Research Methods

?
  • Created by: Honor
  • Created on: 05-03-17 17:38

Variables

Independent - the one you want to change

Dependent - the one you want to measure

Extraneous - anything that needs to be controlled

Confounding - variable that was not accounted for

1 of 32

Experiment Types

Laboratory - Controlled artificial environment, the IV is manipulated and participants are randomly allocated. Good for high levels of control, high replicability and concluding cause and effect. Bad for lacking ecological validity and a higher chance of investigator and participant effects.

Field - Natural environment, IV is manipulated and mostly random allocations. Good for concluding cause and effect with a higher ecological validity and reduction in participant effects. Bad for less control over the EV, more time consuming to set up and random allocation may be difficult.

Natural - IV controlled naturally, no manipulation and no random allocations. Good for high ecological validity and useful where manipulating the IV would be unethical or impossible. Bad for concluding cause and effect as there is no control so lower replicability.

2 of 32

Types of Effects

Investigator Effects - experiment may do something to influence the participants to get the desired results

Demand Characteristics - participants start to realise the experiment so they change their behaviour accordingly

Social Desirability - when people answer questions differently just so that they are presented in a good light

Mundane Realism - Extent to how well it matches real life tasks

3 of 32

Question Types

Open - allows respondent to write their own answer to produce qualitative data. Good for rich, detailed data so they can express how they really feel in real life. Bad for analysing as it can time consuming and difficult

Closed - chooses their response from a limited number of fixed answers to produce quantitative data. Good for analysing as can be statistically graphed in less time and interpreted easily. Bad for its lack of richness and unclear on whether they understood the question

4 of 32

Self Report Techniques

Questionnaires - research method composed of a series of questions designed to gather information about a certain topic. Good for efficiency as researchers do not have to be present and can collect large amounts of data about what people think. Bad for social desirability and difficult to phrase questions that have the same interpretation for everyone

Interviews - Verbal conversation to collect relevant research. Good for allowing participant to freely express their opinion, interviewer can ask them to clarify what they mean and data is more rich and detailed. Bad for being more time consuming, requires skilled researchers and social desirability is still an issue

5 of 32

Interview Types

Unstructured - may have a set of discussion topics but are less constrained about how the conversation goes. Good for flexiblity as questions can be adapted and changed depending on answer and interviewer can ask for clarification. Bad for being much harder to analyse the data as it is all different and may take longer to conduct as there is no end to the questions

Structured - any interview in which the questions are decided in advance and fixed for all participant. Good for analysing easily and quickly conducting so more data can be collected in the time. Bad for being unflexible and lacking detail.

6 of 32

Aims and Hypothesis

Aim - general statement about the investigation's purpose

Hypoethsis - precise statement about the expected outcome

Directional (one-tailed) - hypothesis that states which direction the results will go in

Non-directional (two tailed) - less clear on the direction but acknowledges there will be an effect

Null hypothesis - when there is no effect or result

Operationalise - clearly defining what the variable is and how to specifically measure it

7 of 32

Case Studies

In depth knowledge about an individual of small group of people that are unique.

Studied with the aim of uncovering answers that the scientific world needs.

E.g. Clive Wearing could remember procedural memory but not his episodic

Good for being in depth and detailed and more ecologically valid as its in an everyday natural environment.

Bad for being population valid as individual differences may affect how well it can be applied and investigator effects may interept the information as their aim.

Ethical issues may also include lack of privacy as it is spread for research and having informed consent about being shared.

8 of 32

Experimental Design

Repeated Measures - when the same particpants are used in both conditions. Good for using less partipants, less time and the particpant variables are limited. Bad for needing more stimulus material, demand characteristics and order effects.

Independant Groups - particpants are randomly allocated to different condition groups. Good for having no order effects and reduced chance of demand characteristics. Bad for needing more participants, time to randomly allocate and cannot control participant variables.

Matched Pairs - pairs of participants are closely matched and are then randomly allocated to one of the conditions. Good for having no order effects, participant variables are minimised and there is less chance of demand characteristics. Bad for being difficult and time consuming to match and more participants are needed.

9 of 32

Sampling Techniques

Opportunity - selected by people who are the most easily available at the time. Good for needing less time to locate. Bad for being a small possibly biased population so cannot be generalised.

Volunteer - volunteers make up the sample, sometimes called self-selected. Good for having more variety so more representative and being less likely to withdraw. Bad for having volunteer bias as only certain type of person volunteers

Random - every member in target population has equal chance. Good for being unbiased as there is a equal chance for everyone and also higher population validity. Bad for taking time to locate and contact the particpants.

