Error análisis de confiabilidad

Comments

• EJ

  April 2020

  I'll ask our expert, Don.

  @Don: Google translate the above text gives:

  "I'm analyzing the reliability analysis of a variable. The items are coded with 0 (error) and 1 (hit), in one of the items all the participants were right, so JASP does not analyze that item because it indicates it as an error. How can I solve that????"

  Cheers,

  E.J.
• vandenman

  April 2020

  Unfortunately, there is no way to compute the reliability for an item with zero variance. If you try this in R, you get the following:

  library(psych)
  df <- data.frame(x = rnorm(10), y = rnorm(10), z = 1)
  var(df$z) # no variance
  [1] 0
  
  alpha(df)
  Some items ( x ) were negatively correlated with the total scale and
  probably should be reversed.  
  To do this, run the function again with the 'check.keys=TRUE' option
  Reliability analysis   
  Call: alpha(x = df)
  
    raw_alpha std.alpha G6(smc) average_r   S/N  ase  mean   sd median_r
       -0.55     -0.65   -0.24     -0.24 -0.39 0.86 -0.31 0.66    -0.24
  
   lower alpha upper     95% confidence boundaries
  -2.24 -0.55 1.14
  
   Reliability if an item is dropped:
    raw_alpha std.alpha G6(smc) average_r S/N alpha se var.r med.r
  x     -0.24     -0.24    0.06     -0.24  NA       NA -0.24 -0.24
  y      0.06     -0.24      NA        NA  NA       NA  0.06 -0.24
  
   Item statistics
     n raw.r std.r r.cor r.drop  mean   sd
  x 10  0.82  0.61   NaN  -0.24 -0.38 1.27
  y 10  0.35  0.61   NaN  -0.24 -0.25 0.76
  Warning messages:
  1: In alpha(df) : Item = z had no variance and was deleted
  2: In alpha(df) :
    Some items were negatively correlated with the total scale and probably
  should be reversed.  
  To do this, run the function again with the 'check.keys=TRUE' option
  3: In matrix(unlist(drop.item), ncol = 10, byrow = TRUE) :
    data length [16] is not a sub-multiple or multiple of the number of columns [10]
  4: In sqrt(Vtc) : NaNs produced
  

  The first warning message is:

  1: In alpha(df) : Item = z had no variance and was deleted
  

  JASP throws an error rather than deleting the items with a warning. If you still want to do a reliability analysis on the remaining items, I'd suggest deleting items with zero variance.
• MabelZR2020

  May 2020

  OK, thank you