Home > Preview
The flashcards below were created by user
Anonymous
on FreezingBlue Flashcards.

texts
any kind of communication that can be documented

Cohen's Kappa
 K = #agree/#possible  1/n
 _______________________
11/n

minimum acceptable interrelater reliability
>.70

what causes low interrelater reliability?
 1. definition problems
 2. coding rules
 3. coder drift

6 types of texts
 1. public/private
 2. mediated/nonmediated
 3. scripted/unscripted


How do we increase interrater reliability?
 1. good definitions
 2. clearly defined coding rules
 3. test interrelater reliability often

3 criteria for good coding system
 1. mutually exclusive
 2. exhaustive
 3. equivalent

strengths of content analysis
  less biased & more comprehensive than surveys
 unobtrusive
 less expensive to make errors

weaknesses of content analysis
 limited to recorded information
 validity problems
 coding can be too simplistic
 assuming there is a uniform relationship between symbols and meaning

How are experiments unique?
only method that implies causality

What are the requirements for inferring causality?
 1. IV + DV are correlated
 2. IV comes before DV
 3. Rule out alternative explanations

strengths of experiments
  ability to isolate the experimental variable and effect over time
 experiments don't involve large number of subjects and can be replicated many times

weaknesses of experiments
 artificiality
 little likelihood situations will occur in real life

Double  Blind
neither experimenter nor anyone who comes in contact with the participants has knowledge of whether they are in control group or experimental group

Control group
subject to exact same conditions as experimental group, except without the IV

random group assignment
 each participant has an equally likely chance of being in control or experimental group
 gives assurance that each group is equivalent

matching
rather than random assignment, experimenters predict what variables might affect the DV and account for those in the group assignment

pretesting/posttesting
 testing the DV before and after IV is administered
 gives researchers a baseline to compare effects of IV

manipulation check
asks participants after the study if the independent variable actually meant what it was intended to

experimenter effect
experimenter unconsciously treats subjects differently if they know the purpose of the study, affecting how participants behave

observer bias
if researchers/observers know which subjects are in experimental/control group, what they see may be biased

researcher attribute effect
traits of the researcher affect behavior/data of subjects

Hawthorne effect
participant's responses are influenced because they are aware of being observed

Testing effect
participants are primed by the test to respond in a certain way

maturation effect
subjects naturally change over time during the study regardless of the independent variable

experimental mortality effect
participants drop out before the study is complete

selection bias effect
when subjects are allowed to choose their groups, the groups will not be the same to begin with

intersubject bias
participants influence one another's behaviors

compensatory rivalry
one group compensates for missing experimental condition by working harder

demoralization
sense of not progressing causes control group to give up

History effect
a historical effect occurring outside the study affects the outcomes

instrumentation
unreliable scale yields unreliable results

statistical regression
outliers/extremes will naturally test closer to mean on posttest

Maxiconmini
 maxi: maximize IV
 con: control for third/extraneous variables
 mini: minimize error variance

How do researchers minimize the weaknesses of experiments?
  make conditions the same for control & experimental group
 doubleblindness
 random selection
 manipulation check

Solomon 4
 2 control and 2 experimental groups
 1 experimental and 1 control are pretested

alpha level
significance of the probability that the stats are due to sampling error, measures statistical significance


standard deviation
measures dispersion of statistics, higher =more dispersed data

correlation
the extent to which two variables are related, the extent of agreement/consistency, shows reliability of results

ttest
examines the difference between two groups, will two groups with same IV vary on DV?

