Skip to main content

Table 3 LCA results: Model fit, class size and profiles in 2-class solution

From: Using natural language processing to explore heterogeneity in moral terminology in palliative care consultations

 

LL

BIC (LL)

L2

df

No. of parameters

1-Class

− 1210.1

2474.7

854.5

221

10

2-Class

− 1079.7

2273.8

593.6

210

21

3-Class

− 1073.0

2320.1

580.1

199

32

4-Class

− 1068.1

2370.1

570.3

188

43

5-Class

− 1064.4

2422.7

563.0

177

54

 

Class 1

Class 2

Size (%)

68.3

31.7

Indicators (# words)

 Harm-Virtue (Wald = 67, p < 0.00)

0.25

0.99

 Harm-Vice (Wald = 33, p < 0.00)

0.20

0.90

 Fairness-Virtue (Wald = 42, p < 0.00)

0.06

0.57

 Fairness-Vice (Wald = 0.3, p < 0.62)

0.01

0.03

 Ingroup-Virtue (Wald = 56, p < 0.00)

0.23

1.17

 Ingroup-Vice (Wald = 19, p < 0.00)

0.00

0.06

 Authority-Virtue (Wald = 71, p < 0.00)

0.14

1.27

 Authority-Vice (Wald = 4, p = 0.04)

0.00

0.02

 Purity-Virtue (Wald = 24, p < 0.00)

0.01

0.43

 Purity-Vice (Wald = 8, p = 0.01)

0.01

0.10

Covariates

Gender (%)

  Male

52

48

  Female

48

52

Age (%) (Wald = 15, p < 0.00)

   < 22

20

17

  23–30

21

18

  31–39

22

21

  40–45

20

18

  46–61

17

26

Race and Ethnicity (%) (Wald = 8, p < 0.00)

  Black or Hispanic

20

25

  White

80

75

Education (%) (Wald = 4, p = 0.04)

  Did not graduate high school

18

11

  High school graduate or GED

33

19

  Associate Degree / Technical School

24

31

  Bachelor’s Degree

12

15

  Masters or Doctorate Degree

13

24

Financial Security (%) (Wald = 1, p = 0.23)

  None of the time

8

8

  Some of the time

25

28

  Most of the time

29

27

  All of the time

38

37

Spiritual Needs supported (%) (Wald = 3, p = 0.08)

  Completely

22

19

  Quite a bit

18

23

  Moderately

17

21

  Slightly

9

8

  Not at all

28

26

  Missing

6

2

Emotional Distress (%) (Wald = 0.4, p = 0.51)

  Completely

17

17

  Quite a bit

22

19

  Moderately

12

20

  Slightly

21

26

  Not at all

22

17

  missing

6

1

EOL preference focus comfort-directed treatment (%) (Wald = 0.4, p = 0.55)

  Strongly disagree

8

4

  Disagree

3

2

  Not sure

23

18

  Agree

12

19

  Strongly agree

50

56

  missing

4

1

Total words used (excluding stop words) (%) (Wald = 22, p < 0.00)

   < 32

29

0

  33–73

29

0

  74–110

28

3

  111–151

13

36

  152–197

1

61

Christian (%) (Wald = 8, p = 0.01)

  0

32

28

  1

68

72

Other religion (%) (Wald = 7, p = 0.01)

  0

93

86

  1

6

14

  1. LL log-likelihood, L2 likelihood-ratio chi-squared statistic, df degrees of freedom
  2. Results for indicators represent percentages; numbers for covariates represent counts (n)