Psychpt7

Three main characteristics of experiments

  1. One variable is manipulated (the independent variable)
  2. Another variable is measured to determine what effect, if any, results from the manipulation (the dependant variable)
  3. Other extraneous factor that might influence results are controlled for at least as much as possible.

    Unlike other methods, experiments allow inferences about cause and effect to be made This is due to the manipulation of independent variable while controlling other factors with the ability to influence the dependant variables. If a researcher can accurately predict something by introducing or removing a variable and everything else is held constant, a causal explanation related to that variable can be very compelling. This is reinforced if you can replicate the results across many experiments If you think you understand a causal relationship between two factors, you should be able to add/remove that behaviour by adding or removing that causal factor. General logic of experimentation

  1. Start with equivalent groups of participants (try not to have one group full of grannies and the other full of toddlers)
  2. Treat them equally in all respects except for the variable that is of particular interest
  3. Isolate this variable and manipulate it
  4. Measure how the groups respond and compare the results.

examples- Research question- does using a cellphone negatively impact driving performance? IV- cell phone use Group 1- cell phone use while driving Group 2- No cell phone use while driving (control group) This as a real-world experiment in unethical, but driving simulators may be used to test this. DV- driving performance Operationalize- One option is to relate it to breaking reaction time (faster breaking would be expected for better overall driving.) Prediction- If cell phone use negative;y impacts driving performance, then those in the cell phone group will have slower reaction times than those in the control group. IV usually have at least two levels (one is usually the control level)

Within-subjects design: the same participant tests all conditions corresponding to a variable (subjects drives with and without cell), and this may preserve the variables since the person would have the same experience, age, and so on but doing one part of the experiment before the other may affect how they react to the next part of the study.

 Between subjects design: different participants are assigned to different conditions corresponding to a variable (one subject drives w/cell phone and another drives w/o a cell phone)

 Random assignment: When using between-subjects design, random assignment does not eliminate differences across groups but attempts to balance them by distributing them randomly across conditions and this helps them minimize the possibility of alternative explanations accounting for the results

If you let folks choose their groups, those w/more practice with distracted driving may bo more likely to choose that condition, which could bias/skew the results

ow ow ow ow ow ow ow

Counterbalancing: involves alternating the order of condition assignment across subjects, reducing the possibility of potentially confounding order effects influencing the results.

Sub 1 does condition 1 first, then condition 2. Sub 2 does condition 2 first, then condition 1. ETC. If using a within-subjects design, it's good to make sure there's not one condition that everyone is going into with more or less practice. hungry

Complex designs can include measuring multiple variables in different conditions (low density with phone, low density w/o phone, high density w/ phone, high density w/o phone) Scringly dingly

Validity: how well an experimental procedure test what it is design to do

Internal validity: degree to which an experiment supports clear causal conclusions(flawed designs have low internal validity)

Common threats to internal validity can be

Confounds- two variables are intertwined so one cannot tell which is the dependant or independent.