Research Methods and Statistics 1 Público

Research Methods and Statistics 1

K Pedroso
Curso por K Pedroso, actualizado hace más de 1 año Colaboradores

Descripción

Research methods

Información de los módulos

Sin etiquetas
Descriptive Statistics - Used to describe or summarise raw data Inferential statistics - Used to infer effects in data, inferences are made by performing a statistical test to test the null hypothesis with a p-value Measures of Central Tendency - measuring the average 1. Mean - sum of all scores divided by the number of all scores Pros: Makes use of all scores in a data set, so is most representative Cons: Extreme scores (outliers) can affect the me   an 2. Median - middle score Pros: Not affected by extreme scores/outliers Cons: Not very representative if there are large gaps between scores 3. Mode - most common score Pros: Directly represents the most common score Cons: May not represent the range of scores in the dataset   Measures of Dispersion - measures the distance from the average   1. Range - difference between the min and max scores Pros: Easy to calculate and understand Cons: Only takes into account the extremes of the scores, not the rest of the data or the mean 2. Variance - spread of scores around the mean/the average of how much each score varies from the mean Pros: Uses all of the data Cons: Not in an easily interpretable format (original units squared) 3. Standard deviation - spread of scores around the mean, expressed in the original units of the data (square root of the variance)  Pros: Expressed in the original units of the data, so more easy to interpret Cons: Time consuming to calculate manually (using SPSS will make it easier)     Normal vs Non-normal distribution Normal distribution - Data can be plotted on a histogram - Distribution of many variables follow a bell-shaped curve - largest frequencies of scores in the middle, extremes high or low on either side   Skewnesss score - measures how skewed the data is, score must fall within range of -1 to +1 Kurtosis score - measures kurtosis in the data, score must fall within range of -2 to +2 If both skewness and kurtosis scores fall within the range, the data is considered normally distributed.     Types of Variable Categorical variables - Variables are categories - Need to 'code' each category as a number e.g. Gender 1=Male 2=Female - Can't compute mean, standard deviation - can only check frequencies   Continuous vairables - variables are a continuum - Vairables are already in number form e.g. Age 1, 2, 3, 4, 5 - Can compute mean, standard deviation, etc.   Levels of Measurement Nominal - Categories with no order (nationality, gender, etc.) - Categorical Ordinal - Ranked data, so has an order but not equal gaps (education lvl) - Categorical Interval - Equal gaps between points, but no true zero (temperature - farenheit) - Continuous Ratio - Equal gaps between points and true zero, e.g. Reaction time - Continuous   Statistical tests Parametric data - normally distributed and measured at interval or ratio level - use a Parametric Inferential test   Non-parametric data - non-normally distributed and/or measured at nominal or ordinal level - use a non-parametric statistical test
Mostrar menos
Experimental research - requires exerting control - Researches assign subjects to groups and manipulate a variable   Non-experimental research - allows for much less control - Researchers don't assign subjects to groups or manipulate variables   Variables - variables are characteristics that differ from participant to participant, or from condition to condition Variables/Factors That Need To Be Controlled Extraneous variables can influence the data, but have the same effect in each group/condition Demand characteristics – environmental factors that indicate the purpose of the experiment Experimenter effects – Unintentional behaviours by the experimenter that affects participants’ behaviour Participant variables – Participants factors that affect the findings (e.g. prior knowledge) Situational variables – Environmental effects (e.g. noise, time of day, lighting, etc.) Confounding variables can influence the data, but have different effects in each group/condition They can therefore be the cause of the effect instead of the manipulated variable e.g. Level of exercise (IV) on weight gain (DV) – diet could be a confounding variable Between-subjects design Participants are assigned to different groups of the IV Each participant contributes one score of the DV Advantage: avoids carry-over effects Disadvantage: Could be affected by individual participant variability   Within-subjects design Participants in all conditions of the IV Each participant contributes multiple scores of the DV Advantage: Individual participant variability controlled Disadvantage: Potential for carry over effects   Independent Variables - The variable that is being manipulated or changed    Dependent Variables - the variable we expect to be affected by the IV, so the DV is what we measure Hypothesis testing The p-value tells us the actual probability of our finding being due to chance If p ≥ .05 we fail to reject the null hypothesis This would mean our hypothesis is not supported If p ≤ .05 we can reject the null hypothesis This would mean our hypothesis is supported   T-test -a type of inferential statistical test - it can be used on experimental, or quasi-experimental data   Null hypothesis : the IV will have no effect on the DV Experimental hypothesis: The IV will have an effect on the DV   Uni-directional: IV wil increase/descrease DV Bi-directional: The IV will have an effect on DV
Mostrar menos
To report the results, three statistics from the second SPSS output table must be cited:   1. The test statistics t in the table 2. The degrees of freedom df in the table 3. The p value Sig. (2-tailed) in the table   In this case: t = 6.826, df = 17, and Sig. (2-tailed) = 0.000   The format for reporting these statistics is t(17) = 6.826, p < 0.001   The value of t should always be given to two decimal places and the value of p can be given to either two or three decimal places. The leading zero can be omitted but some people prefer not to. When the p appears as ’0.000’ in the table – report as p < 0.001, because p can never truly be zero. Report as:
