all have no serious irregularities in the labour relations from the point of view of the national Labour Code The compensations remain in the field of the management and directorates, and it is not a matter of the collective bargaining The penetration of the MNC does not lead automatically to the transfer of their social policies The MNC respect national legal requirements on the labour relations, but this represent minimal effort for them Based on that we can conclude that the Bulgarian national legislation is not enough exigent in that respect Obviously there are serious challenges to the European social model (based on the FSM) by some succeeding and not so social countries like the US. It means that the EU countries have to look for a balance between higher competitiveness and better social policy on both national and firm levels

NECESSITY OF SAMPLING ¡ ¡ Collect information from each member of the population Collect information from a portion of the population Population here refers not only to people, but also to firms, products and so on ¡ A sample frame is a listing of units from which the actual sample will be drawn ¡ Taking a sample of elements from the larger group, we can infer something about the larger group ¡ Two reasons for taking a sample: The costs of including all units, The time needed to do so The US Bureau of Census uses sample surveys to check the accuracy of the various censuses

TYPES OF SAMPLES Probability sample Each unit has a known, non zero chance of being included in the sample, which allows for statistical inferences Representative sample what has been found in the sample is valid (within certain limits) for the population Important if we are to estimate unknown parameters or draw valid inferences regarding the population Non probability sample It is not possible to make valid inferences about the population, which implies that such samples are not representative Accidental sample units that we find convenient for some reason are selected Judgment sample select units we think are representative of the population Quota sample certain subgroups of units are represented in the sample in approximately the same proportions as they are represented in the popula tion Easy to draw, but they may give misleading results (no basis for evaluating the size of the sampling variation and the error of estimation)

(1) SIMPLE RANDOM SAMPLING ¡ ¡ ¡ All units in the population have the same chance (probability) of being included What variables or parameters are of interest? Parameters describe aspects of variables Variable can be denned as a set of values related to a population in such a way that each unit has one and only one value from the set Value can be denned as a piece of information regarding a particular aspect of a unit In the case of a total listing of all units, the sample can be drawn as in a lottery (Prepared tables of random digits exist as well)

(2) SIMPLE RANDOM SAMPLING Typical parameters to be estimated in a sampling survey are: - population total, population means, population proportions, population variances and population ratios. When more than one variable is involved, additional parameters of interest might be: population correlation coefficients and population regression coefficients

(3) SIMPLE RANDOM SAMPLING Drawbacks: ¡ ¡ ¡ A complete frame (a list of all units in the whole population) is needed Costs of obtaining the sample can be high if the units are geographically widely scattered The standard errors of estimators can be high This is a major reason for applying other sampling procedures, i. e. to reduce standard errors of estimators by the sample size If the units have quite different values for a variable of interest, simple random sampling can be improved by making the probability of inclusion in the sample proportional to the value of the variable. This is called sampling with probabilities proportional to size (Example)

SYSTEMATIC SAMPLING ¡ ¡ PREREQUISITE: units in the population can be ordered in some way. It allows that the units in the population can be numbered from the first (1) to the last unit (N) Example: A 10% systematic sample is obtained by drawing every tenth unit in the ordered population. For instance: select every 10 th unit after a random start (7, 17, 27, etc. ) Advantages: the method is simple, and a frame is not always needed Drawbacks: danger of hidden periodicities, e. g. that a deficiency in producing a specific product occurs at specific intervals (If one happens to get an unfortunate starting point, the whole sample could consist of defective products)

STRATIFIED SAMPLING ¡ ¡ The parent population is divided into a mutually exclusive and exhaustive subset; A simple random sample of units is chosen independently from each subset An important reason for stratified sampling is that variability, and thus standard error of estimates may be reduced ¡ Important if the means (proportions, etc. ) are very different in the different strata. The result will be a smaller sampling variation ¡ Proportional allocation means that the proportion of units included in the sample is the same for each stratum ¡ Stratifying in a fashion that makes the means (or other parameters) rather different in different strata

STRATIFIED SAMPLING Advantages: ¡ Can give higher precision with the sample size or, alternatively, the same precision with a smaller sample ¡ Can give separate results for each stratum! ¡ Simplifies data collection. Drawbacks: ¡ A complete frame is needed. ¡ Additional information, such as knowledge of standard deviations and costs, may be needed for each stratum. ¡ If the population can be divided into strata which are homogeneous within but heterogeneous between, precision can be increased or costs lowered

CLUSTER SAMPLING ¡ ¡ The population is divided into mutually exhaustive subsets Random samples of the subsets are selected One stage cluster sampling all units in the selected clusters are examined Two stage cluster sampling a sample of units is selected probabilistically from the subsets With stratified sampling, a sample of units is selected from each subgroup With cluster sampling, a sample of subgroups is selected ¡ It is desirable for each subgroup to be a small scale model of the actual population ¡ The subgroup should be formed to be as heterogeneous as possible

