In primary data collection, you collect the data yourself using methods such as interviews and questionnaires.

The key point here is that the data you collect is unique to you and your research and, until you publish, no one else has access to it. There are many methods of collecting primary data and the main methods include:

questionnaires interviews focus group interviews observation case-studies diaries critical incidents portfolios.

Click on one of the above icons for the information in LRN format. The primary data, which is generated by the above methods, may be qualitative in nature (usually in the form of words) or quantitative (usually in the form of numbers or where you can make counts of words used). We briefly outline these methods but you should also read around the various methods. A list of suggested research methodology texts is given in your Module Study Guide but many texts on social or educational research may also be useful and you can find them in your library. Questionnaires Questionnaires are a popular means of collecting data, but are

difficult to design and often require many rewrites before an acceptable questionnaire is produced. Advantages:

Can be used as a method in its own right or as a basis for interviewing or a telephone survey. Can be posted, e-mailed or faxed. Can cover a large number of people or organisations. Wide geographic coverage. Relatively cheap. No prior arrangements are needed. Avoids embarrassment on the part of the respondent. Respondent can consider responses. Possible anonymity of respondent. No interviewer bias.

Disadvantages:

Design problems. Questions have to be relatively simple. Historically low response rate (although inducements may help). Time delay whilst waiting for responses to be returned. Require a return deadline. Several reminders may be required. Assumes no literacy problems. No control over who completes it. Not possible to give assistance if required.

Problems with incomplete questionnaires. Replies not spontaneous and independent of each other. Respondent can read all questions beforehand and then decide whether to complete or not. For example, perhaps because it is too long, too complex, uninteresting, or too personal.

Design of postal questionnaires Theme and covering letter The general theme of the questionnaire should be made explicit in a covering letter. You should state who you are; why the data is required; give, if necessary, an assurance of confidentiality and/or anonymity; and contact number and address or telephone number. This ensures that the respondents know what they are committing themselves to, and also that they understand the context of their replies. If possible, you should offer an estimate of the completion time. Instructions for return should be included with the return date made obvious. For example: It would be appreciated if you could return the completed questionnaire by... if at all possible. Instructions for completion You need to provide clear and unambiguous instructions for completion. Within most questionnaires these are general instructions and specific instructions for particular question structures. It is usually best to separate these, supplying the general instructions as a preamble to the questionnaire, but leaving the specific instructions until the questions to which they apply. The response method should be indicated (circle, tick, cross, etc.). Wherever possible, and certainly if a slightly unfamiliar response system is employed, you should give an example. Appearance Appearance is usually the first feature of the questionnaire to which the recipient reacts. A neat and professional look will encourage further consideration of your request, increasing your response rate. In addition, careful thought to layout should help your analysis. There are a number of simple rules to help improve questionnaire appearance:

Liberal spacing makes the reading easier. Photo-reduction can produce more space without reducing content. Consistent positioning of response boxes, usually to the right, speeds up completion and also avoids inadvertent omission of responses. Choose the font style to maximise legibility. Differentiate between instructions and questions. Either lower case and capitals can be used, or responses can be boxed.

Length There may be a strong temptation to include any vaguely interesting questions, but you should resist this at all costs. Excessive size can only reduce response rates. If a long questionnaire is necessary, then you must give even more thought to appearance. It is best to leave pages unnumbered; for respondents to flick to the end and see page 27 can be very disconcerting! Order Probably the most crucial stage in questionnaire response is the beginning. Once the respondents have started to complete the questions they will normally finish the task, unless it is very long or difficult. Consequently, you need to select the opening questions with care. Usually the best approach is to ask for biographical details first, as the respondents should know all the answers without much thought. Another benefit is that an easy start provides practice in answering questions. Once the introduction has been achieved the subsequent order will depend on many considerations. You should be aware of the varying importance of different questions. Essential information should appear early, just in case the questionnaire is not completed. For the same reasons, relatively unimportant questions can be placed towards the end. If questions are likely to provoke the respondent and remain unanswered, these too are best left until the end, in the hope of obtaining answers to everything else.

