week 8, validity & non-experiments I

Question 1

what is a true experiment?

Answer 1

1. population -> sample -> experimental -> measure (observation)
2. population -> sample -> control -> measure (observation)

Question 2

what characteristics does a well designed experiment have?

Answer 2

• like a trap, almost impossible to derive conclusions drawn from thw results
• variability in DV is clearly linked to the changes in the IV

Question 3

what characteristics does a poorly designed experiment have? What does it lead to?

Answer 3

ambiguous, unsure if the change happened because of a confound or the IV
this leads to….
1. type I error -> false positive
3. type II error> false negative

Question 4

what is high internal validity?

Answer 4

variability in DV attributed to changes cause in IV

Question 5

what is low internal validity

Answer 5

ambiguity in source of variability in DV (a confound exists)
high chance of type I and II errors

Question 6

how do you ensure that there is high internal validity in experimental research?

Answer 6

whether an observed covariation between x and y truly reflects a causal relationship from x to y

Question 7

what are threats to internal validity?

Answer 7

1. placebo effect
   2.hawthorne effect
   3.experimenter (pygmalion effect)

Question 7

how do true experiments maximize internal validity (4)

Answer 7

• random assignment
• control for extraneous variables
  -measure DV
  -differences in DV are due to the IV

Question 8

what are non-experimental designs & quasi experiments

Answer 8

no manipulation on the IV (correlational study)

IV is “manipulated” but without complete random assignment of participation to conditions (quasi-experiment)

Question 9

what are 3 non-experimental designs

Answer 9

1. one group posttest-only design
   2.posttest-only design with non-equivalent groups
   3.one-groups pretest-posttest design

Question 10

one group posttest-only design definition and issue

Answer 10

a single group is given treatment, outcome is measures afterward
- no pretest or control group to compare results

Question 11

posttest only design with non-equivalent groups

Answer 11

two groups (1 gets treatment, other doesnt), but no random assignment

-preexisting differences between groups can affect the outcome

Question 12

one-group pretest-posttest design

Answer 12

a single groups is measured before and after treatment
- no control group

Question 13

what are some threats to internal validity

Answer 13

history-> an external event occurs at the same time as the IV and may affect measurement

maturation -> participants change over time

testing or practice effects -> participants sensitive to repeated measurement

selection bias -> ppl who choose to be in study differ from non-participants in non trivial ways

regression to the mean -> statistical tendency for extreme scores to become more moderate when you retest

mortality -> participants die or drop out of study