PY3101 WEEK 1- 5 - Lecture notes 1-5 PDF

Preview

Summary

LECTURE NOTES ...

Description

PSYCHOLOGICAL RESEARCH METHODS AND INTERPRETATIONS

JAMES COOK UNIVERSITY [email protected]

Psychological Research Methods and Interpretations

January 1, 2020

Table of Contents Lecture 1: Subject Overview.................................................................................................................2 Lecture 2: None Experiments and Single-Subject Design......................................................................6 Lecture 3 - Questionnaire Design.........................................................................................................19

1

Psychological Research Methods and Interpretations

January 1, 2020

Lecture 1: Subject Overview

Today’s Recommended Readings -

Research Methods in Psychology e-book o Chapter 2: overview of the statistic method o Chapter 5: Experimental Research o Chapter 12: Descriptive statistics o Chapter 13: Inferential Statistics

Weekly Breakdowns Lectures 1- 5: focus mostly on research methods Lecture 6 – 13: focus mostly on data analysis using SPSS

What is the point of Research? What is Research? Research is the process of gathering information (data) to advance knowledge. It can include: 1. A person reading a guidebook before a holiday so they know what activities to do 2. A marketing company surveying customers to see if they like a new advertising campaign 3. A psychological scientist performing memory studies to understand how memory works

2

Psychological Research Methods and Interpretations

January 1, 2020

The Typical scientific research process The Outcomes of Psychological Research Scientific research typically has one of the following three aims: 1. To describe a phenomenon 2. To predict a phenomenon 3. To explain a phenomenon Most research actually does one or two, but some does all three I will explain each with a forensic psychology example (the impact of race on trial verdicts O.J. Simpson was famously found not guilty of murder. Did his race and the jurors’ race matter?

Outcome 1: Describing a Phenomenon -

-

The most basic psychological research simply describes phenomenon Anwar et al. (2012) found all White juries in the US are more likely to find Black defendants guilty than White defendants There is no explanation here as to why this happens – it is purely descriptive

Outcome 2: Predicting a Phenomenon -

-

More sophisticated psychological research enables us to predict phenomenon Anwar et al. (2012) also found adding one Black juror to an all-White jury made jurors equally likely to find Black and White defendants guilty This research allows us to predict the likely verdict in a trial based on the jury’s composition

Outcome 3: Explaining a Phenomenon -

The best psychological research explains why phenomenon occur. Doing this is not easy. Anwar et al. (2012) suggest mixed-race juries are less biased against Black people as they are more likely to discuss issues such as racial profiling. Their suggestion is an untested theory. How could you use research to find out if their untested theory is correct? 3

Psychological Research Methods and Interpretations

January 1, 2020

What are Theories? - On the right are the annual percentage of reported assaults, rapes, and murders in the Northern Hemisphere - All three peak mid-year: o Why is this phenomenon happening? o How many possible explanations (theories) can you come up with?

Testing the Truth of Theories Theories are likely explanations for phenomenon. They are not facts. They are best guesses, but…. 1. They should be supported by evidence 2. They should make predictions about phenomenon (hypotheses) A theory’s truth can be determined by testing these predictions/hypotheses with research You theorise that violent crimes increase in the summer months due to heat This leads to the prediction/hypothesis that heat makes people more aggressive You could design a study to test this prediction/hypothesis. If heat increased aggression, this would support your theory Types of Research: Goals and Settings Different Goals and Settings Psychological research can be Basic or Applied  Basic and Applied research have very different goals (explained next) Basic and Applied research can take place in the laboratory, field, or both  Laboratory and field research each have their own strengths and weaknesses The Goal of Basic (or Pure) Research

4

Psychological Research Methods and Interpretations   

January 1, 2020

It aims to increase our understanding of why a phenomenon occurs (e.g., why do violent crimes increase during summer months?) It focusses on testing theories to see if they can explain a phenomenon (if not, the theories are adjusted/replaced) It does not to aim to solve real world problems BUT theories arising from basic research often inform solutions to real world problems

The Goal of Applied Research  It aims to solve a real-world practical problem  The research is often informed by theory from basic research and can provide a useful realworld test of existing theories  In my study I discuss how jurors forget trial evidence (a problem). I examined if note taking during a trial improves recall of the evidence. A theory existed suggesting it would. I found it did. I recommend courts allow jurors to take notes to reduce the problem of forgetting What is Laboratory Research? 