Coding If analysis of the results is to be carried out using a statistical package or spreadsheet it is advisable to code non-numerical responses when designing the questionnaire, rather than trying to code the responses when they are returned. An example of coding is: Male [ ] 1 Female [ ] 2

The coded responses (1 or 2) are then used for the analysis. Thank you Respondents to questionnaires rarely benefit personally from their efforts and the least the researcher can do is to thank them. Even though the covering letter will express appreciation for the help given, it is also a nice gesture to finish the questionnaire with a further thank you. Questions

Keep the questions short, simple and to the point; avoid all unnecessary words. Use words and phrases that are unambiguous and familiar to the respondent. For example, dinner has a number of different interpretations; use an alternative expression such as evening meal. Only ask questions that the respondent can answer. Hypothetical questions should be avoided. Avoid calculations and questions that require a lot of memory work, for example, How many people stayed in your hotel last year? Avoid loaded or leading questions that imply a certain answer. For example, by mentioning one particular item in the question, Do you agree that Colgate toothpaste is the best toothpaste? Vacuous words or phrases should be avoided. Generally, usually, or normally are imprecise terms with various meanings. They should be replaced

with quantitative statements, for example, at least once a week.

Questions should only address a single issue. For example, questions like: Do you take annual holidays to Spain? should be broken down into two discreet stages, firstly find out if the respondent takes an annual holiday, and then secondly find out if they go to Spain. Do not ask two questions in one by using and. For example, Did you watch television last night and read a newspaper? Avoid double negatives. For example, Is it not true that you did not read a newspaper yesterday? Respondents may tackle a double negative by switching both negatives and then assuming that the same answer applies. This is not necessarily valid. State units required but do not aim for too high a degree of accuracy. For instance, use an interval rather than an exact figure:

How much did you earn last year? Less than 10,000 [ ] 10,000 but less than 20,000 [ ] Avoid emotive or embarrassing words usually connected with race, religion, politics, sex, money. Types of questions Closed questions A question is asked and then a number of possible answers are provided for the respondent. The respondent selects the answer which is appropriate. Closed questions are particularly useful in obtaining factual information: Sex: Male [ ] Female [ ] Did you watch television last night? Yes [ ] No [ ]

Some Yes/No questions have a third category Do not know. Experience shows that as long as this alternative is not

mentioned people will make a choice. Also the phrase Do not know is ambiguous: Do you agree with the introduction of the EMU? Yes [ ] No [ ] Do not know [ ] What was your main way of travelling to the hotel? Tick one box only. Car Coach Motor bike Train Other means, please specify With such lists you should always include an other category, because not all possible responses might have been included in the list of answers. Sometimes the respondent can select more than one from the list. However, this makes analysis difficult: Why have you visited the historic house? Tick the relevant answer(s). You may tick as many as you like. I enjoy visiting historic houses [ ] [ ] [ ] [ ] [ ]

The weather was bad and [ ] I could not enjoy outdoor activities I have visited the house before and wished to return Other reason, please specify Attitude questions Frequently questions are asked to find out the respondents opinions or attitudes to a given situation. A Likert scale provides a battery of attitude statements. The respondent then [ ]

says how much they agree or disagree with each one: Read the following statements and then indicate by a tick whether you strongly agree, agree, disagree or strongly disagree with the statement. Strongly Agree Disagree agree My visit has been good value for money There are many variations on this type of question. One variation is to have a middle statement, for example, Neither agree nor disagree. However, many respondents take this as the easy option. Only having four statements, as above, forces the respondent into making a positive or negative choice. Another variation is to rank the various attitude statements, however, this can cause analysis problems: Which of these characteristics do you like about your job? Indicate the best three in order, with the best being number 1. Varied work Good salary Opportunities for promotion Good working conditions High amount of responsibility Friendly colleagues [ ] [ ] [ ] [ ] [ ] [ ] Strongly disagree

A semantic differential scale attempts to see how strongly an attitude is held by the respondent. With these scales doubleended terms are given to the respondents who are asked to indicate where their attitude lies on the scale between the terms. The response can be indicated by putting a cross in a particular position or circling a number: Work is: (circle the appropriate number) Difficult Useless 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Easy Useful

