Research Methodology

4.1 Choice of Research Strategy

A number of well-documented strategies exist for investigating a research issue. Robson (1995) broadly categorised these as Experiment, Case Study and Survey. Their selection depends upon a number of factors - such as the nature of the information required, the characteristics of the population being studied, and their geographical dispersion.

Research using experiments is usually directed at identifying a causal relationship and may be either qualitative or quantitative. Experimental research typically involves manipulating a single variable, making a controlled change in order to observe its effect upon others. Such studies allow a hypothesis to be tested - i.e. that x affects y. An experimental research methodology could not easily be used to collect the identified information.

Case studies are frequently used for exploratory investigations - such as to assess phenomena in a new light, or to seek new insights (Robson 1995 pp 42). Case studies usually involve small populations, but rely on a wide variety of sources of evidence such as documents, observations and interviews etc. to build up a detailed picture about the ‘case’ that is the subject of the ‘study’. Such examinations produce a detailed picture of a small study sample but can offer no information as to whether the case is typical of a larger population.

It would be possible to look in some detail at the approach of a single Local Authority, or at the knowledge of an individual Environmental Health Officer, however, such information could offer no details about a larger population.

Surveys are commonly used for descriptive studies - to portray an accurate profile of persons, events or situations (Robson 1995 pp42).

"Surveys are based on the desire to collect information (usually by questionnaire) about a well defined population" (Czaja & Blair 1996 pp3)

Surveys involve the collection of standardised information from groups of people. There are three main survey approaches. These are Face-to-face, Telephone and Mail (or Postal) surveys. Each have their merits. Czaja and Blair (1996) observe that there is no one "best" method and that their selection depends upon:

1. Administrative or resource implications

2. Questionnaire issues

3. Data quality issues

Using surveying techniques the results of a representative sample population should be able to produce a picture of a larger population. The survey strategy is accordingly the most appropriate to collect information about Environmental Health Officer’s knowledge and actions. Therefore it is now appropriate to look more closely at possible survey strategies.

4.1.1 Face to Face Interviews

Face to face interviews involve two way dialogue between the interviewer and the respondent. This allows feedback which enables the interviewer to more fully understand the respondents comments. Answers given can be probed to ensure that the interviewer fully understands the response and that they can be correctly recorded. New issues can be identified and explored.

Face to Face interviews are only possible where the interviewer has direct access to the interviewee.

Face to face interviews have been found to have the following disadvantages:

Ÿ Lack of anonymity for the interviewee

Ÿ Characteristics of the interviewer may influence the response

Ÿ Differential interview techniques may be a source of bias

Ÿ Interviewer cheating

Ÿ Cost

Source: BUMP Research Methods

4.1.2 Telephone Interviews

These are a commonly used by market research companies for ‘public opinion’ type research. Interviews can be conducted from a central location. Assuming that the target sample is available and willing to undertake the telephone interview at the time of the interviewers choosing, the survey method can be fast and inexpensive compared to travelling costs for personal interviews.

Questions put to interviewees can be structured or semi-structured and issues can be further explored as they can be in a face to face situation.

4.1.3 Postal Surveys

These surveys rely upon the efficiency of the written word rather than the skill of the interviewer. They are geographically flexible, and have a relatively low cost compared to the other methods outlined. They can be used to reach a dispersed sample population rapidly.

The self administered questionnaire can be completed at the respondents convenience and can offer a high degree of anonymity. The absence of the interviewer does, however, mean that the respondent can not ask for clarification of the questions and thus selective perception may occur. This technique offers no immediate opportunity for feedback nor probing of answers.

There is a time delay between the posting of the questionnaires and the receipt of the response.

It is recognised that the initial appearance - its design and format, its ease of completion, and its ease of return will affect the response rate.

A covering letter identifying the reason the opinions are sought, appealing for assistance, assuring confidentiality, and identifying the minimal effort required by the respondent is acknowledged to be a valuable means of increasing response rates.

