week 2 rms2

week 2 rms2

1. measures used in psych

   • self report
   • overt behaviour
   • physiological
   • behavior measurements can be based on
     • rate
       • duration
         • frequency
           • latency
             • intensity
               • accuracy
     • measurement of abstract constructs
       • verbal / non verbal behaviour
         • overt actions
           • physiological responses
       • why is measurement important
         • research may fail to find real effect
           • research find false effect
             • findings may be inconsistent across replication

2. RELIABILITY

   • getting consistent measures across time
     • a reliable measure is
       • consistent, stable,trustworthy and predictable
       • necessary due to constant error
         • if reliability is .40 then validity's ceiling is .40
           • measure value = true value + error
             • reducing error increases reliability
   • reliablility is a prereq for validity
     • if not reliable it is not valid
     • being reliable dosnt mean its valid
       • the more reliable= the more potential of validity
     • measurement error :
       • systematic error (bias)
         • will vary in a certain ways across levels of the IV
           • environmental, participant, experimenter, measuring tools
             • major threat to internal validity
             • have potential to bias if changed from experimental conditions.
       • random error
         • is unsystematic
         • does not vary in any particular way across iv levels
         • will influence individual scores
           • increase in some decrease others
         • has less effect on group mean
           • if error truly random it balances out to 0
             • will form a normal distribution
         • will increase variability in the scores
           • make manipulation of iv harder to detect
           • ways to assess reliability
             • test retest reliability
               • give same measure to same subject twice
                 • used for most meaures
                   • indicates how stable measure over time
                     • examine relationship between T1 & T2
                       • retest coefficent
                         • .60 cut off (some .70)
                         • vary from 0 - +1
             • inter observer(rate) agreement
               • inter observer(rate) reliability
                 • Used when measurements are based on ratings/scores
                   • Measures consistency between rater
                     • For quantitative data, it is the square of the correlation between the two raters
                       • range from 0-1
                         • Indicates the percentage of relationship between the two rater
                           • should be above .70
                           • interobserver agreement/reliability improved by
                             • training raters
                             • Simplifying their task ie less catagories
                             • replacing poor rate
               • Used when measurements are based on ratings/scores
                 • indicates extent of difference rater give from same infomation
                   • test 2 raters to see if there are similar
                     • ie 2 rate 10 measures and see how many they both got the same
                       • Will range from 0 to 100% (the higher the better)
             • internal consistency
       • sources of error
         • environment
           • participants
             • experimenter
               • measuring tools
         • Administration should be standardised