Interesting

1 2 3 4 5 6 7

Boring

For summary and analysis purposes, a score of 1 to 7 may be allocated to the seven points of the scale, thus quantifying the various degrees of opinion expressed. This procedure has some disadvantages. It is implicitly assumed that two people with the same strength of feeling will mark the same point on the scale. This almost certainly will not be the case. When faced with a semantic differential scale, some people will never, as a matter of principle, use the two end indicators of 1 and 7. Effectively, therefore, they are using a five-point scale. Also scoring the scale 1 to 7 assumes that they represent equidistant points on the continuous spectrum of opinion. This again is probably not true. Nevertheless, within its limitations, the semantic differential can provide a useful way of measuring and summarising subjective opinions. Other types of questions to determine peoples opinions or attitudes are: Which one/two words best describes...? Which of the following statements best describes...? How much do you agree with the following statement...? Open questions An open question such as What are the essential skills a manager should possess? should be used as an adjunct to the main theme of the questionnaire and could allow the respondent to elaborate upon an earlier more specific question. Open questions inserted at the end of major sections, or at the end of the questionnaire, can act as safety valves, and possibly offer additional information. However, they should not be used to introduce a section since there is a high risk of influencing later responses. The main problem of open questions is that many different answers have to be summarised and possibly coded. Testing pilot survey Questionnaire design is fraught with difficulties and problems. A number of rewrites will be necessary, together with refinement and rethinks on a regular basis. Do not assume that you will write the questionnaire accurately and perfectly at the

first attempt. If poorly designed, you will collect inappropriate or inaccurate data and good analysis cannot then rectify the situation. To refine the questionnaire, you need to conduct a pilot survey. This is a small-scale trial prior to the main survey that tests all your question planning. Amendments to questions can be made. After making some amendments, the new version would be re-tested. If this re-test produces more changes, another pilot would be undertaken and so on. For example, perhaps responses to open-ended questions become closed; questions which are all answered the same way can be omitted; difficult words replaced, etc. It is usual to pilot the questionnaires personally so that the respondent can be observed and questioned if necessary. By timing each question, you can identify any questions that appear too difficult, and you can also obtain a reliable estimate of the anticipated completion time for inclusion in the covering letter. The result can also be use to test the coding and analytical procedures to be performed later. Distribution and return The questionnaire should be checked for completeness to ensure that all pages are present and that none is blank or illegible. It is usual to supply a prepaid addressed envelope for the return of the questionnaire. You need to explain this in the covering letter and reinforce it at the end of the questionnaire, after the Thank you. Finally, many organisations are approached continually for information. Many, as a matter of course, will not respond in a positive way.

Interviews Interviewing is a technique that is primarily used to gain an understanding of the underlying reasons and motivations for peoples attitudes, preferences or behaviour. Interviews can be undertaken on a personal one-to-one basis or in a group. They can be conducted at work, at home, in the street or in a

shopping centre, or some other agreed location. Personal interview Advantages:

Serious approach by respondent resulting in accurate information. Good response rate. Completed and immediate. Possible in-depth questions. Interviewer in control and can give help if there is a problem. Can investigate motives and feelings. Can use recording equipment. Characteristics of respondent assessed tone of voice, facial expression, hesitation, etc. Can use props. If one interviewer used, uniformity of approach. Used to pilot other methods.

Disadvantages:

Need to set up interviews. Time consuming. Geographic limitations. Can be expensive. Normally need a set of questions. Respondent bias tendency to please or impress, create false personal image, or end interview quickly. Embarrassment possible if personal questions.

Transcription and analysis can present problems subjectivity. If many interviewers, training required.

Types of interview Structured:

Based on a carefully worded interview schedule. Frequently require short answers with the answers being ticked off. Useful when there are a lot of questions which are not particularly contentious or thought provoking. Respondent may become irritated by having to give over-simplified answers.

Semi-structured The interview is focused by asking certain questions but with scope for the respondent to express him or herself at length. Unstructured This also called an in-depth interview. The interviewer begins by asking a general question. The interviewer then encourages the respondent to talk freely. The interviewer uses an unstructured format, the subsequent direction of the interview being determined by the respondents initial reply. The interviewer then probes for elaboration Why do you say that? or, Thats interesting, tell me more or, Would you like to add anything else? being typical probes. The following section is a step-by-step guide to conducting an interview. You should remember that all situations are different and therefore you may need refinements to the approach. Planning an interview:

List the areas in which you require information. Decide on type of interview.

Transform areas into actual questions. Try them out on a friend or relative. Make an appointment with respondent(s) discussing details of why and how long. Try and fix a venue and time when you will not be disturbed.

