CHAPTER SEVEN RESEARCH DESIGNS FOR SPORT STUDIES RESEARCH DESIGNS The research design is the overall ‘blueprint’ that guides the researcher in the data collection stages. A number of different research designs exist, including: • Experimental. • Cross-sectional. • Time series. • Longitudinal. • Case study. • Ethnography. 1. Experimental Test effect of independent variable on dependent variable. More for science based projects. For example, interested in influence of crowd on anxiety: pretest post-test effect anxiety(1) crowd anxiety(2) effect = [anxiety (2)] – [anxiety (1)] What about any moderating variables? Learning, increased self-efficacy etc. – all may have an effect. So: group ‘a’ − ‘treatment’ (crowd) group ‘b’ − ‘no treatment’ (control) All other variables remain constant (to some extent!) pretest post-test effect A: anxiety(1) crowd anxiety(2) x = (2 - 1) B: anxiety(3) no treatment anxiety(4) y = (4 - 3) Effect of crowd = x – y You MUST have a control group wherever possible. Ethical issues apply to experimental research – depending on treatment. DEMONSTRATING CAUSALITY USING EXPERIMENTAL DESIGNS Three conditions need to be met: 1. Covariation − As the independent variable changes, then so does the dependent variable. 2. Time order − It is important to ensure that the independent variable (i.e. the cause) actually happens before the effect upon the dependent variable. 3. Non-spuriousness − Some relationships may be due to the existence of an additional variable. Thus, all variables need to be accounted for. 2. Cross-sectional Most common design. Take a (cross-sectional) sample of your population. Each subject provides data only once. Can use this design to approximate an experimental design. Q.1 Rate your anxiety at your last performance. Q.2 Was there a crowd present? Crowd present Anxiety 0 3 10 5 15 6 100 10 Again − need to account for moderating variables – especially as no control group. Need to ask questions about moderating variables – can make questionnaire long! 3. Time series Extended measures before and after introduction of independent variable. Does competing against a particular opponent raise anxiety? time O1 O2 O3 Ind O4 O5 O6 O = Measure Ind = treatment (game against team) Allows you to account for game-by-game fluctuations. Question: How successful is Town ‘X’s Sport Development Policy (introduced 2003 for one year)? Why not measure: • participation at beginning of year • participation at end of year? 20 0 2003 2004 Is the policy successful? 20 0 2002 2003 2004 Is the policy still successful? 20 0 2002 2003 2004 2005 Is the policy still successful? Add a control if possible: 20 0 2002 2003 2004 2005 Is the policy still successful? You need to consider the • time • resources required to do such a study. 4. Longitudinal Same sample measured over an extended time period. How do sport students develop skills over their degree programme? Measure skills at start, end of Year 1, end of Year 2 etc. Can last from 6 months – many years! Therefore not recommended for student projects! What if people drop out (attrition)? If you wanted to do a longitudinal style design, you can approximate longitudinal with Cross-sectional designs. Ask people to recall past attitudes or behaviours, to state their present attitudes and behaviours, and predict their future attitudes and behaviours. Obviously not as reliable or valid… 5. Case study Take one case (group/organisation/team/individual) to study in depth. Choose a case that is: • Typical – greater external validity. • Extreme – shows theory to greatest extent, e.g. study on hooliganism may choose case study of Millwall FC. • ‘Interesting’ – e.g. new surf reef at Bournemouth. 6. Ethnography • Generally qualitative approach. • Associated with researching a group. • Become part of the group. • Understand the group from the group’s (emic) perspective rather than your (etic) perspective. • Collect data using whatever means possible at whatever time you can. Ethnographic approaches: Observation Participant observation Research Interviews question Focus groups What is ethnography? Ethnography is the study of people in naturally occurring settings or ‘fields’ by methods of data collection which capture their social meanings and ordinary activities, involving the researcher participating directly in the setting, if not also the activities, in order to collect data in a systematic manner but without meaning being imposed on them externally. (Brewer 2000, p.189) The ethnographer participates, overtly or covertly, in people's daily lives for an extended period of time, watching what happens, listening to what is said, asking questions; in fact collecting whatever data are available to throw light on the issues with which he or she is concerned. (Hammersley & Atkinson, 1995, p.2) Could I do an ethnography? • Am I looking to study one group in depth? • Am I looking to gain an understanding of that group? • Can I gain access to the group? • Will I be able to collect valid and reliable data? SELECTING A SAMPLE 1. Define the population. The population consists of every individual case that possesses the characteristic that is of interest to the researcher. 2. The second stage is to determine your sampling method. A number of sampling techniques can be used. The most common associated with quantitative research are: • Random sampling. • Stratified random sampling. • Cluster sampling. • Systematic sampling. RANDOM SAMPLING A random sample is where every member of the population has an equal probability of being selected. It is the best technique to obtain a representative sample, and produce findings that will be generalisable to the overall population. OBTAINING A RANDOM SAMPLE The first stage is to define your population. The next stage is to ensure that each member of this population has an equal chance of being selected: 1. Place the names of all the population in a container, and pick names until the desired sample size has been achieved. 