(1) DETERMINING SAMPLE SIZE ¡ ¡ ¡ “What is the sample size needed? ” It depends on the desired precision from the estimate Precision is the size of the estimating interval (you want the sample estimate to be within ± £ 100 of the true population mean. This is more precise than to be within ± £ 200 of the true value) The concept of standard error (of the mean) is central to determining the size of a sample. The formula for the standard error (SE) is: (1) SE = SD √n where SD = standard deviation (of mean) and n = sample size ¡ To know SE we first must know or estimate the standard deviation. ¡ The degree of confidence associated with the estimate also needs to be taken into account.

(2) DETERMINING SAMPLE SIZE Assume that a researcher wants the estimate to be within ± £ 25 of the true population value, and to be 95% confident that the interval (25¯x+ 25) will contain the true population mean. This implies constructing an interval ± z. SD around the observed mean, in which z is approximately 2. This can be expressed as; H = Z. SE = Z x SD √n where H is half of the interval, i. e. 25. ¡ Also assume that early studies have demonstrated the standard deviation to be around 100, Thus: 25 = 2 x 100 √n √n = 2 x 100 25 2 n = 2 x 1002 = 64 252 ¡ Note what happens if the estimate must be twice as precise, i. e. a desired interval x¯ ± 12. 5 n = 22 x 1002 = 256 12. 52 ¡ ¡ There is a trade off between degree of confidence and degree of precision with a sample of fixed size. Thus, doubling the precision interval increased the required sample size by a factor of four. If the standard deviation, SD, is not known it must be estimated.

(3) DETERMINING SAMPLE SIZE Often the population proportion, π, is another parameter of interest, e. g. percentage of voters, percentage with a specific interest and so on. The distribution of sample proportions is centered about the population proportions. The standard error of a proportion SDp is equal to: SDp = √ π (1 π)/n To estimate the required sample size we need to decide on the precision and confidence wanted: H = z SDp (where H is half of the interval, i. e. 25) H = z. √π (1 π)/n Let us assume that a political party wants to conduct a poll to estimate the % voting for the party within ± 2 percentage points and that the party wishes to be 95% confident of the result. Also assume that the percentage voting for the party is believed to be 40%. To estimate the required sample size, an estimate of the proportion is also needed. To estimate the sample size, we apply formula: n = 22 π (1 π) H 2 n = 22. (0. 40) (1 0. 40) = 2400 0. 022

Typical sample size Number of subgroup analyses People or households National Regional or special None or few 1000 1500 200 500 Institutions National Regional or special 200 50 200 Average 1500 2500 1000 200 500 Many 2500+ 1000+ 500 +

Non-response ¡ A serious potential threat to the validity of results from sampling surveys is non response, which reduced the effective sample size. ¡ But this is not the main problem, since it can easily be remedied. Thus, if we need a sample of 400 units and we expect a 50% response rate, we could take a sample of 800 units to counteract the non response. ¡ The real problem with non response is that those who do not respond are usually different from those who do respond (Example the majority of the real drinkers will probably not respond for several reasons, but they make up an important part of the whole picture)

Sampling in qualitative research ¡ ¡ ¡ Samples always applied in qualitative research. Statistical conclusion validity plays a major role in quantitative research. In qualitative research the purpose is to understand, gain insights and create explanations (theory).

(Example) Understanding purchase decisions ¡ Who possesses (and is willing to share) the needed informa tion, which impliesselecting the most relevant respondents (subjects). ¡ We may start with one person, e. g. the manager of the research department, and by asking about specific purchases, also asking: “Were other persons involved? ” and thus gradually uncovering participation and influences in buying decisions.

(Example) opinions ¡ How many focus group interviews should be conducted? Let us assume that the researcher starts with one focus group, and the data is transcribed analyzed. Then s/he conducts another focus group interview, and that also uncovers points of view not present in the first one. ¡ The researcher continues the procedure until no new opinions/points of view are uncovered. This way of reasoning corresponds to sequential sampling, i. e. one continues to add observations until a (final) conclusion is arrived.

Theoretical sampling ¡ ¡ ¡ Consider a study designed to examine the (potential) relationship between organizational form and innovativeness. In order to study this research question, variability of organizational forms of the firms (organizations) included is needed. Let us also assume that the researcher, based on review of the literature, knows that forms, F 1, F 2…exist. This insight is then useful when deciding on which firms (organizations) should be included (the sample units are chosen for theoretical reasons).