Conducting an interview: Personally arrive on time be smart smile employ good manners find a balance between friendliness and objectivity. introduce yourself re-confirm the purpose assure confidentiality if relevant specify what will happen to the data. speak slowly in a soft, yet audible tone of voice control your body language know the questions and topic ask all the questions. recorded as you go on questionnaire written verbatim, but slow and timeconsuming summarised by you taped agree beforehand have alternative method if not acceptable consider effect on respondents answers proper equipment in good working order sufficient tapes and batteries minimum of background noise. ask if the respondent would like to give further details about anything or any questions about the research thank them.

At the start

The questions

Responses

At the end

Telephone interview This is an alternative form of interview to the personal, face-toface interview. Advantages:

Relatively cheap. Quick. Can cover reasonably large numbers of people or organisations. Wide geographic coverage. High response rate keep going till the required number. No waiting. Spontaneous response. Help can be given to the respondent. Can tape answers.

Disadvantages:

Often connected with selling. Questionnaire required. Not everyone has a telephone. Repeat calls are inevitable average 2.5 calls to get someone. Time is wasted. Straightforward questions are required. Respondent has little time to think. Cannot use visual aids. Can cause irritation. Good telephone manner is required. Question of authority.

Getting started

Locate the respondent:

Repeat calls may be necessary especially if you are trying to contact people in organisations where you may have to go through secretaries. You may not know an individuals name or title so there is the possibility of interviewing the wrong person. You can send an advance letter informing the respondent that you will be telephoning. This can explain the purpose of the research.

Getting them to agree to take part:

You need to state concisely the purpose of the call scripted and similar to the introductory letter of a postal questionnaire. Respondents will normally listen to this introduction before they decide to co-operate or refuse. When contact is made respondents may have questions or raise objections about why they could not participate. You should be prepared for these.

Ensuring quality

Quality of questionnaire follows the principles of questionnaire design. However, it must be easy to move through as you cannot have long silences on the telephone. Ability of interviewer follows the principles of faceto-face interviewing.

Smooth implementation

Interview schedule each interview schedule should have a cover page with number, name and address. The cover sheet should make provision to record which call

it is, the date and time, the interviewer, the outcome of the call and space to note down specific times at which a call-back has been arranged. Space should be provided to record the final outcome of the call was an interview refused, contact never made, number disconnected, etc.

Procedure for call-backs a system for call-backs needs to be implemented. Interview schedules should be sorted according to their status: weekday call-back, evening call-back, weekend call-back, specific time call-back.

Comparison of postal, telephone and personal interview surveys The table below compares the three common methods of postal, telephone and interview surveys it might help you to decide which one to use. Postal survey Cost (assuming Often lowest a good response rate) Ability to probe No personal contact or observation Telephone survey Usually inbetween Some chance for gathering additional data through elaboration on questions, but no personal observation Perhaps, but usually no Personal interview Usually highest Greatest opportunity for observation, building rapport, and additional probing Perhaps, if interview time is prearranged with respondent

Respondent ability to complete at own convenience

Yes

Interview bias No chance

Some, perhaps Greatest due to voice chance inflection Some Greatest

Ability to

Least

decide who actually responds to the questions Impersonality Greatest Some due to lack of faceto-face contact Least

Complex questions Visual aids Potential negative respondent reaction Interviewer control over interview environment Time lag between soliciting and receiving response Suitable types of questions

Least suitable Somewhat suitable Little opportunity Junk mail No opportunity Junk calls

More suitable Greatest opportunity Invasion of privacy

Least

Some in selection of time to call Least

Greatest

Greatest

May be considerable if a large area involved Greatest opportunity for openended questions

Simple, mostly dichotomous (yes/no) and multiple choice

Some opportunity for openended questions especially if interview is recorded Medium

Requirement for technical skills in conducting interview Response rate

Least

Greatest

Low

Usually high

High

Table 3.1: Comparison of the three common methods of

surveys Focus group interviews A focus group is an interview conducted by a trained moderator in a non-structured and natural manner with a small group of respondents. The moderator leads the discussion. The main purpose of focus groups is to gain insights by listening to a group of people from the appropriate target market talk about specific issues of interest. Observation Observation involves recording the behavioural patterns of people, objects and events in a systematic manner. Observational methods may be:

structured or unstructured disguised or undisguised natural or contrived personal mechanical non-participant participant, with the participant taking a number of different roles.

