Research Methods

• The independent variable (cause) is manipulated by the researcher to produce a change in the dependent variable (effect).
• All other variables (extraneous variables) are held constant or eliminated.
• Participants are randomly allocated into their experimental conditions.
• Experiments look for a difference between two conditions.

Laboratory Experiment: The researcher has complete control over the situation.

Field Experiment: The researcher doesn't have control over the situation as it occurs in the real world. The participants don't know they are taking part.

Natural Experiment: The IV varies naturally and cannot be deliberately manipulated by the researcher for ethical or practical reasons.

Quasi Experiment: The IV occurs naturally as a difference between people that already exists e.g. gender.

• No IV is deliberately manipulated by the researcher. Therefore it is hard to see a clear cause and effect.
• Naturally occurring behaviour is studied meaning there is high ecological validity.
• Observation involves careful watching and recording of behaviour.

Naturalistic Observation: Naturally occurring situation beyond the researcher's control.

Controlled Observation: A setup scenario to observe a particular behaviour.

Overt Observation: The observer is in plain sight. The participants are aware they are being observed.

Covert Observation: The observer is hidden. The participants are unaware they are being observed.

Participant Observation: The observer becomes actively involved in the situation to gain more valuable insights.

Non-participant Observation: The observer doesn't become involved in the situation.

Interviews

Involve asking the participant questions face to face, the interviewer records their answers.

Structured Interview: All the questions are decided on beforehand and may have a limited range of answers which gains more quantitative data.

Unstructured Interview: Resemble conversations, the interviewer can ask any questions about a certain topic that will get the best response. Response to the answers gains a clear insight into their thoughts and feelings and gains more qualitative data.

Questionnaires

Involve giving the participant a list of questions to write down their own answers.

• A correlation looks to see whether two variables are related. 
• There is no manipulation of one variable, it is just the measurement of two variables to see a relationship between them.
• They don't show a cause and effect between two variables.
• Correlations can be carried out where it would be unethical to manipulate the IV.

Positive Correlations: Involve a relationship where as one variable increases the other increases.

Negative Correlations: Involve a relationship where as one variable increases the other decreases.

Correlation Co-efficients: Can range from +1 (a perfect positive correlation) to -1 (a perfect negative correlation) 0 indicates no relationship at all between the variables.

An in-depth study, using a range of methods on one person or a small group. They usually involve analysis of an unusual individual or event. It can use a range of different research methods, this increases reliability.

Advantages:

• A good method to gain qualitative data into the situation of the person or group.
• Gives an opportunity to investigate something that can't be manipulated.

Disadvantages:

• The results can't be generalised because it is only based on a small sample.
• There could be ethical issues with interference in the person's life.

• The aim is what you are trying to find out, what the purpose of the study is.
• A hypothesis is a testable statement about what you expect to happen in the study.
• The aim is more general than the hypothesis and the hypothesis has to be a specific prediction.
• The hypothesis predicts that there will be an effect on the IV or the DV or a relationship between two co-variables.

Directional Hypothesis: Used when we are predicting which way the results will go. This could be a positive or negative effect/relationship.

Non-directional Hypothesis: Used when we don't know whether it will be a positive or negative effect/relationship.

• A sample is the set of people from the general population that we are interested in.
• We can very rarely do studies on the whole target population so a sample is used to make predictions about the whole target population.
• The size of the sample is important to represent the population, generally for a small-scale study 30 participants are enough to make accurate predictions.

Random Sampling: Every person in the target population has an equal chance of being chosen to take part.

Systematic Sampling: Every nth person from a list is taken to create a sample. E.g. 1000 people and a sample of 20 is needed; 1000/20=50 so every 50th person on a list is chosen for the sample.

Stratified Sampling: A representative sample is obtained. The population is divided into subgroups e.g. gender. The same percentage of that subgroup in the population is kept for the sample.

Opportunity Sampling: People who happen to be available are used for the sample.

Volunteer Sampling: Participants volunteer to take part usually through advertising.