4.1.4 The Chosen Research Strategy

For the type of information required to satisfy the identified research goal it was decided that the principal survey methodology should be one of Postal Survey directed toward Local Authority Environmental Health Officers. However personal interviews and telephone interviews would also be used where appropriate e.g. for collecting information from Employers, their Safety Professionals, Managers and workers in local retail outlets.

4.2 Design Considerations for the Postal Survey

The postal survey is a familiar research format to Local Authority Environmental Health Departments. Most authorities will receive on average at least two or three requests of this nature each week on a wide variety of subject matters and for reasons ranging from research into consistency of enforcement through to product marketing. Accordingly it is widely accepted, usually seen as non threatening and is usually a successful medium.

4.2.1 Questionnaire Design and Format

The questionnaire design consisted of fifteen questions spread over four A4 Pages. Questions were divided by clear sub headings which gave a clear indication of the question subject matter. Format and layout were chosen to closely follow surveys of a similar subject nature compiled by the Health and Safety Executive, the Health and Safety Commission and the Local Authority Unit.

It was printed, double sided, on two A4 pages, stapled in the top left hand corner. The front page contained the Newcastle-under-Lyme Borough Council logo printed in its blue and green colour scheme. The questions were printed using black ink.

A variety of questions styles were used to make completion varied. The bulk of the questions were closed questions giving a limited ranges of responses and using a tick box selection process for respondents.

Responses were, where possible, kept clear and simple - such as ‘Yes’ / ‘No’ (Single Dichotomy Responses). Where a qualitative or quantitative response was required a four or a maximum of a five point scale was used for example: ‘None’ / ‘Insufficient’ / ‘Sufficient’ / ‘Excessive’ or ‘Every Time’ / ‘Often’ / ‘Rarely’ / ‘Never’. Czaja & Blair suggested that the number of categories should be limited where possible to ease the respondents choice. Rules for question composition suggested by Sudman & Bradburn (1982) were considered. The use of neutral responses such as ‘sometimes’ and ‘no opinion’ considered in some depth by Sudman & Bradburn (1982) and Schuman & Presser (1981) were avoided in order to encourage the respondent to express an opinion.

A single question was included which used the choices ‘good’ and ‘fair’. Such phraseology has been criticised by researchers such as Bradburn and Miles (1979) who felt that such vague terms, which do not have an absolute meaning, rely upon the respondents value judgement. The excessive use of ‘Don’t Know’ responses was avoided.

A small number of free response (open-ended) questions were also used where listing options would lead the respondent - for example examining the criteria an Officer would consider when inspecting a supermarket checkout. Check list type questions were also used inviting the reader to mark multiple answers to a single question from non overlapping categories. The mix of question styles was designed to reduce respondent boredom and fatigue with a change of pace.

Questions were spaced so that the questionnaire did not appear unwieldy. Space was intentionally provided around questions and at the foot of the questionnaire for the respondent to add comments.

Questionnaires were given a survey reference number so that responding and non-responding Local Authorities could be identified and enable any necessary follow ups. Individual respondents were given the option to supply their contact information, and be assured of the confidentiality of their replies.

4.2.2 The Questions Asked.

Questions were written to collect information about the ways in which Authorities became aware of checkout safety issues, the information available to them, their knowledge of issues, their perception of the relative importance of the checkout as a safety concern, the occasions and ways in which they performed inspection of checkouts, the criteria they would look at, their perceptions of worker conditions and their view on the adequacy of existing controls and levels of consistency of enforcement.

Questions were designed in a way that responses could be made clearly and quickly, and that information could be readily collated for subsequent analysis.

Care was taken to ensure that the number of questions were neither over simplified nor excessive in number or complexity. Questions were ordered to provide a logical sequence for the respondent.

4.3 Pilot Survey & Question pre-testing

The questionnaire was first pre-tested by inviting fellow Environmental Health Officers, from my Local Authority to complete draft versions. After participating they were debriefed and their comments incorporated in subsequent redrafts. Final drafts were submitted to the Dissertation Supervisor, Liz Falconer at Salford University and Dr Claire Dickinson at Health Sciences Division of the Health and Safety Executive for comment.