Structured or unstructured In structured observation, the researcher specifies in detail what is to be observed and how the measurements are to be recorded. It is appropriate when the problem is clearly defined and the information needed is specified. In unstructured observation, the researcher monitors all aspects of the phenomenon that seem relevant. It is appropriate when the problem has yet to be formulated precisely and flexibility is needed in observation to identify key components of the problem and to develop hypotheses. The potential for bias is high. Observation findings should be treated as

hypotheses to be tested rather than as conclusive findings. Disguised or undisguised In disguised observation, respondents are unaware they are being observed and thus behave naturally. Disguise is achieved, for example, by hiding, or using hidden equipment or people disguised as shoppers. In undisguised observation, respondents are aware they are being observed. There is a danger of the Hawthorne effect people behave differently when being observed. Natural or contrived Natural observation involves observing behaviour as it takes place in the environment, for example, eating hamburgers in a fast food outlet. In contrived observation, the respondents behaviour is observed in an artificial environment, for example, a food tasting session. Personal In personal observation, a researcher observes actual behaviour as it occurs. The observer may or may not normally attempt to control or manipulate the phenomenon being observed. The observer merely records what takes place. Mechanical Mechanical devices (video, closed circuit television) record what is being observed. These devices may or may not require the respondents direct participation. They are used for continuously recording on-going behaviour. Non-participant The observer does not normally question or communicate with the people being observed. He or she does not participate. Participant In participant observation, the researcher becomes, or is, part of

the group that is being investigated. Participant observation has its roots in ethnographic studies (study of man and races) where researchers would live in tribal villages, attempting to understand the customs and practices of that culture. It has a very extensive literature, particularly in sociology (development, nature and laws of human society) and anthropology (physiological and psychological study of man). Organisations can be viewed as tribes with their own customs and practices. The role of the participant observer is not simple. There are different ways of classifying the role:

Researcher as employee. Researcher as an explicit role. Interrupted involvement. Observation alone.

Researcher as employee The researcher works within the organisation alongside other employees, effectively as one of them. The role of the researcher may or may not be explicit and this will have implications for the extent to which he or she will be able to move around and gather information and perspectives from other sources. This role is appropriate when the researcher needs to become totally immersed and experience the work or situation at first hand. There are a number of dilemmas. Do you tell management and the unions? Friendships may compromise the research. What are the ethics of the process? Can anonymity be maintained? Skill and competence to undertake the work may be required. The research may be over a long period of time. Researcher as an explicit role The researcher is present every day over a period of time, but entry is negotiated in advance with management and preferably with employees as well. The individual is quite clearly in the role of a researcher who can move around, observe, interview and participate in the work as appropriate. This type of role is the most favoured, as it provides many of the insights that the

complete observer would gain, whilst offering much greater flexibility without the ethical problems that deception entails. Interrupted involvement The researcher is present sporadically over a period of time, for example, moving in and out of the organisation to deal with other work or to conduct interviews with, or observations of, different people across a number of different organisations. It rarely involves much participation in the work. Observation alone The observer role is often disliked by employees since it appears to be eavesdropping. The inevitable detachment prevents the degree of trust and friendship forming between the researcher and respondent, which is an important component in other methods. Choice of roles The role adopted depends on the following:

Purpose of the research: Does the research require continued longitudinal involvement (long period of time), or will in-depth interviews, for example, conducted over time give the type of insights required? Cost of the research: To what extent can the researcher afford to be committed for extended periods of time? Are there additional costs such as training? The extent to which access can be gained: Gaining access where the role of the researcher is either explicit or covert can be difficult, and may take time. The extent to which the researcher would be comfortable in the role: If the researcher intends to keep his identity concealed, will he or she also feel able to develop the type of trusting relationships that are important? What are the ethical issues? The amount of time the researcher has at his disposal: Some methods involve a considerable amount of time. If time is a problem alternate approaches will have to be

sought.

