A Comparison of Statistical Tests for Likert-Type Data: The Case of Swearwords

Roald Eiselen; Gerhard B. van Huyssteen

doi:10.5334/johd.132

Figures & Tables

Table 1

Summary of statistical tests used in previous studies on swearwords.

PUBLICATION	TASK	ASPECTS MEASURED	SCALE	# USABLE RESPONDENTS	STATISTICAL TESTS
Beers Fägersten (2007)	Word ratings	Offensiveness	1–10 (Not offensive – Very offensive)	60	Descriptives (mean, mode, SD); One-way ANOVA test for significance
Beers Fägersten(2012)	Multiple choice questions	Frequency of swearing	Never, Rarely, Sometimes, Often	60	Descriptives (percentages)
	Likelihood judgements	Frequency of swearing	0–100 (Not likely at all – Most likely possible); 1–9 (Never heard at all – Heard very frequently)	53; 59	Descriptives (mean)
Jay(1992)	Situation ratings	Offensiveness and tabooness	1–9 (Not offensive/ obscene at all – Most offensive/ obscene word imaginable)	52; 59; 90†	Descriptives (mean, rank-order); ANOVA
Janschewitz (2008)	Word ratings	Personal use; familiarity; offensiveness; tabooness; valence; arousal; imageability	1–9 (Positive/low –Negative/high)	78	One-way ANOVA; 2×5 mixed ANOVA; post hoc t test with Bonferroni corrected alphas
Dewaele(2004)	Multiple choice questions	Perception of emotional force	1 = Does not feel strong; 2 = Little; 3 = Fairly; 4 = Strong; 5 = Very strong	1039 + 50††	Multivariate analyses of variance (MANOVA); Scheffe’ post-hoc test; linear regression analysis
Dewaele(2018)	Multiple choice questions; word ratings	Familiarity with meaning; offensiveness; frequency	0–5 (Very low – Very high)	2347	One-sample Kolmogorov-Smirnov test; Mann-Whitney test; Kruskal-Wallis H test

[i] Note: † = Three different experiments. †† = Two modes of collection (online and paper-based).

Table 2

Median Likert scale (1–9) values of the four response variables for the four swearwords.

WORD	PRODUCTION	PERCEPTION	OFFENSIVENESS	TABOONESS
*feeks*	3	5	2	3
*piele*	1	3	5	7
*moffie*	2	4	7	7
*jissis*	1	4	9	8

Table 3

Sociodemographic summary of respondents for the four swearwords.

SOCIODEMOGRAPHIC FACTOR	LEVEL	FEEKS (n= 147)	PIELE (n= 167)	MOFFIE (n= 179)	JISSIS (n= 152)
SOCIODEMOGRAPHIC FACTOR	LEVEL	n (%)
Age group	18–39	55 (37.41)	65 (38.92)	67 (37.43)	63 (41.45)
	40–59	62 (42.18)	66 (39.52)	76 (42.46)	59 (38.82)
	60+	30 (20.41)	36 (21.56)	36 (20.11)	30 (19.74)
Sex	Male	55 (37.41)	63 (37.72)	73 (40.78)	59 (38.82)
Sex	Female	92 (62.59)	104 (62.28)	106 (59.22)	93 (61.18)
Religious views	Very religious	30 (20.41)	35 (20.96)	34 (18.99)	33 (21.71)
	Religious	56 (38.10)	65 (38.92)	65 (36.31)	58 (38.16)
	Moderate	19 (12.93)	23 (13.77)	26 (14.53)	24 (15.79)
	Not really	18 (12.24)	18 (10.78)	21 (11.73)	17 (11.18)
	Not at all	24 (16.33)	26 (15.57)	33 (18.44)	20 (13.16)
Political views	Very Conservative	2 (1.36)	2 (1.20)	2 (1.12)	4 (2.63)
	Conservative	11 (7.48)	14 (8.38)	13 (7.26)	12 (7.89)
	Moderate	61 (41.50)	75 (44.91)	75 (41.90)	69 (45.39)
	Liberal	34 (23.13)	39 (23.35)	43 (24.02)	36 (23.68)
	Very liberal	39 (26.53)	37 (22.16)	46 (25.70)	31 (20.39)
World view	Very Conservative	2 (1.36)	2 (1.20)	2 (1.12)	2 (1.32)
	Conservative	17 (11.56)	22 (13.17)	19 (10.61)	19 (12.50)
	Moderate	37 (25.17)	44 (26.35)	44 (24.58)	46 (30.26)
	Liberal	34 (23.13)	44 (26.35)	43 (24.02)	36 (23.68)
	Very liberal	57 (38.78)	55 (32.93)	71 (39.66)	49 (32.24)

Table 4

Significant ANOVA results.