Testing confirmed that completion of the questionnaire took approximately ten minutes - which had been one of its intended design features.

Pre-testing was completed by sending a small number of questionnaires out to Local Authorities and studying the way in which they were completed - i.e. terms which caused confusion. Some respondents were contacted and their opinions on the questionnaire format and content requested. All non respondents were contacted and their reasons for not replying solicited. In most cases it was found that non response was due to Local Authority policy on participation in surveys and not the format/content of the documents received. These investigations suggested that was no evidence of respondent bias.

Upon completion of the pre-testing phase, and after ensuring that guidance offered by Reynolds et al (1992) had been closely followed it was determined that the questionnaires could be supplied to all the selected recipients without further change.

4.4 The Survey Sample

The Chartered Institute of Environmental Health (CIEH) estimates that there are approximately 4,000 Environmental Health Officers working for in the region of 360 local authorities across the country. It is known that approximately 10% of their total time is spent on Health and Safety issues, but some will be specialists, others multidisciplinary and other will have no involvement in this work area.

In contrast the Health and Safety Commission estimate that there are 1,530 ‘full time equivalent’ Health & Safety enforcement staff (HSC 1997). Consequently from these records it is clear that it would not be possible to identify named individuals to form the respondents for the questionnaires.

Therefore it was decided that the survey should be supplied to ‘selected’ Local Authorities, with the request that it be passed to an Officer who would normally be involved in the Health & Safety inspection of Supermarkets. It was assumed that the Officers response would be characteristic of that of their work colleagues, this being suggested by pre-testing. A question was also included in the survey to gain an insight into whether they felt that their views would be representative of their colleagues and test this hypothesis.

The Local Authority survey sample was selected using a Department of Health Listing of Local Authority Environmental Health Departments. Half the Authorities listed were selected. These Authorities are listed in Appendix Two.

Sample population size was reviewed against the comments of Cohen (1988), Fleiss (1981) and Moser & Kalton (1972) and considered appropriate.

4.5 Respondent bias / respondent characteristics

Postal surveys are known to suffer with respondent bias. This is discussed in detail in Groves 1989. Care was taken with the design of the survey to encourage a frank response and not to word questions in a manner that inferred that there was a preferred answer.

A known failing is that the survey offers no information about non respondents. It is not possible to see how their views would differ from those who replied, nor their reason for non-completion. However, from the range of responses obtained by the pilot study and subsequent follow up investigations, it appeared that there were no specific qualities of non respondents which were not expressed amongst those who replied.

4.6 Main Survey

Questionnaires, with a covering letter - explaining the aims and objectives of the survey and the use that the data would be put to were sent by first class post and addressed to the Chief Officer at each of the respective Environmental Health Departments. Full details of how to contact the sender for further information / clarification were given and a pre paid addressed envelope supplied for their ease of return.

A copy of the questionnaire is attached as Appendix Three.

4.7 Follow Up procedures

Attempts were made to contact non-respondents of the main survey using the EHC-net electronic mail system, a medium recommended by the Department of Health and supported by CIEH. A total of two reminders were sent, the first two weeks after the original questionnaire had been supplied, and the second (a final reminder), a further week later. A number of Authorities made contact suggesting that they had not received a questionnaire and requesting a further copy. Duplicate questionnaires and covering letters were re-sent to named individuals. Addresses used for the repost were in all cases confirmed to be the same as those initially used. Repost documentation was marked accordingly and its response rate was 100%.

It was considered that further attempts at increasing returns would have a negligible effect.

Eligible responses were deemed to be those received within six weeks of the original mailing. Responses received after this deadline (two weeks after the stated deadline in the covering letter) were not included in the main study.

4.8 Analysis of Results

Returned questionnaires were reviewed and the responses to closed answer questions coded. Questions left blank or not appropriately completed were coded as invalid responses. Data generated was entered into the SPSS/PC version 5 software from Knowledge Dynamics Corp., using the Data Entry II Module.

Specifics of the information collected and its method of analysis is documented in the following section.