Case-studies The term case-study usually refers to a fairly intensive examination of a single unit such as a person, a small group of people, or a single company. Case-studies involve measuring what is there and how it got there. In this sense, it is historical. It can enable the researcher to explore, unravel and understand problems, issues and relationships. It cannot, however, allow the researcher to generalise, that is, to argue that from one casestudy the results, findings or theory developed apply to other similar case-studies. The case looked at may be unique and, therefore not representative of other instances. It is, of course, possible to look at several case-studies to represent certain features of management that we are interested in studying. The case-study approach is often done to make practical improvements. Contributions to general knowledge are incidental. The case-study method has four steps: 1. Determine the present situation. 2. Gather background information about the past and key variables. 3. Test hypotheses. The background information collected will have been analysed for possible hypotheses. In this step, specific evidence about each hypothesis can be gathered. This step aims to eliminate possibilities which conflict with the evidence collected and to gain confidence for the important hypotheses. The culmination of this step might be the development of an experimental design to test out more rigorously the hypotheses developed, or it might be to take action to remedy the problem. 4. Take remedial action. The aim is to check that the hypotheses tested actually work out in practice. Some action, correction or improvement is made and a recheck carried out on the situation to see what effect the change has brought about. The case-study enables rich information to be gathered from

which potentially useful hypotheses can be generated. It can be a time-consuming process. It is also inefficient in researching situations which are already well structured and where the important variables have been identified. They lack utility when attempting to reach rigorous conclusions or determining precise relationships between variables. Diaries A diary is a way of gathering information about the way individuals spend their time on professional activities. They are not about records of engagements or personal journals of thought! Diaries can record either quantitative or qualitative data, and in management research can provide information about work patterns and activities. Advantages:

Useful for collecting information from employees. Different writers compared and contrasted simultaneously. Allows the researcher freedom to move from one organisation to another. Researcher not personally involved. Diaries can be used as a preliminary or basis for intensive interviewing. Used as an alternative to direct observation or where resources are limited.

Disadvantages:

Subjects need to be clear about what they are being asked to do, why and what you plan to do with the data. Diarists need to be of a certain educational level. Some structure is necessary to give the diarist focus, for example, a list of headings. Encouragement and reassurance are needed as completing a diary is time-consuming and can be

irritating after a while.

Progress needs checking from time-to-time. Confidentiality is required as content may be critical. Analyses problems, so you need to consider how responses will be coded before the subjects start filling in diaries.

Critical incidents The critical incident technique is an attempt to identify the more noteworthy aspects of job behaviour and is based on the assumption that jobs are composed of critical and non-critical tasks. For example, a critical task might be defined as one that makes the difference between success and failure in carrying out important parts of the job. The idea is to collect reports about what people do that is particularly effective in contributing to good performance. The incidents are scaled in order of difficulty, frequency and importance to the job as a whole. The technique scores over the use of diaries as it is centred on specific happenings and on what is judged as effective behaviour. However, it is laborious and does not lend itself to objective quantification. Portfolios A measure of a managers ability may be expressed in terms of the number and duration of issues or problems being tackled at any one time. The compilation of problem portfolios is recording information about how each problem arose, methods used to solve it, difficulties encountered, etc. This analysis also raises questions about the persons use of time. What proportion of time is occupied in checking; in handling problems given by others; on self-generated problems; on toppriority problems; on minor issues, etc? The main problem with this method and the use of diaries is getting people to agree to record everything in sufficient detail for you to analyse. It is very time-consuming!

Sampling

Collecting data is time consuming and expensive, even for relatively small amounts of data. Hence, it is highly unlikely that a complete population will be investigated. Because of the time and cost elements the amount of data you collect will be limited and the number of people or organisations you contact will be small in number. You will, therefore, have to take a sample and usually a small sample. Sampling theory says a correctly taken sample of an appropriate size will yield results that can be applied to the population as a whole. There is a lot in this statement but the two fundamental questions to ensure generalisation are: 1. How is a sample taken correctly? 2. How big should the sample be? The answer to the second question is as large as possible given the circumstances. It is like answering the question How long is a piece of string? It all depends on the circumstances. Whilst we do not expect you to normally generalise your results and take a large sample, we do expect that you follow a recognised sampling procedure, such that, if the sample was increased generalisation would be possible. You therefore need to know some of the basics of sampling. This will be done by reference to the following example.

