Experimental methods Flashcards
- Demonstrating cause and effect
An experiment is one of the most powerful ways gathering evidence to support theories.
It is more likely than other methods to demonstrate a relatively unambiguous connection between cause and effect and these connections are what science tries to establish
Alternative explanations
There are always alternative interpretations of findings
We can gather data that supports a hypothesis but very often we cannot rule out competing explanations; that is, we cannot confidently point to a clear cause effect relationship between events and human behaviour.
Explanation 1: heat causes aggresion
But heat is also related to: explanation1: number of people outside.
Explanation 2: lenght of queues (filas)
Any of these could be cause of aggression
The other variables that might have effect on the dependent variable are cofounding variables
An experiment is a type of research in which researcher carefully manipulates a limited number of factors (IVs) and measures the impact on other factors (DVs)
In psychogy, you would be looking at the effect of the experimental change (IV) on a behavoir (DV).
How do experiments support causal statements?
Experiments can meet three causal rules:
- Covariance: Do the variables change together? Are there changes? Variables need to be related.
The studys results show that as A changes, B changes - Temporal precedence: Usually yes, because the IV is manipulates, it comes before the DV.
- Internal validity: The most important! Can alternative explanations be ruled out? Are there selection effects? Are there order effects?
The studys method ensures that there are no plausible alternative explanations for the change in B; A is the only thing that changed.
GROUPS
Experimental group
The one that is being manipulated
Control groups
Act as a BASELINE MEASURE of behaviour without treatment
A control group is not always neceessary
Placebo groups
One group of patients is often given an inert substance (a placebo) so that the patients think thay have had the treatment.
A PLACEBO group is similar to a control group in that they experience exactly the same conditions as the experimental group except for the level of the independent variable that is thought to affect the dependent variable
Critiques to experiments
We can’t always do experiments:
Innapropiate, unethical
Experiments can be highly artificial and can severely limit the range of what can be studied – all variables need to be strictly operationally defined and occur in a specific manner in the experimental setting – be it in the laboratory or the field
Non-experimental methods are better suited to the study of naturally occuring phenomena, such as reactions to parental discipline, gender specific behaviour, large-scale investigation of attitudes and prejudice, everyday health behaviour and so on.
Independent samples design
AKA ‘Between subjects/groups design’.
One group of participants is subject to the experimental condition while an entirely different (independent) group of people participates in the control condition.
Participants are always randomly assigned to experimental conditions
Participants randomly allocated to… experimental group or control group and the procedures for all participants is exactly the same… except one group got the treatment administered and the other got no treatment… and then everything is measured on the dependent variable
Just to clarify two similar-sounding terms…
Random sampling is how you get a pool of research participants that represents the population you’re trying to learn about
Random assignment of participants to control or experimental groups is how you control all variables except the one you’re manipulating.
First you sample, then you sort (assign).´
Independent samples designs
Multilevel designs.
More than two leves of the IV. More real. Nonlinear effects can be discovered
PARTICIPANT VARIABLES
In an independent samples design it would always be difficult to rule out participant differences as a possible source of variation between the results of groups in our study,
Threat to internal validity
STEPS to reduce participant variables
Random assignment
Matching: group 1 (one person with a high IQ, one person with middle IQ, one with low IQ) group 2 (one person…)
REPEATED MEASURES DESIGN
AKA ‘within subjects/groups design’
The same measure is repeated in each participants but under the various conditions of the independent variable
If the participants are the same for both conditions, and all other variables are controlled, any differences within participants should be the effect of the manipulated independent variable.
Individuals serve as their own control
Multiple testing is not repeated measures
ORDER EFFECTS
Effects from the order in which people participate in conditions
Practice may improve performance but also fatigue may reduce performance
How to deal with order effects?
Counterbalancing (balancear el conteo) Having two conditions (A and B), one group does the AB order while the other groups does the BA order.
Complex counterbalancing:
To balance asymmetrical order effect all participant take conditions in the order ABBA.
Score in condition A is taken as the mean of the two A trials and likewise for B.
When there are more than 2 conditions, you divide the participants that many groups as possible orders.
Randomization of condition order
Present the conditions to each participant in a different randomly arranged order.
Randomization of stimulus items
Present items from different conditions
Elapsed time
Leave enough time between conditions for any learning of fatigue effects to dissipate
Using another design
Move to an independent samples design
Independent samples designs
More sample needed, too much variance make the analysis harder, no contamination across independent variable levels
Repeated measures
Order effects, effect of attrition (desgaste), taking both conditions create demand characteristics bias
Practice effect
Need of equivalent stimuli
- FACTORIAL DESIGNS
A study with two independent variables
Each IV has at least two levels.
IVs are crossed with each other, creating all possible combinations of the levels.
IVs are crossed with each other, crating all possible combinations of the levels.
IVs can be participant variables or manipulated variables.
IVs can be within-groups variables or independent-groups variables