Problems that might occur with the research can often be spotted by carrying out a pilot study. This is a small-scale prototype of the study, carried out on a smaller number of participants.

Pilot studies are useful to find out:

• Potential problems with the design.
• Whether the instructions are clear.
• In a self-report study, whether the questions are clear.
• Potential problems with measuring the DV.
• How long the study will take to carry out.

Pilot studies give the researcher useful practice in approaching participants and using any technical apparatus and materials. Participants can be informed that it is a pilot study and can be asked for feedback on any problems with the procedure.

Independent Groups Design: Two completely different groups of people are used for each condition.

Advantages: No order effects, the participants can't practice performing better the next time.

Disadvantages: Participant variables can affect the results.

Repeated Measures Design: Only one group of participants is used to do both conditions.

Advantages: No participant variables, same person's performance is measured.

Disadvantages: Order effects.

Matched Pairs Design: Two groups of participants are used, but each one is matched on a relevant variable to a participant in the other group to make them as alike as possible.

Advantages: Solves the problem of participant variables and order effects.

Disadvantages: Can be difficult and time-consuming, not always obvious what variable needs to be matched.

• When observing behaviour, most psychologists categorise behaviour rather than recording every gesture.
• Behavioural categories need to be clearly defined (operationalised).
• These behaviours can then be recorded every time they are observed.
• Most observations involve producing a checklist or tally chart to keep track of how many times the behaviour has occurred. 
• Video recordings are often made to check the accuracy of the observers. This is particularly important when there is more than one observer.

Event Sampling: The researcher records an event every time it occurs. 

Time Sampling: The researcher decides on a time and then records what's happening at that time.

The sort of questions asked and how they are phrased needs careful consideration.

• Open and Closed Questions: Open questions give qualitative data and closed questions give quantitative data.
• Order of the Questions: Sensitive questions are usually asked later on in the questionnaire so people aren't put off at the start.
• The use of Jargon: If specialist terms are used, some people may not be able to understand the question so plain English is essential.
• The use of Leading Questions: A question that makes an assumption makes it difficult to answer honestly.
• The use of Double-Barrelled Questions: A two-part question may lead the participant to want to give different answers to each part, meaning they don't know how to answer.
• The use of Emotive Questions: Certain topics produce emotion and this needs to be made clear, giving the respondent a choice whether or not to answer.
• The use of Vague or Ambiguous Questions: The question needs to be interpreted in the same way by everyone to produce useful data.
• The use of Inappropriate Assumptions: It may be embarrassing or upsetting if the questions make assumptions about them.

• Interviews have a lot in common with questionnaires so the issues are very similar.
• This is particularly for structured interviews which are like face to face questionnaires.
• However, some interviews are almost completely unstructured and generate long, detailed answers.
• It is particularly important for the interviewer to have a clear plan, or the research will be too vague and the results impossible to analyse.
• The interviewer may record the interview and also take notes. 
• Details can be missed when notes are made, therefore recording the interview means that nothing is missed as it can be listened back to, however, the participant may be put off by it.
• Body language, facial expressions and tone of voice can also help gain insight into the topic.

There are factors that can affect the outcome of an experiment other than the independent variable.

• Extraneous Variables: Any variable other than the IV that may have an effect on the DV if it is not controlled. Where possible they are identified at the start of the study and are minimised. 
• Confounding Variables: Any variable other than the IV that may have had an effect on the DV so that we cannot be sure of the true source of the changes to the DV. Extraneous variables can become confounding variables if they are not controlled before the study.
• Variables should be Operationalised, this means they should be clearly defined so that it can be manipulated and measured.
• Random Allocation controls for participant variables in an independent groups design, ensuring each participant has the came chance of being in one condition as the other.
• Counterbalancing controls for order effects, condition A then B then B then A.
• Standardisation means the exact same procedures and instructions are used for all participants to avoid investigator effects caused by different instructions. Scoring and environment should also be kept the same.
• Randomisation Order of tasks or presentation of data is decided on the toss of a coin or other random method of selection to control for order effects

Demand Characteristics: This happens when the participants know they are in a study so they try to make sense of it. They become a problem when participants change their behaviour to act differently to how they would normally. They may involve:

• Trying to guess the purpose of the research and trying to act the way they think is expected.
• Displaying the social desirability bias; trying to make themselves appear in the most favourable way.

