Conducting Pilot Studies and The Experimental (Lab
A pilot study is a small scale study carried out with a few participants to assess and eliminate any possible problems with the planned study. Some of the things which can be testes are shown below:
Standardised instructions : are these clear? Do all participants understand them? Are questions in an attitude scaled ambiguous? How much time should participants be given to complete a task?
An experiment is the most rigorous of all techniques used in psychology.
This is because the experimenter has full control over the independent variable (IV) which can be manipulated to produce a change in the dependent variable (DV) and has the best chance to control extraneous variables.
A good way to achieve this within an experiment is to have a control group which remains untreated. The control group provides a baseline against which to measure the findings.
Field Experiment,Natural Experiments and Experimen
The field experiment meets all the criteria for an experiment but is carried out in natural setting. An example is the study by Hofling et al 1966.
A natural Experiment takes advantage of naturally occurring events. An example is the adoption studies of Heston 1966 which compared incidents of schizophrenia in children or schizophrenia parents who were adopted before the age of one month, with children of normal parents raised in their own home.
Such studies have the advantage of being able to study things that it wouldn't be ethical to set up but can be utilised as they happen.
The main disadvantage is that the experimenter has no control over allocation of participants to conditions. Therefore, internal validity is low.
Experimental Designs, apart from considering the IV, DV and possible extraneous variables, the design of the study needs to be chosen carefully. There are three possible basic designs that we can choose from.
Repeated Measures Designs
Sometimes called related or within-group designs.
Suppose we overheard a conversation in a coffee shop that more wars seem to take place in hot countries. From this chance conversation, we decide to test the theory that heat makes people aggressive. We could ask participants to perform a problem solving task in cold conditions and again, in hot conditions. We could then look for signs of aggression in the participants when the task became difficult.
The main advantage of a repeated measures design is that it eliminates the influence of individual differences since the same people are tested in each condition.
Problems with repeated measure designs, order effects. These can be a problem if the same people perform a task under two different conditions. Performance may improve through practice (practice effects) or worsen due to fatigue or boredom (fatigue effects). Demand characteristics may also be a problem since the participant may guess the true purpose of the experiment by the time he participates in the second condition.
Solution, one solution to this problem is counterbalancing.
Independent Measure Designs
Sometimes called unrelated or within-groups designs.
Sometimes an experiment cannot be counterbalances to get rid of order effects. In this case an independent measures design ( where different people participate in the two conditions0 must be used.
Such cases could be: When order effects can't be dealt with or are asymmetrical. When testing differences, between males and females or introverts and extroverts. Where a naive participant must be used because the real purpose of the experiment can be guessed after the first condition. Where a control group is needed, testing a new drug against a placebo.
The main advantage of this design is that it eliminates order effects since fresh participants take part in the two conditions. It also reduces the effects of demand characteristics since the participant only takes part in one condition.
The main drawback with this design is that individual differences can never be totally eliminated. Another problem is that independent measures designs require twice as many people which can be expensive and time consuming.
Matched Pairs Designs
Sometimes the way to avoid fatigue and practice effects is to find different people but match them for certain criteria, such as race, age, sex, intellectual ability and socio-economic status. This type of experiment is called a matched pairs design.
It's a related design because the pairs of scores are treated just as if one person had obtained a set of scores by participating in each condition. The pairs of participants are assigned randomly to the different conditions.
This type of design utilises the advantages of both the repeated measures and independent measures designs while avoiding some of the problems. However, this type of design can be expensive since if one person drops out of the study, the paired participant is also lost.
These focus on naturally occurring behaviour. There are many types of observational study. The two being focused on here are naturalistic observation, controlled observation.
Naturalistic observation involves the researcher observing behaviour as it occurs in the natural environment. The observer has no control over the experiment and cannot manipulate the IV. However, it does have high ecological validity and overcomes the artificial environment of the laboratory.
Controlled observation is one in which the researcher attempts to control certain variables in order to be able to observe behaviour under a certain set of circumstances. Ainsworth 1970 performed a controlled observation to look at attachment behaviour in infants.
Methods: Participant observation - observer joins and becomes part of the group being observed.Non-participants observation - observer remains outside the group being observed.
Disclosed observation - participants aware they are being observed. Undisclosed observation - participants unaware they are being observed.
Structured observation - researcher determines precisely what types of behaviour are to be observed and constructs a standardised checklist. Unstructured observation - data fathered in unplanned ad hoc way.
Systems used for written observations
Rating scale - behaviour rated on a pre-designed scale, coding of an aggressive act on a scale of 1 to 10 ( mild to severe )
Coding system - behaviour recorded using number or letter system, V for verbal aggression, P for physical aggression.
Categorisation system (tally chart) - pre-categorised behaviour recorded as frequency of occurrence.
Advantage is that it has high ecological validity and avoids demand characteristics
Disadvantage is that there may be observer effects and replication may be difficult.
A correlation is an association or relationship between two variables.
An example is the study by Rahe et al 1970 found a correlation between LCU ( life change unit) scores and illness scores. Rahe found a positive correlation. That is, the higher the number of LCUs, the greater the risk of developing illness.
