Class Notes (1,100,000)
CA (650,000)
UTSG (50,000)
PSY (4,000)
Lecture

Lecture on Research Methodology


Department
Psychology
Course Code
PSY320H1
Professor
Dax Urbszat

This preview shows half of the first page. to view the full 2 pages of the document.
Reliability: true variance/total variance
Test-retest : test is administered twice at two different points in time assuming that theres no
inequality nor change (ie: IQ score should be the same for the same person/ *reliability is the most at
short intervals of time because after a year, significant changes may occur)
Inter-rater reliability : two or more independent judges score the test
Parallel-forms reliability : administer the two different tests that were created using the same content
and then compare the score (ie: version A and version B => if the score is the same, it has a good
quality)
Internal consistency : when a test has the same item more than once (ie: the same question may be
asked in different words, and they should have the same response in order to make sure theres
reliability)
Validity: you cant never get rid of random error; different methods have different systematic errors
Face validity : if a measure appears (on the face of it) to measure what it is supposed to measure (ie:
depression = have questions about isolation, sadness; so that the test looks like whats supposed to
measure -> more direct)
Convergent validity : the degree to which scores on a test correlated with scores on other tests that are
designed to assess the same construct (ie: depression scale: measure the same people on one
depression scale PLUS a new scale => if the score is the same, theres convergent validity)
Discriminant validity : the degree to which scores on a test DO NOT cor relate with scores from other
tests that ARE NOT designed to assess the same construct (ie: measure depression scale => test it
against other scales such as anxiety scale to make sure theres no correlation)
Self report: unstructured questions vs. scaling (=a continuum from very –ve to +ve attitudes)
Bogarduss social distance : measure attitudes towards members of social or ethnic groups that is
based on the assumption that ones liking for a group is reflected in the social distance that one finds
acceptable in relationships with members of other group (willingness to allow certain ethnic groups to
their own group)
Thurstone : many possible scale items are rated by a panel of judges; the group ranges from very
positive to neutral to very negative; ask to state if they agree wit h each statement
Liker t (method of summated ratings ): most common five scale is SA-A-Neutral-DA-SDA; outliers
are possible, so only the items that cor relate well with the total score are kept on the scale
Osgoods semantic differential : include various subscales that measure connotative meaning of the
attitude object; bipolar (good ---- bad) subscales (ie: evaluation (good/bad), potency (strong/weak),
activity (active/passive)
One-item scale : often used in sur veys and in experiments b/c they can measure attitudes & can avoid
redundancy; form of a thermometer scale” such as from 0 to 100
Problems
Dimensionality : a test is supposed to measure only the construct of interest, BUT a score may reflect
more than one construct (ie: depression scale may also provide information for other constructs such
as anxiety = items maybe the measures of more than one construct)
Ambiguous items : (ie: men and women should have equal rights => in what aspects?)
Items with different variability : when the item does not have the same strength as the others (ie: 3
extreme hates and 1 moderate hate measures)
Extremity bias : some people are just willing to use the endpoints of a scale
Life-right bias : people like one side of the paper more than the other (solution: can be handled
statistically by using reverse code)
www.notesolution.com
You're Reading a Preview

Unlock to view full version