bambam

Experimental Psychology
Ad 0:
Try a new drinks recipe site
2005-02-28 17:49:38 (UTC)

Hill 2-28

More design/Conduct Threats

-Testing-effects on DV produced by previously measuring it
-sometimes learn or remember how responded; result of
reactivity
-realize what is being studied
-Reduce by test-retest reliability, control group
-problematic for w/i Ss
-Instrumentation- Some feature of msmt of DV changes
during study
-E.g., msmt instruments are different, human observers
different manner of Ss responding character changes,
ratings shift, experimenter bias
-Regression to the mean-can occur when S groups based on
"extreme" scores on measure
-Extreme scores are less reliable b/c tend to move toward
mean on retesting (e.g., Iq groups, height of children,
abilitys)

Participant/Subject problems

-Selection-groups different to start with, due to nonrandom
assignment, could influence DV
-E.g., volunteers for b/w Ss, not w/i Ss
-Problamatic for b/w Ss not w/i Ss
-Attrition-drop outs unequally distributed
-may cluster in a treatment condition or grp
-must be reported - by treatment, #s, reasons,
demographics
-problamatic for w/i Ss & longitudinal
-Selection interaction-w/another threat or a "family" of
threats
-Selection problem can combine w/ another threat to
influence DV for some treatments & not for others
-E.g., selection X history; selection x attrition;
selection x testing; selection x maturation

External Validity -An Introduction

-Going beyond the experiment
-Do findings generalize for the IV
-Not an "either/or" issue, but a "continuum"
-some exp't more, some less
-2 requirements to be met:
-internaly valid (sets limit on external validity)
-Findings can be replicated

Applictains of External Validity

-Across Subject (to other participants & pops)
-Want to infer from sample to pop to other pops
-need truly random samples-volunteer problem
-Problems if only M or F (e.g.,Kohlberg, Freud)


Ad:0