2. Each name can be assigned a number, and a random number table or computer software can be used to select the sample. STRATIFIED RANDOM SAMPLING If there are certain subgroups within the population, for example based on age, sex and so on, then it may be necessary to ensure that they are adequately represented in the final sample. In this case, the population is divided into subgroups. Random samples are then taken from within these groups. Thus, you may divide your population into ‘male’ and ‘female’, and randomly select 50 per cent of your sample from the list of your male participants, and the remaining 50 per cent from the list of your female participants. CLUSTER SAMPLING Where groups are randomly selected, rather than individuals. Thus, if the researcher was interested in the attitudes towards intimidatory behaviour in little league baseball, then a number of teams could be selected at random, and all players within those teams questioned. It is important to select a number of clusters to ensure generalisability with this method. SYSTEMATIC SAMPLING Involves selecting every Kth case, for example taking every fourth name from a list, or every seventh person to enter a sports facility. Systematic sampling is best recommended when the list from which the names are taken is randomly ordered, otherwise some bias is likely to occur. NON-PROBABILITY SAMPLING METHODS There are a range of non-probability sampling methods, which will be of interest to the qualitative researcher. The term ‘sample’ is perhaps less appropriate here, with ‘selection’ more reflective of the process, as generalisability is not the prime concern of the qualitative researcher. The focus is more upon a sample who can describe, explain and illuminate the phenomena that is being explored. SNOWBALL SAMPLING Locate your initial participants, and these initial participants identify further potential participants themselves. Thus, you may find access to a ‘gatekeeper’, or influential member of your population. They can then introduce you to other participants, who themselves will be able to give you access to further participants. One potential advantage of this is that by being introduced by a known member of the population, you may be able to engender greater trust between researcher and subject, with subsequent improvement in the quality of your data. THEORETICAL SAMPLING Identify cases that demonstrate a particular theory particularly well. Use where a random sample would be unlikely to demonstrate the theory or concept under investigation. TYPICAL CASES Your sample is chosen on the basis that they are ‘typical’ of a particular theory. EXTREME CASES You choose cases that are extreme cases of a theory. An example of this would be an investigation into the personality characteristics of elite athletes. You may wish to sample Olympic medal winners as extreme cases. OPPORTUNISTIC You select samples as they arise, taking advantage of unexpected opportunities. For example you may be introduced to a particular key informant at an unexpected time. CONVENIENCE The sample is chosen as it is convenient in terms of location, accessibility, etc. Try to avoid convenience samples as far as you can – it is always tempting to hand out questionnaires to those you are in day to day contact with, or interview people that you know! KEY INFORMANT TECHNIQUE Individuals are chosen on the basis of specific knowledge that they possess, for example they may have a particular role or responsibility within an organisation. MISTAKES MADE IN SELECTING A SAMPLE Some of the common errors made in sampling include: • Selecting individuals who are convenient, or readily available, for example individuals that are already known to the researcher. • Selecting individuals who volunteer to take part as well as those who are more randomly selected without reference to the potential differences between these. • Introducing bias through selecting a non-random sample. • Using a random sample when other sampling methods would be more appropriate. • Not obtaining a large enough sample for the purposes of the project. ASSESSING THE ETHICS OF YOUR RESEARCH Make sure you read the codes of conduct laid down by your specific discipline (e.g. American Psychological Association or the British Psychological Society). Such codes generally include the following key requirements: • Risks to participants are outweighed by the benefits of the research programme. • Participation should be voluntary. • Risks to participants are eliminated or minimised as far as possible, including psychological and social, as well as physical risks. • All information should be treated as strictly confidential. • The participants have the right to be informed of the purpose of the study. • Participants may withdraw at any time. • Participants should be debriefed after the research programme. • Ensure that you gain ethical approval from the relevant committee at your institution before any research programme involving ethical issues commences. SUMMARY 1. You should have a carefully considered approach to the collection of your data, in terms of who you will collect data from, when such data will be collected, and how such data will be collected. This is your research design. 2. We have identified six research designs: experimental, cross-sectional, time series, longitudinal, case study and ethnography. 3. Whatever research design you adopt, it is unlikely that you will be able to collect data from the entire population. Thus you will have to collect data from a smaller group within that population – your sample. 4. You will also have to choose the methods by which you are going to collect data from that sample. 5. It may be possible to increase the validity of your research by undertaking some form of triangulation. Triangulation means collecting data from more than one perspective; for example it may involve collecting qualitative, as well as quantitative data. 6. You will also need to consider the ethical issues involved in your research. If in doubt, you should gain approval from the committee responsible for ethical issues at your institution.