Stratified - sub groups are identified like age and chosen randomly within that sub-group. Good for being random but still representative of sub group and more variety of people. Bad for being time consuming to identify sub groups and pick them.

Systematic - using a predetermined system to random locate them e.g. pick every 1st and 3rd name. Good for being unbiased as it is using an objective system so investigator will not have influence. Bad for missing subgroups so may not be truly representative

10 of 32

Pilot Studies

Study conducted with a small sample before the real study

Good for helping the researcher identifying any possible problems with the method, design, instructions to participants and so on. Can also check quesionnair items are easy to answer and unambiguous. Can do all this checking without having to invest in a large amount of time and money for a full-scale study

11 of 32

Observational Methods X

Naturalistic - carried out in an everyday setting, does not interfere in anyway. Good for having high ecological validity and applying to real life. Bad for not controlling all the extranous variables.

Controlled - behavior is observed under controlled lab conditions. Good for having more control over extranous variables. Bad for having lower ecologicsal validity.

Covert - when the researcher observes in secret. Good for hearing the opinions that only a member of the study group may hear. Bad for having ethical problems as deception and consent may pose as an issue.

Overt - when the researcher tells the group they are observing them. Good as there is no ethical issues regarding consent. Bad as they may display social desiribility effects.

Participant - when the researcher joins in and becomes part of the group they are studying. Good for. Good for hearing the opinions that only a member of the study group may hear. Bad for reducing validity if they get too involved as they may produce a bias

Non-participant - whe

12 of 32

Sampling in Observations

Event Sampling - counting the number of times a certain behvaiour occurs from population in a given time frame. Good for not missing any behaviouir and being in depth and detailed. Bad for not being to record everything if too many happen at once and may be time consuming.

Time Sampling - recording behaviours in a given time frame. Good for being convient as the observer has time to record what they have seen. Bad for not being representative and may miss important behaviour.

Behavioural Catergories - specific type of behaviour which is deinfed before the study takes place and allows researchers to focus on a certain target so that the most valid and reliable data is produced

13 of 32

Types of Data

Qualitative - typically descriptive and cannot be made into statistics. Good for being in depth and descriptive. Bad for being harder to analyse.

Quantitative - data that can be made into useable statistics. Good for being easier to anaylse. Bad for lacking in depth and detail.

Primary - data collected yourself. Good for being more reliable and it can answer your direct research questions. Bad for being more expensive and not immediately available as it takes time

Secondary - data collected by someone else e.g. meta analysis. Good for being inexpensive, easily accessible and available. Bad for being outdated, potentially unreliable and may not totally answer your research questions directly like primary would

14 of 32

Summarising Data in Graphical Form

Histograms - bars are joined together with equal widths and uses continous data i.e. x axis will increase or decrease. Likely to be used in data analysis to show distribution of items in a data set

Bar Charts - bars are not joined together and bars are an equal width. The data is single and seperate and likely to be used when comparing classes or groups of data 

Scattergram - more points in a cluster mean a stronger relationship. A forward slant means a positive correlation and a backward slant means there is a negative correlation. Likely to be used when a good visual is needed to represent a relationship between two variables.

15 of 32

Measures of Central Tendency

Mean - the statistical average. Calculated by adding up all the scores in a set of data and then dividing them by the total number of scores

Median - middle value of a data set. Calculated by putting the data in order and finding the middle score. If there is an even number, the two middle numbers are added together and divided by 2

Mode - most common occuring score. Easiest to calculate as the data is put into order and it is easy to see which one occurs the most

16 of 32

Measures of Dispersion

Range - subtracts the lowest score from the highest

Standard Deviation - tells us how far the scores are scattered around the mean

Big standard deviation - many of the data points are far away from the mean

Small standard deviation - data was closely clustered around the mean

17 of 32

Correlational Analysis

Investigates strength and direction of a relationship between two variables

Correlation Coefficient measure the strength e.g. +1 means a perfect positive positive

A correlation coefficient of 0 means that there is no correlation between the two variables

Good for establishing a relationship and the strength of the relationship so trends can be established

Bad as cause and effect cannot be established and the strength can be hard to interpret potentially

18 of 32

Content Analysis

Qualitative data cannot be analysed easily using measures of central tendency and dispersion

Aims to identify patterns and trends in the topics

Step 1 - Sampling the data e.g. picking every 5th name on the register for a book scrunity

Step 2 - Coding the data e.g. picking out criteria to meet like names on files and progress planners