Mostrar menos
Code of Ethics and Conduct   1. Respect Respect for the dignity of persons and peoples is one of the most fundamental and unversal ethical principles across geographical and cultural oundaries and across professional disciplines   Psychologists should always consider:  - Privacy and Confidentiality - Consent   2. Competence Refers to their ability to provide specific services to a requisite professional standard    Psychologists should always consider: - Possession or otherwise of appropriate skills and care needed to serve persons and peoples; - The limits of their competence and the potential need to refer on to another professional.​ ​​​​​​ 3. Responsibility Psychologists must accept appropriate responsibility for what is within their power, control or management.   In applying these values, Psychologists should consider: • Responsible use of their knowledge and skills; • Respect for the welfare of human, non-humans and the living world;     4. Integrity Acting with integrity includes being honest, truthful, accurate and consistent in one’s actions, words, decisions, methods and outcomes.   In applying these values, Psychologists should consider: • Honesty • Fairness • Addressing misconduct   Definitions of terms • ‘Research’ is defined as any form of disciplined enquiry that aims to contribute to a body of knowledge or theory. • ‘Research ethics’ refers to the moral principles guiding research from its inception through to completion and publication of results. • ‘Research Ethics Committee (REC)’ refers to a multidisciplinary, independent body responsible for reviewing research proposals involving human participants to ensure that their dignity, rights and welfare are protected. • ‘Protocol’ refers to a filed document which specifies for a research project the procedures for recruiting participants and gathering and managing data, with which all project staff agree to comply. • ‘Participant’ It is now common practice to refer to a person who serves as a data source for research as a ‘participant’.   Principles Research that involves humans addresses a wide range of topics and utilizes many different methodologies. The types and severities of risks associated with human research range widely 1. Respect for the autonomy, privacy and dignity of individuals and communities - Means that there is a clear duty to participants - Psychologists respect the knowledge, experience, and expertise of participants and potential participants - They respect individual, cultural and role differences - Researchers will respect the privacy of the individuals, and will ensure that individuals are not personally identifiable   2. Scientific Integrity  - Research should be designed, reviewed and conducted in a way that ensures its quality, integrity and contribution to the development of knowledge and understanding. - Quality relates primarily to the scientific design of the research and the consideration of potential risks of harm and protocols for addressing such difficulties (should they arise)   3. Social Responsibility - A shared collective duty for the welfare of human beings, both within the societies in which psychology researchers live and work, and beyond them, must be acknowledged by those conducting the research. - - The aim of generating psychological knowledge should be to support beneficial outcomes.   4. Maximizing benefit and minimizing harm - Psychology researchers should seek to maximize the benefits of their work at all stages, from inception through to dissemination - Harm to participants must be avoided. Where risks arise as unavoidavle and integral elecment of the research, robust risk assessment and management protocols should be developed and complied with. - Normally, no risk of harm should be greater than that encountered in ordinary life.   Risk Risk can be defined as the potential physical or psychological harm, discomfort or stress to human participants that a research project may generate The following research would normally be considered as involving more than minimal risk: • Research involving vulnerable groups (such as children aged under 16; those lacking capacity; or individuals in a dependent or unequal relationship); • Research involving potentially sensitive topics (such as participants’ sexual behaviour; their legal or political behaviour; their experience of violence; their gender or ethnic status); • Research involving a significant and necessary element of deception; • Research involving access to records of personal or confidential information (including genetic or other biological information); • Research involving access to potentially sensitive data through third parties (such as employee data); • Research that could induce psychological stress, anxiety or humiliation or cause more than minimal pain (e.g. repetitive or prolonged testing); • Research involving invasive interventions (such as the administration of drugs or other substances, vigorous physical exercise or techniques such as hypnosis) that would not usually be encountered during everyday life;   Valid Consent Researchers should ensure that every person from whom data are gathered for the purposes of research consents freely to the process on the basis of adequate information.   Assessment of Risk A prior assessment of potential risks should inform the preparation of the information to be given to potential participants and the procedures for seeking consent. This assessment should be used to determine the appropriate form of consent and the nature of any risk management required. Who can give consent? The consent of participants in research, whatever their age or competence, should always be sought, by means appropriate to their age and competence level. For children under 16 years of age and for other persons where capacity to consent may be impaired the additional consent of parents or those with legal responsibility for the individual should normally also be sought.   