estimate_meta_pdiff_two.md
In esci: Estimation Statistics with Confidence Intervals

Some tests for meta_pdiff_two; needs development

Code
  estimate
Output
  Analysis of raw data:

  -- es_meta --
         effect_label moderator_variable_name moderator_variable_level
  x.flip    My effect                 setting                  Overall
  1         My effect                 setting                In-Person
  2         My effect                 setting                   Online
         effect_size          LL        UL        SE k diamond_ratio
  x.flip   0.5579311  0.11751343 0.9983488 0.2247070 4             1
  1        0.7654911 -0.02552414 1.5565063 0.4035866 2             1
  2        0.4646801 -0.06551995 0.9948802 0.2705152 2             1
         diamond_ratio_LL diamond_ratio_UL I2 I2_LL    I2_UL       PI_LL
  x.flip                1         5.079969  0     0 94.67705  0.11751343
  1                     1        17.550734  0     0 99.65651 -0.02552414
  2                     1        26.193770  0     0 99.58311 -0.06551995
             PI_UL          p     width FE_effect_size RE_effect_size FE_CI_width
  x.flip 0.9983488 0.01303079 0.8808354      0.5579311      0.5579311   0.8808354
  1      1.5565063 0.05786481 1.5820305      0.7654911      0.7654911   1.5820305
  2      0.9948802 0.08584046 1.0604001      0.4646801      0.4646801   1.0604001
         RE_CI_width
  x.flip   0.8808354
  1        1.5820305
  2        1.0604001

  -- es_heterogeneity --
          effect_label moderator_variable_name moderator_level       measure
  1          My effect                 setting         Overall Diamond Ratio
  11         My effect                 setting       In-Person Diamond Ratio
  12         My effect                 setting          Online Diamond Ratio
  H^2        My effect                 setting         Overall           H^2
  H^21       My effect                 setting       In-Person           H^2
  H^22       My effect                 setting          Online           H^2
  I^2(%)     My effect                 setting         Overall        I^2(%)
  I^2(%)1    My effect                 setting       In-Person        I^2(%)
  I^2(%)2    My effect                 setting          Online        I^2(%)
  tau        My effect                 setting         Overall           tau
  tau1       My effect                 setting       In-Person           tau
  tau2       My effect                 setting          Online           tau
  tau^2      My effect                 setting         Overall         tau^2
  tau^21     My effect                 setting       In-Person         tau^2
  tau^22     My effect                 setting          Online         tau^2
          estimate LL         UL
  1              1  1   5.079969
  11             1  1  17.550734
  12             1  1  26.193770
  H^2            1  1  18.786568
  H^21           1  1 291.125405
  H^22           1  1 239.872084
  I^2(%)         0  0  94.677048
  I^2(%)1        0  0  99.656505
  I^2(%)2        0  0  99.583111
  tau            0  0   2.200472
  tau1           0  0  10.000000
  tau2           0  0  10.000000
  tau^2          0  0   4.842076
  tau^21         0  0 100.000000
  tau^22         0  0 100.000000
  [1] "As of version 1.0.2 esci has implemented an improved method for calculating the CI for the diamond ratio; these will no longer match those presented in the 2nd edition of <i>Introduction to the New Statistics</i>."

  -- es_meta_difference --
                   type effect_label moderator_variable_name    moderator_level
  Comparison Comparison    My effect                 setting             Online
  Reference   Reference    My effect                 setting          In-Person
  Difference Difference    My effect                 setting Online ‒ In-Person
             effect_size          LL        UL        SE          p
  Comparison   0.4646801 -0.06551995 0.9948802 0.2705152 0.08584046
  Reference    0.7654911 -0.02552414 1.5565063 0.4035866 0.05786481
  Difference  -0.3008110 -1.25308049 0.6514585 0.4858607 0.53583069


  Note: LL and UL are lower and upper boundaries of confidence intervals with 95% expected coverage.