Skip to main content

Table 3 Effects of estimation targets, category, skew & difficulty on observed or estimated chance agreement and reliability (dr2)

From: Interrater reliability estimators tested against true interrater reliabilities

   

A.

B.

C.

D.

E.

F.

G.

H.

1

Right: Source or Author

Observation

%-agreement

Bennett et al.

Perreault & Leigh

Gwet

Scott

Cohen

Krippendorff

Effects on Intcdr Reliability Obsv & Ests

2

Right: Obsd / Estd Interrater Reliability as Dependent Variables

Down: Independent Variables

ori

ao

S

Ir

AC1

Ï€

κ

α

3

Observed Reliability (ori)

1.00***

.841***

.691***

.599***

.721***

.312***

.312***

.312***

4

Category (C)

.003

−.002

.175***

.185***

.123***

.001

.001

.001

5

Distribution Skew (sk)

.000

.000

.000

−.000

.003

−.293***

−.292***

−.293***

6

Difficulty (df)

−.774***

−.778***

−.566***

−.434***

−.554***

−.389***

−.389***

−.389***

Effects on Chance Agrt Obsv & Ests

7

Right: Obsd / Estd. Chance Agreement as Dependent Variables

Down: Independent Variables

oac

aoac = 0a

Sac

Irac

ACac

Ï€ac

κac

αac

8

Observed Chance Agreement (oac)

1.00***

–

.021**

.021**

.075***

−.151***

−.152***

−.151***

9

Category (C)

−.019**

–

−.863***

−.863***

−.661***

−.013*

−.014*

−.013*

10

Distribution Skew (sk)

−.001

–

.000

.000

−.039***

.437***

.434***

.437***

11

Difficulty (df)

.585***

–

.000

.000

.009

−.123***

−.125***

−.123***

N

12

Nc (number of rating sessions)

384

384

384

384

384

384

384

384

13

Nd (number items within each session)

100

100

100

100

100

100

100

100

  1. Main cell entries are directional r squared (dr2), which are r squared with the directional sign of r, dr2 = r•|r|
  2. *: p<.05; **: p<.01; ***: p<.001
  3. a  As aoac, the chance estimate of ao, is a constant, its correlations (dr2) with other variables cannot be calculated