Experimental Research 2024 PDF

Experimental research Ivan Buljan Assistant professor  Lecture overview 0 1 Laboratory experiments Essentials for true experiments Independent and dependent groups More complex experimental...

Experimental research Ivan Buljan Assistant professor  Lecture overview 0 1 Laboratory experiments Essentials for true experiments Independent and dependent groups More complex experimental designs  Expected outcomes 0 2 Understand importance and concepts related with laboratory experiment Differ between true and quasi experiments Define internal and external validity Differ between within subjects and between subjects design  Laboratory experiment 0 3 Main advantage of laboratory experiments: researcher can control the conditions Main disadvantage of laboratory experiments: the conditions can be too artificial and distort generalization Internal vs external validity Experiment is the principal study design with which we can determine causality!  Variables experiments 0 4 Independent variable: it has an effect on the dependent variable (in experiment, it is usually the one under control of the researcher) Dependent variable: The variable under influence of the independent variable, often represents the thing we try to measure  Laboratory experiment 0 5 Basic distinction: INDEPENDENT GROUPS: different participants in different conditions DEPENDENT GROUPS: same groups of participants in different conditions  Laboratory experiment 0 6 Examples: https://www.youtube.com/watch?v=GjHFvOzNiCo&ab_channel=AndreaEllinger https://www.youtube.com/watch?v=_4mevfKefs4&ab_channel=DriveSafeTownsville  Essentials for TRUE experiments 0 7 Experimental manipulation Standardization of procedures Random assignment  Experimental manipulation 0 8 Important concepts Independent variable Dependent variable Experimental manipulation defines independent variable Manipulated variable is believed to have effect on the dependent variable  Experimental manipulation 0 9 Manipulation must have at least two different levels (at least one experimental and one control group) Internal validity: When manipulating IV researcher must make sure that there are no confounding factors Recommended: do the checking of intervention before the research begins High internal validity can decrease External validity: The degree on which the results of the study are valid for other people, situations and settings (the conditions are sometimes too artificial).  Standardization of procedures 1 0 The only difference between control group and experimental should be in experimental manipulation Interaction should be standardized across groups and participants Standardization of potentially relevant conditions: time of the day, temperature, etc. IMPORTANT: HOWITT mentions that you should decrease variation in order to obtain statistically significant results. This is wrong! You should control conditions in order to increase probability to get correct results. Systematic biases can lead to distortion of results  Randomization 1 1 Primarily used to allocate participants in experimental and control group in unrelated designs In ideal research, randomization helps to eliminate all biases except PUBLICATION BIAS! Participants can be sometimes matched by relevant characteristics and randomly allocated to conditions as a pair (in those cases unrelated design becomes related one) Randomization can be done by throwing a coin, a dice or using computer programs (recommended: https://www.randomizer.org/) R can also be used for randomization  A neurofeedback video game (MindLight) to prevent anxiety in children: A 1 randomized controlled trial 2 (Schoneveld, 2019) Background: Childhood anxiety is a global mental health concern. Interventions are needed that are effective, but also cost less, are more accessible and engage children long enough to build emotional resilience skills through practice. Methods: The present randomized controlled study aimed to examine the prevention effects of a neurofeedback video game, MindLight, developed based on evidence-baed practices with anxious youth. Over 750 children (7-13 years old) in elementary schools were screened for elevated anxiety; 136 selected children were randomly assigned to play Mindlight or a control game. Self- and parent-reported anxiety was assessed at pre-, post-intervention and 3-month follow up. Results/conclusions: Intent-to-treat analyses revealed an overall significant reduction in child- and parent-reported anxiety, but the magnitude of improvements did not differ between conditions. Future research comparing MindLight to cognitive-behavioral interventions is suggested, as well as testing a range of specific (e.g., exposure) and non-specific (e.g., expectations, motivation) therapeutic factors as mediators of outcomes.  Within subjects designs 1 3 Same participants in all situations (conditions) We control for lots of differences between participants (since it is common that differences are greater between different people than within ourselves in different situations) Potential problems in within subjects designs: Fatigue or boredom Practice Carryover effects (especially if the intervals are short)  More advanced research designs 1 (experimental) 4 Key concepts: Pre test: to determine basic state and whether there are differences between groups Participants can become alerted on the purpose of the experiment and influence their behaviors (lack of external validity) Post-test: to determine the effects of the intervention More complex approach: Solomon’s four group design Solomon’s four group design 1 5 It is very difficult to control the experiment of this size and complexity, and to collect as many participants.  Pre-test implementation 1 6 If implemented, should be implemented in all groups It is recommended that pre-test is disguised (embedded in another task) The interval between the pre- and post test should be sufficiently big (the bigger the better) so the participants forget about the pretest In some cases, when the intervention effects are short lived, it is possible to do pre-test AFTER the post test (e.g. the effects of alcohol on making the errors in the test can be first tested under influence of alcohol, and AFTER when the effects have worn out, to avoid sensitization)  Deception 1 8 Intentionally giving false information to subjects or intentionally misleads them about some key aspect of the research. Frequent in experimental studies in psychology Following conditions must be met: No other nondeceptive method exists to study the phenomenon of interest The study makes significant contributions to scientific knowledge The deception is not expected to cause significant harm or severe emotional distress to research participants The deception is explained to participants as soon as the study protocol permits Boynton MH, Portnoy DB, Johnson BT. Exploring the ethics and psychological impact of deception in psychological research. IRB. 2013 Mar-Apr;35(2):7-13. PMID: 23672145; PMCID: PMC4502434.  Debriefing 1 9 The process of giving participants in a completed research project a fuller explanation of the study in which they participated than was possible before or during the research (APA dictionary). Must be done if the deception has been performed.  Ethical aspects of conducting an 1 experiment 7 Arrive to study session early (before your participants) and prepare all materials Treat people with courtesy and respect; dress nicely Start session with informed consent form Have a copy of a protocol with you Practice the experiment couple of times before real application Be alert to distress of your participants Prepare (understandable) debriefing Thank participants at the end Thank you for your attention! 2 0 Questions?