Investigator Effects: Happen when the researcher's behaviour affects the outcome of the research. This can happen by:

• The investigator expectations of what will happen in the study may make them biased in their observations. These expectations can also communicate to the participants.

Ethics are linked to morality. The rights and wrongs of our actions need to be weighed up such as whether they hurt or benefit others. The research has to be ethically justified so it can be carried out. If the benefits to humankind outweigh the costs such as psychological and physical harm, the research is justified.

British Psychological Society (BPS) Code of Ethics:

• Consent: Full disclosure of the objectives of the research is required for the participant to give full consent.
• Deception: Participants must not be deceived as to the nature of the study. If they are deceived full consent can't be gained.
• Confidentiality: It should not be possible to identify the participants from the report.
• Protection: Participants should be protected from mental and physical harm.
• Observational Research: If consent is not obtained then only behaviour observed in public is allowed.
• Giving Advice: If the investigator obtains evidence of a psychological/physical problem, the participant should be informed and recommended an appropriate advisor.
• Right to Withdraw: The participant has the right to leave the study whenever they feel like it.

• Lab Experiment:
• Withdrawal: participants may feel pressured to continue with the experiment when they don't want to.
• Protection from harm: the experiment may put the participant under stress, especially when they feel they can't withdraw.
• Field Experiment:
• Consent: no consent is asked for because participants need natural behaviour.
• Deception: participants don't know they are taking part in a study and believe the situation is real.
• Natural Experiment: 
• Consent: participants don't know they are being studied.
• Confidentiality: need to be careful not to pry into private matters or reveal individuals identities.
• Naturalistic Observation:
• Consent: participants don't know they are being studied.
• Withdrawal: because they don't know they're being studied they can't withdraw.
• Confidentiality: there is need to be careful with private matters.

• Questionnaires and Interviews:
• Withdrawal: participants may feel drawn into a conversation which they find it difficult to back out of.
• Confidentiality: it is sometimes possible to identify someone from what they say.
• Protection: especially in unstructured interviews, they interviewer may delve into personal issues that the participant may find painful to discuss.
• Giving Advice: if it becomes apparent that the participant has psychological problems, the interviewer should get them help.
• Correlations:
• Protection: researchers need to be careful that their results aren't misinterpreted as this could cause harm to a group.
• Case Studies:
• Confidentiality: because it is based on individuals, it may be possible to identify someone.
• Protection: the intrusion of being researched could cause psychological damage.

Consent: If you are unable to gain the consent of participants in the study, others who are similar to the participants can be asked if they think it is okay to be carried out.

Deception: Debriefing is carried out, here the situation and the results are explained to the participants, they then have the right to withdraw their result.

Protection from Harm: The participants have the right to withdraw at any point in the research. They are also debriefed and explained to that they have been harmed, they can then get help to resolve the situation.

• The researcher carries out the research and prepares a manuscript for publishing.
• This manuscript is sent to experts in the relevant field who act as peer reviewers, they review all aspects of the manuscript and they comment on the content, methodology and suitability for publication.
• The reviewer is asked to make an explicit recommendation of what to do with the manuscript. Most are along the lines of:
  • unconditionally accepting the manuscript.
  • accepting it in the event that it's authors improve it in certain ways.
  • rejecting it, but encourage revision and invite a resubmission.
  • rejecting it outright. 
• The manuscript gets sent to the editor who publishes it.

Implications of psychological research on the economy refer to how useful research findings are and how governments could use them. Of particular interest is the economic worth of these findings.

Attachment Research: Bowlby said that babies need the constant care of the mother for healthy psychological development. Later evidence showed that this is not the case and babies can have good substitute care, therefore mothers can return to work after having a child, staying economically active.

