Confounding variables
also referred to as the third variable (alternative explanations) outside influences that changes the effect of a dependent and independent variable
Environmental Variables
Time of day, temperature, noise levels, colours
Participant variables
IQ, age, gender, education, income, birth order, health
Time related variables
fatigue, practice
What are three types of extraneous variables?
Environmental, Participant and time related
What does control refer to?
controlling the extraneous variables (we must stop extraneous variables from becomming confounding variables
How can we control extraneous variables?
- Holding constant (same time of day, same room,same researcher, same noise, same age, gender)
- Matching across groups (same gender mix in each group, same average age, IQ, counter balancing)
- Randomization (the quick and easy solution?)
is there a perfect solution?
Between subject research design (BSD)
(also known as independent-measures experimental design)
One score per participant
- different participants in each group
- thus the scores are not dependent on the same participant thus independent-measures design
Between subject design advantages (BSD)
scores are independent so there is no need to worry about
- practice effects
- fatigue effects
- carry over effects
Between subject design disadvantages (BSD)
individual differences (what is these introduce confounding variables?) - what if random assignment to groups has failed thus assignment bias?
Assignment bias
a threat to internal validity when the process assign different participants to different treatments produces groups of individuals with noticeably different characteristics
Differential Attrition (withdrawal)
treatment group may cause more withdraw than the comparison group
Diffusion
When the treatment spreads from one treatment group to the control group (participants talk)
Within-subject experimental design (WSD)
Also known as Dependent-measures experimental design
advantages of within-subject design (WSD)
reduce error in measurement (participants serve as their own control group) -individual differences have been eliminated (IQ, gender
-reducing error variance thus increasing power
disadvantages of within subject design
testing (order) effects such as practice, fatigue (progressive error), sequence, and carry over.
- practice and fatigue come from the experiment in general
- carry over from a specific treatment (treatment A affecting outcome of treatment B) to avoid use counterbalancing
counterbalancing
technique used to deal with order effects when using repeated measures design. participants sample is divided in half (one half completing two conditions in one order and the other completing the conditions in the reverse order)
Quasi Experimental design
No control over group assignment (also called nonequivalent group design) e.g. male vs female.
- there is assignment bias in the study when using this strategy
Examples of quasi-experimental designs
DONT WORRY TOO MUCH ABOUT THESE
1) differential research design (race, gender ect.)
2) postest-only non-equivalent control group design (smokers with treatment vs non-smokers with no treatment)
3) pretest-postest nonequivalent control group design
pretest-postest nonequivalent control group design
this design is usually referred to as a quasi-experimental design
- pretest observations (O) limit somewhat the threat due to assignment bias
- the researcher can compare the groups before treatment (X)
Developmental research designs
two types
Cross-sectional research design (one observation)(cohort effect)
Longitudinal research design (multiple observations) (no cohort effect)
Cohort effect
particular impact of a group bonded by time or common life experience.
correlation
demonstrates a relationship (between a and b for example)
- no control or manipulation of the variables
- cannot talk about cause and effect
- no IV DV
- all variables are equal
Advantages of correlation
High external validity (measures what is there, no manipulation or artificial settings)