Participants are studied in a controlled setting that is contrived specifically for the research (not their everyday environment)

What is Field Research? 

Participants are studied in their everyday environment (e.g., school, the workplace)



It is possible to study the exact same issue in both laboratory and field settings…..

Laboratory and Field Combined 

On the right are two studies by the same authors examining the own-race bias in face recognition



Laboratory study: Black and White participants studied pictures of own/other race faces in a laboratory and later had to recognise the faces



Field study: Black and White people in a shopping centre were briefly approached by a Black or White stranger and later had to recognise this stranger

5

Psychological Research Methods and Interpretations

January 1, 2020

Research Designs in Psychology Three Categories of Research Basic or applied research AND laboratory or field research can fit into any of these categories: 1. Experiments 2. Quasi-experiments 3. Nonexperiments They differ in terms of: 1. Whether or not there is a control condition and experimental condition 2. Whether or not participants are randomly assigned to a condition What are Experiments? They always have: 1. An experimental condition that receives a manipulation/treatment 2. A control condition that does not receive the manipulation/treatment 3. Random assignment to either condition Performance in each condition is compared What are Experiments? In an experiment: • The of the The being If the experimental the IV likely caused •

Experiment In the example on the right there are: 1. Experimental and control conditions 2. Random assignment to the conditions The IV differentiates the two conditions: • Information received about the actor The DV is what was being measured: • The diagnoses (psychoses or not)

Independent Variable (IV) is the name manipulation/treatment differentiating the conditions Dependent Variable (DV) is what is measured/assessed and control condition differ on the DV, this Example

Between and Within-Subjects Designs Experiments can be: Between-subjects (multiple groups)  Participants randomly assigned to control or experimental conditions Within-subjects (one group)  Same participants tested in all conditions. Testing order randomised

6

Psychological Research Methods and Interpretations

January 1, 2020

What are Quasi-Experiments? They have: 1. An experimental condition 2. A control condition 3. No random assignment  Between-subjects: No random assignment to different conditions  Within-subjects : No randomising/ counterbalancing of testing order Quasi-Experiment Example In the example on the right there are: 1. Experimental and control conditions 2. No random assignment (it is impossible) The IV differentiates the conditions: • Smoking history The DV is what was being measured: • Recognition memory scores

What are Nonexperiments? They have no control group and no random assignment: Between-subjects example:  Comparing Males and Females’ memory. Random assignment impossible. There is no control group Within-subjects example:  Comparing memory performance in a happy or sad mood but always testing them in a happy mood first (no counterbalancing). There is no neutral mood control condition showing normal performance Nonexperiment Example Correlational studies are also nonexperiments as they look for a relationship between variables 1. No control condition 2. No random assignment possible My study, right, looked for a correlation between levels of creativity and false remembering This subject mostly focusses on nonexperiments

7

Psychological Research Methods and Interpretations

January 1, 2020

Choosing the Right Test The Two Varieties of Statistical Tests Most statistical tests look for one of two things: 1. Statistically significant differences between groups/conditions: 

t-tests, ANOVA, and MANOVA, etc

2. Statistically significant relationships/associations between variables: 

correlation, linear regression, and multiple regression, etc

Decision Trees Choosing the correct test to analyse your data can be confusing Decision trees can make the process easier 

The one on the right is from Field’s ‘Discovering Statistics…’ textbook



We will look at each step next

8

The Decision Tree Steps The test you use is determined by up to seven factors: 1. The number of DV’s 2. The type of DV’s (continuous or categorical) 3. The number of IV’s 4. The type of IV’s (continuous or categorical): a) If categorical, how may conditions? b) If categorical, is the design between-subjects, within-subjects, or mixed-subjects? 5. Whether or not statistical assumptions are met An Example Study Evidence suggests students who sit near the front of lectures (even if randomly assigned seats) achieve better grades* I want to replicate this, so randomly assign: 