WORD	RESPONSE	PREDICTOR	n	MEDIAN	ANOVA p	KW-H p	OLR p
feeks	Production	Age	147	3	.010	.004	.005
feeks	Offensiveness	Sex	147	2	.038	.009	N/A
piele	Perception	World view	165†	3	.022	.009	.001
piele	Offensiveness	Age	167	5	.001	.001	.016
piele	Offensiveness	Sex	167	5	<.0005	<.0005	<.0005
piele	Offensiveness	Political view	165†	5	.038	.028	.007
piele	Offensiveness	World view	165†	5	.013	.012	.001
piele	Tabooness	Sex	167	7	.007	.005	.004

[i] Note: N/A = Violation of the assumption of proportional odds for OLR; † = Respondents from “Very conservative” level excluded since there were fewer than five respondents.

Table 5

Significant KW-H results.

WORD	RESPONSE	PREDICTOR	n	MEDIAN	ANOVA p	KW-H p	OLR p
piele	Production	Age	167	1	.088	.043	.251
piele	Production	Sex	167	1	N/A	<.0005	<.0005
piele	Production	Political view	165†	1	.069	.008	N/A
piele	Production	World view	165†	1	N/A	.0134	.001
piele	Perception	Political view	165†	3	.074	.034	N/A
piele	Tabooness	Age	167	7	.058	.004	.015
moffie	Production	Sex	179	2	N/A	.008	.002
moffie	Offensiveness	Sex	179	7	N/A	.004	.002
jissis	Production	Sex	152	1	N/A	.001	.002
jissis	Production	Political view	148†	1	N/A	.001	N/A
jissis	Production	Religious view	152	1	N/A	<.0005	<.0005
jissis	Production	World view	150†	1	N/A	<.0005	N/A
jissis	Offensiveness	Sex	152	9	.0526	.044	.031
jissis	Offensiveness	Political view	148†	9	N/A	.001	N/A
jissis	Offensiveness	Religious view	152	9	N/A	<.0005	N/A
jissis	Offensiveness	World view	150†	9	N/A	<.0005	<.0005
jissis	Tabooness	Religious view	152	8	.054	.016	.010

[i] Note: N/A = Violation of assumption of equal variance for ANOVA, or assumption of proportional odds for OLR; † = Respondents from “Very conservative” level excluded since there were fewer than five respondents.

Table 6

Significant OLR results.

WORD	RESPONSE	PREDICTOR	n	MEDIAN	ANOVA p	KW-H p	OLR p
piele	Perception	Religious view	167	3	.207	.125	.040
piele	Offensiveness	Religious view	167	5	.132	.086	.016
moffie	Perception	Political view	177	4	.338	.304	.030

Table 7

Dunn’s post-hoc test results for Production of the word feeks.

AGE GROUP	18–39	40–59	60+
18–39	1.000	.006	.037
40–59	.006	1.000	1.000
60+	.037	1.000	1.000

Table 8

Tukey’s HSD test results for Production of the word feeks.

AGE GROUP	18–39	40–59	60+
18–39	1.000	.014	.053
40–59	.014	1.000	.999
60+	.053	.999	1.000

Box plot of different Age levels for the Production of *feeks*.

Table 9

OLR significance values for the Production of feeks.

AGE GROUP	18–39	40–59	60+
18–39	1.000	.001	.008
40–59	.001	1.000	.983
60+	.008	.983	1.000

Table 10

OLR odds ratio values for the Production of feeks.

AGE GROUP	18–39	40–59	60+
18–39	1.000	.333	.335
40–59	.333	1.000	1.008
60+	.335	1.008	1.000