Psychopathology Research: Anything to do with treatment and people's ability to work and contribute as effective members of society would be relevant. Research shows that people with a disorder such as depression are less likely to suffer a relapse after having cognitive therapy then, even though it may be more expensive than drug therapy, in the long term it would mean people would have less time off work.

Memory Research: Any evidence relating to a more effective use of public money would be relevant. Research showing the cognitive interview facilitates accuracy of eyewitness reporting enables better use of police time and resources.

Psychologists need to be able to measure variables consistently; this is called reliability and is used to assess both experimental procedures and 'tools' in tests, interviews and behavioural categories. A reliable psychological test will always produce the same findings.

• Internal Reliability: Consistency within itself e.g. a questionnaire should measure all the same phenomenon.
• External Reliability: Consistency with different occasions, researchers and settings.
• Test-Retest: Used with questionnaires and psychological tests and can be used with interviews, it is tested then retested.
• Inter-Observer: Several observers watch and record the same behaviour categories and compare to see if they record the same thing.
• Improving Reliability:
  • Questionnaires: measure over time using test-retest, should produce a correlation of +0.8, questions can be changed and ambiguous questions should be avoided.
  • Interviews: the same interviewer used each time, trained interviewer, avoid ambiguous/leading questions and use structured interviews.
  • Experiments: lab rather than field used for control and no extraneous variables and tests should be done under the same conditions.
  • Observations: operationalised behavioural categories, self-evident and no overlap.

• Face Validity: Does it appear to measure what it is supposed to?
  • To assess: one or more judges assess whether the test seems appropriate and suggest changes if necessary.
• Ecological Validity: Can the findings be generalised to other settings and situations?
  • To assess: test it out on other settings/situations to see if the findings remain the same.
    
• Concurrent Validity: Does the measure fit with other existing measures?
  • To assess: correlate the two findings to see if the findings remain the same.
• Temporal Validity: Can the findings be generalised to other times?
  • To assess: test it out at a different time period to see if findings remain the same.
• Predictive Validity: Can we use the findings to accurately predict from?
• Population Validity: Can we use the findings for different cultures etc.?

Experimental Research: Have a control group to see if the IV has and effect on the DV, standardise the procedure and use single-blind (participant doesn't know who is receiving a particular treatment) or double-blind (participant and investigator don't know).

Questionnaires: A lie scale could be used: questions are flipped to see if the participants are lying.

Observations: Covert observations: participants aren't aware of the observation and the behavioural categories observed need to be operationalised, clear, unambiguous and narrow. 

Qualitative Methods: There could be a problem with the interpretation of participants answers, to improve this a reflexive practice is used, they should repeat what they said.

Objectivity: Scientists agree that objectivity must be maintained as part of their investigations, they must keep a 'critical distance' during their research. This means that their opinions don't affect the results making it biased. Objective methods in psychology include:

• Questionnaire
• Lab experiment

The Empirical Method: Empiricism is derived from the Greek for 'experience'. Empirical methods emphasise the importance of data collection based on direct experience (primary data) rather than using secondary methods such as a meta-analysis or 'guess work'. Empirical methods include:

• Observations
• Lab experiments
• Field experiments

Replication: If a scientific theory is to be trusted then the findings from it must be able to be repeated in a number of different contexts and circumstances, this helps to determine how valid it is. For replication to become possible psychologists must report their investigations with as much precision as possible. This allows other researchers to verify their work and findings.

Falsifiability: Karl Popper argued that key criteria of science are the notion of falsifiability. He states that for a theory to be genuine they should be able to be tested and the possibility of being proved wrong should exist. Therefore true science shows that the data is true but also shown to be untrue.

Theory Construction and Hypothesis Testing: Theories are constructed through the gathering of evidence via direct observation. It should be possible to make clear and precise predictions on the basis of a theory, this is the role of hypothesis testing. An essential component of a theory is that it can be scientifically tested and that it could give rise to a number of potential hypotheses.

Paradigm: A set of shared assumptions and agreed methods within a scientific discipline.