The theory of sampling is based on random samples where all items in the population have the same chance of being selected as sample units. Random samples can be drawn in a number of ways but are usually based on having some information about population members. This information is usually in the form of an alphabetical list called the sampling frame. Three types of random sample can be drawn a simple random sample (SRS), a stratified sample and a systematic sample. Simple random sampling Simple random sampling can be carried out in two ways the lottery method and using random numbers. The lottery method involves:

transferring each persons name from the list and putting it on a piece of paper the pieces of paper are placed in a container and thoroughly mixed the required number are selected by someone without looking the names selected are the simple random sample.

This is basically similar to a game of bingo or the national lottery. This procedure is easy to carry out especially if both population and sample are small, but can be tedious and time consuming for large populations or large samples. Alternatively random numbers can be used. Random numbers are strings of digits that have been generated by the lottery method and can be found in books of statistical tables. An example of these is: 03 47 43 73 86 97 74 24 67 62 16 76 62 27 66 12 56 85 99 26 36 96 47 36 61 42 81 14 57 20 56 50 26 71 07 96 96 68 27 31

55 59 56 35 64

38 04 80 46 22

Random numbers tend to be written in pairs and blocks of 5 by 5 to make reading easy. However, care is needed when reading these tables. The numbers can be read in any direction but they should be read as a singe string of digits i.e. left to right as 0, 3, 4, 7 etc, or top to bottom as 0, 9, 1, 1, 5, 3, 7, etc. It is usual to read left to right. The random number method involves:

Allocating a number to each person on the list (each number must consist of the same number of digits so that the tables can be read consistently). Find a starting point at random in the tables (close your eyes and point). Read off the digits. The names matching the numbers are the sample units.

For the example of selecting nine people at random from 90: a) The sampling frame is the list of 90 people. Number this list 00, 01, 02, , 89. Note that each number has two digits and the numbering starts from 00. Suppose a starting point is found at random from the random number tables and let this number be 16. Then the person that has been numbered 16 is the first sample unit. Let the next two digits be 76, then the person numbered 76 is the second sample unit. This procedure is repeated until the nine people have been identified. d) Any number occurring for second time is ignored as is any two-digit number over 89.

b)

c)

Simple random number sampling is used as the basis for many other sampling methods, but has two disadvantages:

A sampling frame is required. This may not be

available, exist or be incomplete.

The procedure is unbiased but the sample may be biased. For instance, if the 90 people are a mixture of men and women and all men were selected this would be a biased sample.

To overcome this problem a stratified sample can be taken. In this the population structure is reflected in the sample structure, with respect to some criterion. For example, suppose the 90 people consist of 30 men and 60 women. If gender is the criterion for stratification then: 30 of the sample should be men 90 ie. 30 90 x 9 = 3 men

60 of the sample should be women 90 ie. 60 90 x 9 = 6 women

Thus the sample reflects the population structure in terms of gender. The three men and six women would then be selected by simple random sampling e.g., random numbers. The problem with this approach is the criterion for stratification, (e.g., age, sex, job description), is chosen by you it is subjective and may not be the best or more appropriate criterion. Also a more detailed sampling frame is required. Systematic sampling Whilst not truly random this is a method that is used extensively because it is easy to operate and quick, even when the population and the sample are large. For example, for the population 90 and sample of nine:

Split the sampling frame in to nine equal groups. i.e. 1 to 9 10 to 19 etc 80 to 89

Select a number between 1 and 9 using random number tables. Suppose this number is 6. Person numbered 6 is chosen. Then the 16th, 26th, 36th, 46th, 56th, 66th, 76th, and 86th people are the remaining sample units. If no sampling frame is available access to the population is necessary, such as customers of a business such as a leisure centre, restaurant or museum. Systematic sampling can be used by selecting a random number say 25. Then the 25th person to enter is the first sample unit. The 50th person to enter is the second sample unit. This process is carried on until the required sample size is met. This approach usually generates a good cross section of the population. However, you may need a team of people when no sampling frame exists to help with counting, interviewing, etc.

Data recording and analysis When you are at the planning stage of your research design, it is worth thinking about how you are going to record your data and, even more importantly, how you are going to analyse it. It is pointless collecting data in a form that you cannot understand or analyse! In the next two units, we help you with this. Make sure you check these issues out before you carry out your primary data collection.

Secondary data collection

Accuracy of the research method