Alternatively, we can have negative correlations where more of one thing leads to less of another, the hotter it gets, the less clothes we wear.
Correlations are measured from +1 to -1. +1 is a perfect positive correlation. -1 is a perfect negative correlation. 0 means that there's no correlation at all.
In practice, perfect correlations are rarely found but usually fall somewhere between these extreme values. A correlation of 0.7 would show a fairly strong positive relationship between two variables, while a correlation of -0.3 would show a fairly weak negative one.
A scattergram can be used to determine visually if there's a relationship between two variables. The scores on one variable are plotted against the scores on the second variable.
Interviews and Surveys
Interviews can be structured where every respondent answers the same questions, or unstructured where there is more flexibility in what is asked. Alternatively, the semi-structured approach has the advantages of both the structured and unstructured approaches.
An example of semi- structures approach is the clinical interview, used to assess a person with a mental disorder.
The technique allows the respondent to provide detailed information about his/her family history and symptoms, at the same time, structures tests can be administered to provide information on the person's cognitive or social functioning.
This approach allows the clinical interview to be used for diagnosis and treatment although of its disadvantages is that it can still be limited by the respondents ability to express himself clearly.
In addition, the interviewer must be sensitive or he may created demand characteristics, especially if the person's future depends on their responses in the interview.
Designing Interviews, interviews must be planned carefully taking procedural and ethical issues into account. Below is a checklist for planning interviews.
Before the Interview- clearly describe the research problem, state the aim, link the problem to an appropriate theory, identify the categories of data to be collected.
The questions- generate an appropriate set of questions, plan the order in which questions will be presented, plan the interview to obtain the right balance between structured and unstructured items.
The interview- consider how you will present yourself, identify and approach potential respondents, plan a pre interview meeting, plan a post interview debriefing, decide how much information is to be collected during the interview, consider the ethical issues raised by the research and seek advice as necessary.
Evaluation- detailed information can be obtained and interviewer can clarify data collected at time, unstructured interviews can raise new lines of psychological inquiry, produce qualitative data which can be difficult to analyse, increased risk of investigator effects.
Questionnaires used in surveys may use closed ended questions whereby the respondent is provided with alternative answer choices and must pick one, or open ended questions which allow them to answer in any way they wish. Both of these have advantages and disadvantages.
Closed questions are easier to analyse quantitatively since the respondents' choices are arranged in categories but the researcher may not get access to the respondent's true feelings towards a particular topic and could miss important information they have not pre-categorised. Alternatively, open questions allow the researcher to collect rich, detailed information but are much more difficult to analyse.
Care must be taken when designing questionnaires since you'll not be there to clarify any questions which respondent may find ambiguous or difficult.
Type of question- closed questions are easy to analyse since you can provide a box for respondent to tick yes or no. Alternatively you can offer a scale along which respondent can indicate the strength of their feelings towards a particular topic or a list of options from which to choose. Open questions don't constrain respondent but allow them to answer in any way they like. They are much harder to analyse but may give access to information you haven't thought of and pre-categorised.
Number of questions- only include questions which are necessary for the purpose of the research. Include demographic question on age, sex, marital status, at the end of the questionnaire. Don't put highly sensitive questions at the beginning of the questionnaire.
Langauge- use language which is clear and unambiguous, avoid jargon or technical language.
Leading questions- never word a questions that leads a respondent in a particular way, never include value judgements in questions.
Ask one questions at a time- take care that you don't ask two questions rolled into one, you won't be sure which part of the question the participant has answered.
Avoid using Emotive language- can bias response.
Avoid making inappropriate or insensitive assumptions- avoid questions which include assumptions that could cause embarrassment.
Evaluation- can be used to question a large sample of people relatively quickly and can collect large amount of data, reduces investigator effects, answers may be affected by social desirability, low response rates self selecting sample.
The case study provides a detailed account, usually of a single participant. It contains detailed personal history, background, test results IQ score or personality profile, rating and records of interviews. This method gives rich information about the person but the data are often qualitative and therefore it's difficult to make interferences about cause and effect. Below is an extract from a famous case study of a patient with 'multiple personality disorder' Eve White.
As if seized by sudden pain, she put both hands to her head. After a tense moment of silence, both hands dropped. There was a quick, reckless smile, and, in a bright voice that sparkled, she said "Hi there doc!" The demure and constrained posture of Eve White had melted into buoyant repose. This new and apparently carefree girl spoke casually of Eve White and her problems, always using 'she' or 'her' in every reference, always respecting the strict bounds of a separate identity. When asked he name, she immediately replied "oh i'm eve black". Source "The three faces of eve" Thigpen and Cleckly 1954.
Advantages of case studies such as this one can be useful to study an unusual set of circumstances and lend themselves to a longitudinal approach where the person is followed up over time. They're also useful when investigating a new line of psychological research.
Problems are data collection for case studies relies on the memory of the participant and close family. Also a close relationship develops between experimenter and client which may lead to bias. Cause and effect are difficult to establish especially if there's no way to prove the details of the individual's past. The limited sample makes it difficult to generalise and they can be extremely time consuming and expensive.