Paradigm Shift: The result of a scientific revolution, a significant change in the dominant unifying theory within a scientific discipline.

Kuhn argues that the social sciences are fundamentally different to the physical sciences. He suggests that psychology belongs to the social science grouping. Kuhn argues that the physical sciences could hold one unifying theory or paradigm which may then shift when new evidence is discovered. However, he argues that this cannot apply to the social sciences as there is not one unifying theory upon which we could test our ideas. Behaviourism was once seen as a possible unifying theory for psychology, it was popular for its scientific simplicity; simple laws of cause and effect.

However, we now consider it too simplistic to explain the complex phenomenon of human behaviour. Likewise, biological theories are considered reductionist. In reality, it is the complex interaction between nature and nurture which is required to understand human behaviour. The unique attribute of free will means that studying human behaviour will never be straightforward or easy.

• Title: this determines who reads the full report, needs to be concise but informative.
• Abstract: this is written last and is a 150-200 word clear, concise summary of the topic studied, description of participants, sampling technique, procedure, results, conclusion, implications etc.
• Introduction: explains the existing background research to the reader, the reasoning behind the investigation and predictions derived from the hypothesis.
• Method: this describes how the study was conducted, should contact enough information for the study to be replicated.
  • Design: a brief outline of the method and design used.
  • Participants: key features such as age or gender etc. and how they were selected.
  • Apparatus/Materials, Procedure: a precise explanation of how the study was conducted, how participants were allocated, instructions used and how data was collected.
• Results:
  • Descriptive Statistics: measures of central tendency and dispersion displayed in tables.
  • Inferential Statistics: used to analyse the data, whether null was accepted or rejected.
• Discussion: summary of the findings and explaining them. Other factors such as problems and relating it back to existing information from the introduction.
• References: all previous research that has been read should be listed here in alphabetical order.

Quantitative Data: quantity, numbers.

• Objective
• Precise numerical measures
• Lacks detail
• High in reliability
• Used for behaviour
• Collected in an 'artificial' setting

Qualitative Data: quality, descriptive.

• Subjective
• Imprecise non-numerical
• Rich and detailed
• Low in reliability
• Used for attitudes, opinions, beliefs
• Collected in 'real-life' settings

Primary Data: data gathered first hand from a source directly from the researchers.

Secondary Data: data already been gathered by someone and is used by someone else for further research.

Meta-Analysis: a technique that allows findings of several studies to be combined to give a set of overall conclusions. It can be useful as large quantities of data allow a trend and relationship to be seen more clearly than in a smaller study. A meta-analysis is especially helpful when smaller studies have found weak or contradictory results.

Measures of Central Tendency: gives average values, they provide a single value that is representative of the whole set of data by indicating the most typical value.

• Mean: the average, to find this all the scores are added up and divided by the total number of scores.
• Median: the middle number, to find this the scores are listed in order of size, the middle number is then found to give the median.
• Mode: the number that appears the most in the list of scores.

Measures of Dispersion: examines the spread of scores within the data set, this measures the variability between the values.

• Range: the difference between the largest and smallest scores, to find this the smallest number is taken away from the largest number.
• Standard Deviation: tells us how far each score is from the mean. The smaller it is, the more clustered the scores are around the mean.

Percentage: an amount out of 100, to find a percentage of something the difference is worked out between two numbers, this value is then divided by the original number then multiplied by 100.

Correlations: seeks to establish the extent to which two variables are related. It is possible that:

• As one variable increases, so does the other (positive correlation).
• As one variable increases, the other decreases (negative correlation).
• There is no correlation.

A correlation co-efficient can range from +1 (a perfect positive correlation) to -1 (a perfect negative correlation). 0 indicated no relationship.

Bar Chart: a diagram consisting of columns (bars), the heights of which indicate frequencies, so data on the x-axis is discrete, this means it falls into categories (nominal data). There should be a gap between each bar, this shows the data is discrete or non-continuous.

Histogram: somewhat similar to bar charts, however, they are used for continuous data meaning there are no gaps between the bars. They display interval data.