18 students to the front rows



18 students to the back rows

I later compare their final exam scores Step 1: How many DV’s? Studies can have multiple DV’s: One DV example: The impact of seating position (IV) on exam performance (DV) Three DV’s example: The impact of seating position (IV) on exam performance (DV1), enjoyment of lectures (DV2), and self-reported mind-wandering (DV3)

Step 2: What type of DV’s? DV’s have continuous or categorical data Continuous Data: The data is numerical: 

Exam performance as a % (0 – 100%)

Categorical Data: Data falls into discrete groups/categories: 

Exam performance as pass or fail or as a grade (F, P, C, D, HD)

Psychological Research Methods and Interpretations Step 3: How many IV’s? 

Studies also vary in the number of IV’s:



One IV example: The impact of seating position (IV) on exam performance (DV)



Three IV example: The impact of seating position (IV1), attendance at lectures (IV2), and self-reported hours spent revising (IV3) on exam performance (DV)

Step 4: What type of IV’s? They can be categorical or continuous Categorical IV’s: Participants in groups or conditions (between/within-subjects designs)  Lecture seating position (front or back) Continuous IV’s: These are on a scale:  Number of lectures attended (0 – 12)  Self-reported hours revising (0 - ?) Step 5: Statistical Assumptions Many statistical tests can only be used if assumptions about the data are met. A common assumption is: 

Normality: Data has a normal enough distribution (it is never perfectly normal)

If a test is used/the data did not meet the assumptions for it, results can be incorrect

Decision Trees In my study, I decide to have: 

One continuous DV (exam % score)



One between-subjects categorical IV (seating position: front or back)

The decision tree recommends I run a between-subjects t-test on the data (if the data meets the assumptions for it)

Review of Statistical Concepts

January 1, 2020

Psychological Research Methods and Interpretations

January 1, 2020

Two Types of Statistics Part my data is pictured. Two types of statistics will be used to analyse my data: Descriptive Statistics: These summarise data 

Includes the Mean and Standard Deviation

Inferential Statistics: These test for significant differences/associations 

Includes the t-test, ANOVA, regression

Descriptive Statistics Looking at (part of) the data, which group performed best? You can’t tell from this Calculating each group’s Mean (average) exam grade tells us students at the front scored higher: We want to know if this difference is statistically significant or not Inferential Statistical Tests Before analysis, an Alternative Hypothesis (H1) should have been formed: 

H1: Students who sit at the front of my lectures will obtain higher exam scores than those who sit at the back

This is contrasted to a Null Hypothesis (H0): 

H0: Students who sit at the front and back of my lectures will have equivalent exam scores

How Do Inferential Statistical Tests Work? 1. The test starts off assuming The Null Hypothesis is true (e.g., no differences between conditions) 2. The test tells you how much evidence there is in your data that the Null Hypothesis is incorrect (this is shown in the test statistic) 3. The test also tells you how likely it is you would obtain this much evidence (or more) against the Null Hypothesis, if the Null Hypothesis were actually correct (this is the probability or pvalue) 4. If it is highly unlikely you would obtain this much evidence against the Null Hypothesis if it were correct (a small probability), you can reject the Null Hypothesis Let’s examine this in relation to our study…. How Do Inferential Statistical Tests Work? The test assumes the Null Hypothesis is true (i.e., students at the front and back of the lecture will have similar exam scores)

2

Psychological Research Methods and Interpretations

January 1, 2020

The test statistic shows how much evidence there is against the Null Hypothesis (i.e., did students who sit at the front consistently score higher on the exam?)

The larger the test statistic, the more the data contradicts the null hypotheses. My t-test statistic is 2.21. This information alone is not very informative. Something else is needed….. How Do Inferential Statistical Tests Work? Each statistical test also produces a probability (or p-value) that ranges from 0 to 1:

This tells you how likely it is you would obtain a test statistic so large/so much evidence of differences between the groups if no differences actually existed My p-value is .034, so there is a very small probability we would obtain a test statistic so large/so much evidence of differences between our groups if no differences actually existed How Do Inferential Statistical Tests Work? In Psychology, we accept that any p-value below .05 (p...