Informing participants Giving potential participants sufficient information about the research in an understandable form requires careful drafting of the information sheet. • The aim(s) of the project. • The type(s) of data to be collected. • The method(s) of collecting data. • Confidentiality and anonymity conditions associated with the data. • The time commitment expected from participants. • The right to decline to offer any particular information requested by the researcher.   Confidentiality Subject to the requirements of legislation information obtained from and about a participant during an investigation is confidential unless otherwise agreed in advance.   Deception To many outside the psychology profession, and to some within it, the idea of deceiving the participants in research is seen as quite inappropriate. Since there are very many psychological processes that are modifiable by individuals if they are aware that they are being studied, the statement of the research focus in advance of the collection of data would make much psychological research impossible.​​​​​​​   Double blind study Neither the participants nor the experimenters know who is receiving a particular treatment, This is utilised to prevent bias in research results
Mostrar menos
Correlations Correlational research involves observing what naturally goes on in the world without directly interfering with it. T-tests examine the mean differences between levels of one Independent Variable (IV) on a Dependent Variable (DV)   Relationship - how one variable changes as a function of another variable   We cannot establish a cause-effect relationship because there could e a third variable which is influencing both our variables   Pearson's Correlation test - Two variables - Must be parametric (continuous, normally distributed) - Use Spearman's rank correlation test if data are not parametric   Spearman's rank correlation - Pearson's correlation coefficient - We can measure the strength and direction of a linear relationship between two variables using a correlation coeffricient (r) - Correlation coefficient lies between -1 and +1  Direction of a linear relationship r=+1 indicates a perfect positive relationship  r=-1 indicates a perfect negative relationship r=0 indicates no relationship    Predictor variable - A variable thought to predict an outcome variable Outcome variable -A variable thought to change as a function of changes in a predictor variable   - Scatterplot is used    Questionnaires - a research tool for data collection - a list of written questions - can be effective means of measuring the behaviour, attitudes, preferences, opinions and intentions  - self-administered or interview - close-ended or open-ended questions Strengths - large amount of information can be collected - target large number of people - relatively easy to get information from people quickly - relatively low cost in time and money - can be used to explore embarrasing areas more easily than other methods if done in anonymous Limitations - difficult to obtain a good response rate - lack of confidence in results - question wording can have a major effect on answers - misunderstandings cannot be corrected - difficult to account for cultural and language differences   Stages in Questionnaire design Define the Research aims Identify the population and sample Decide how to collect replies (mode of administration: self-administered or interview) Design the questionnaire (question types; question wording) Pilot test the questionnaire to identify the potential problems Administer the questionnaire Analyse the data Question wording - Be concise - Avoid ambiguous words - Avoid double-barreled questions - Avoid double negative questions -Avoid leading questions   Likert scale - Most common scale to measure attitudes  - Ordinal scale measures level of agreement
Mostrar menos
What is regression? Regression is the degree to which one or more predictor variables can predict the value of an outcome variable Outcome variable: what you are trying to predict Predictor variable: the variable you use to try to predict the outcome variable   Criteria for Simple Regression: Two variables (predictor and outcome) Both must be parametric - Predictor and outcome must be continuous (interval or ratio) Predictor and outcome must be normally distributed   Variance is a measure of how far observed values differ from the average of predicted values i.e., their difference from the predicted value mean Explained variance/explained variation - is used to measure the discrepancy between a model and actual data; it is the part of the model's total variance that is explained by factors that are actually present   Higher percentages of explained variance indicates a stronger strength of association, and means that you make better predictions   Interpreting Linear Regression Anova tells us whether our regression model explains a statistically significant proportion of variance (F (1,18) = 39.1, p < .001)   The regreesion coefficient (B) - tells us how much the outcome variable is expected to increase or decresse when the predictor variable increases by 1 unit.  (B = .69, t(18) = 6.25, p < .001)   Coefficient of determination (R2 ) • Regression tells us how good the equation is at predicting the Outcome • This is represented by ‘R2 • Minimum R2 is 0, maximum is 1 (or 0% and 100%) • The closer R2 is to 1, the better the equation is at prediction The regression model was statistically significant, and explained 73% of the variance of [outcome variable], R2 = .73, F(1, 8) = 21.28, p = .002. R^2 = p value
Mostrar menos
Mostrar resumen completo Ocultar resumen completo