Pie Chart: suitable for displaying how a population is divided up into different parts and what proportion of the whole each portion represents. Pie charts are used to display nominal data.

Line Graph: similar to a histogram in that it is continuous data. However, the data is plotted in a series of points that are joined by straight lines. It is particularly useful for comparing two sets of data. Data on the x-axis must be continuous.

Scattergram: used to display results from a correlation. pairs of scores are plotted against each other to show their relationship. Patterns or trends can be calculated.

Normal Distribution: data can be normally distributed to show a bell-shaped curve. Characteristics include:

• a bell-shaped curve
• symmetrical
• the mean, median and mode all fall on the same central point.
• the two tails never touch the horizontal axis.

Curves can be positively skewed, negatively skewed or bimodal. Unless a distribution of scores is symmetrical, as in the normal distribution, it will be skewed. This is caused by outliers or extreme scores. A positively skewed distribution will have a long tail on the right and scores peak on the left, and the opposite for a negatively skewed distribution.

Inferential Statistics: this helps to discover if the results are statistically significant. A statistically significant result is one which is unlikely to have occurred through chance. Usually, psychologists use the 5% level of significance (0.05) this means they can be 95% sure the result is due to the manipulation of the IV. if it is significant we can reject the null hypothesis and if not we accept it.

Null Hypothesis: Any change between the two conditions is due to chance.

Sign Test: this involves counting up the number of positive and negative signs. The sign that least appears is the value of S. The number of scores listed minus the ones where there was no change is the value of N. If the hypothesis is directional the 0.05 level of significance is used for a one-tailed test and if it is non-directional the 0.05 level is used for a two-tailed test. S must then be equal to or less than the critical value shown for there to be significance.

Nominal: results that fall into two or more named categories, the groups are not related in any way that would mean they fall on a scale.

Ordinal: results are points from a scale and relate to each other so they could be put into order. The value of points increases along the scale but the gap between the points doesn't have to be equal. Data can be ordered in some way.

Interval: points are scored on a scale and increase in value, the divisions within the scale are equal.

Content Analysis: a method used to analyse qualitative data. It is used to indirectly observe the presence of certain words, images or concepts. Researchers familiarise themselves with the content and go back to the source and quantify and analyse the presence, meanings and relationships of words and concepts and make inferences about the messages within the media, the writer, the audience, the culture and the time. The media is coded and broken down into manageable categories and examined.

Thematic Analysis: can follow on from content analysis, once the data has been coded recurrent themes may emerge. This means the idea is either clearly shown or alluded to. This is more likely to be descriptive rather than quantitative.

Strengths: depth and detail, high validity shows true measures of the theme, idiographic approach.

Weaknesses: open to interpretation, low reliability as it can't be replicated to gain the same results as everyone is unique.

Once a statistical test has been calculated the result is the calculated value. This must be compared to the critical value and it's comparison tells us whether we can reject or accept the null hypothesis. Each statistical test has it's own table of critical values, those tests with the letter 'R' in their name are those where the calculated value must be equal to or exceed the critical value to be significant. In order to look up the critical value you need to know these:

• If the hypothesis is directional: if so a one-tailed test is carried out and if not then a two-tailed test is carried out.
• The significance level: most commonly 0.05
• The number of participants in the sample

Type 1 error: where the null hypothesis is rejected when it should actually be accepted because the results are due to chance. This is sometimes referred to as an optimistic error or a false positive and is most likely to occur at less stringent levels such as 0.05.

Type 2 error: where the null hypothesis is accepted when it should actually be rejected because the results are not due to chance. This is sometimes referred to as a pessimistic error or false negative and is most likely to occur at more stringent levels such as 0.01.

Type of Hypothesis:        Level of Measurement:        Design

                                                                             Independent               Repeated

Difference                     Nominal                              Chi-squared               Sign test

                                   Ordinal                                Mann-Whitney            Wilcoxon    

                                   Interval                                Unrelated t-test          Related t-test

Correlation                   Ordinal                                                                 Spearman's rank

                                   Interval                                                                Pearson's r