Step 3 - Representing the data e.g. describing the instance in each category

Good for being true to life and representative so high ecological validity, less likely to be investigator effects as they are not there to influence the participants, easy to carry out and cheap

Bad for having researcher bias as one observer carrying it out may twist the data as it is not always objective, extremely time consuming and only describes the data and does not explain

19 of 32

Thematic Analysis

In order to identify themes so qualitative data can be organised and conclusions can be made

Step 1 - read every transcript carefully as all items are included. No notes should be taken and their persepctive should try to be understood

Step 2 - break up the data up in to smaller units. Each unit has a code and may have more than one

Step 3 - combine the smaller codes into larger themes

Step 4 - identify the most common themes. General conclusions can be drawn and made into summarised tables and graphs

20 of 32

Features of a Science

Science, which comes from "knowledge" in latin, is the systematic approach to learning more

Empirical Methods - when information is gained through observation or experiment rather than beliefs so its meaningful and not made up

Objectivity - not being affected by the expectations of the researcher and carefully controlling conditions. More objectivity makes it more measurable and eliminates bias

Replicability - repeating it and getting the same outcome makes the original results more valid. Important that procedures are recorded carefully so it may be followed again

Theory Construction - facts make sense when they are explained with a theory

Hypothesis Testing - theories can be altered through testing the validity and if support is not found then the theory needs to be modified

21 of 32

Inductive Method

Induction is reasoning from the partciular to the general

Starts with evidence which is then used to generate a theory

1. Observations

2. Testable hypothesis

3. Conduct a study to test the hypothesis

4. Draw conclusions

5. Propose theory

E.g. Pavlov observed dogs that salivated and then generated his classical conditioning theory

22 of 32

Deductive Method

Deduction is reasoning from the general to the particular

Starts with a theory then looks for evidence to support it

1. Observations

2. Propose theory

3. Testable hypothesis

4. Conduct a study to test the hypothesis

5. Draw conclusions

E.g. Darwin formulated his theory and then tested it through collection of evidence like fossils

23 of 32

Paradigm

A paradigm is a shared set of beliefs or assumptions about a aprticular subject matter

A paradigm shift is the changing in beliefs about a subject e.g. when astromers suggested that the Earth is not the centre of the universe

24 of 32

Probability and Significance

Probability is a numeircal meausre of likelihood or chance that a certain event will happen e.g. the change of rolling an odd number on a dice is 50%

Significance is a statistical term used to indicate whether the research findings are sufficiently strong enough to rejct the null hypothesis and accept the experimental hypothesis

In psychology the accepted level of signifance is usally p≤0.05

25 of 32

Type I and II Errors

A type I error is an error which occurs when a researcher rejects a null hypothesis which is true

More likely to occur if the significant level is too lenient e.g. 0.1 not 0.05

Saying it is signifcant when there is actually no difference

A type II error occurs when a researcher accepts a null hypothesis which was not true

More likely to occur if the significant level is too stringent e.g. 0.01 not 0.05

Saying it is not significant when there is actually a difference

26 of 32

Levels of Measurement

Nominal data is represented in the form of categories e.g. how many males and females there are in a group

Ordinal data is ordered in some way e.g. asking everyone to rate how much they like something from 1 to 10

Interval data is based on numerical scales that include units of equal size e.g. measuring temperature or timing with a stop watch

27 of 32

Statistical Testing Grid

NAME

TESTING FOR

DATA

DESIGN

Mann-Whitney

Difference

Ordinal

Unrelated

Wilcoxon

Difference

Ordinal

Related

Unrelated t-test

Difference

Interval

Unrelated

Related t-test

Difference

Interval

Related

Sign Test

Difference

Nominal

Related

Spearman’s Rho

Correlation

Ordinal

Pearson’s R

Correlation

Interval

Chi-squared

Association

Nominal

Unrelated

28 of 32

Mann-Whitney

Testing for significant difference

Ordinal data

Using unrelated design

Calculated value must be equal to or less than the critical value in the table

29 of 32

Wilcoxon

Testing for significant difference

Ordinal data

Using related design

Calculated value must be equal to or less than the critical value in the table

30 of 32

Unrelated T-test

Looking for a difference

Interval data

Unrelated design

Calculated value must be equal to or more than the critical value

31 of 32

Related T-test

Looking for a difference

Interval data 

Related design

Calculated value must be equal to or more than the value from the table

32 of 32

Comments

No comments have yet been made

Similar Psychology resources:

See all Psychology resources »See all Research methods and techniques resources »