Summary Statistics of Abalones’ - Dimensional Measurements

A useful R package for making tables is the arsenal package. Learn more at:

The key function is tableby(), see the vignette at https://mayoverse.github.io/arsenal/articles/tableby.html.

	Overall (N=4169)
length
Mean (SD)	0.524 (0.120)
Range	0.075 - 0.815
diameter
Mean (SD)	0.408 (0.099)
Range	0.055 - 0.650
height
Mean (SD)	0.139 (0.039)
Range	0.010 - 0.515

Summary Statistics of Abalones’ - Weight Measurements

	Overall (N=4169)
wholeWeight
Mean (SD)	0.830 (0.490)
Range	0.002 - 2.825
shuckedWeight
Mean (SD)	0.360 (0.222)
Range	0.001 - 1.488
visceraWeight
Mean (SD)	0.181 (0.110)
Range	0.000 - 0.760
shellWeight
Mean (SD)	0.239 (0.139)
Range	0.002 - 1.005

Abalone Dimensional Measurements by Sex - default statistical tests

Now we can add a grouping variable such as comparing these summary statistics between the 3 biological sex groups: Male, Female and Infant.

Notice that the default settings produce a p-value. The arsenal::tableby() function is performing an ANOVA (analysis of variance) for each of these measurements.

	F (N=1306)	I (N=1335)	M (N=1528)	Total (N=4169)	p value
length					< 0.001¹
Mean (SD)	0.579 (0.086)	0.428 (0.109)	0.561 (0.103)	0.524 (0.120)
Range	0.275 - 0.815	0.075 - 0.725	0.155 - 0.780	0.075 - 0.815
diameter					< 0.001¹
Mean (SD)	0.455 (0.071)	0.327 (0.088)	0.439 (0.084)	0.408 (0.099)
Range	0.195 - 0.650	0.055 - 0.550	0.110 - 0.630	0.055 - 0.650
height					< 0.001¹
Mean (SD)	0.157 (0.030)	0.108 (0.032)	0.151 (0.035)	0.139 (0.039)
Range	0.015 - 0.250	0.010 - 0.220	0.025 - 0.515	0.010 - 0.515

Linear Model ANOVA

Abalone Dimensional Measurements by Sex - change statistical test

Suppose you decide that diameter is skewed and really need non-parametric statistics and the Kruskall-Wallis non-parametric ANOVA test performed. We can customize the statistics - learn more at https://mayoverse.github.io/arsenal/articles/tableby.html#change-summary-statistics-within-the-formula-1 and see more options at https://mayoverse.github.io/arsenal/articles/tableby.html#available-function-options-1.

You’ll notice this automatically creates footnotes for each customized statistic in the output table.

	F (N=1306)	I (N=1335)	M (N=1528)	Total (N=4169)	p value
length					< 0.001¹
Mean (SD)	0.579 (0.086)	0.428 (0.109)	0.561 (0.103)	0.524 (0.120)
diameter					< 0.001²
Median (Q1, Q3)	0.465 (0.410, 0.505)	0.335 (0.270, 0.395)	0.455 (0.395, 0.500)	0.425 (0.350, 0.480)
height					< 0.001¹
Mean (SD)	0.157 (0.030)	0.108 (0.032)	0.151 (0.035)	0.139 (0.039)

Linear Model ANOVA
Kruskal-Wallis rank sum test

Abalone Weight Measurements by Sex

	F (N=1306)	I (N=1335)	M (N=1528)	Total (N=4169)	p value
wholeWeight					< 0.001¹
Mean (SD)	1.047 (0.430)	0.432 (0.286)	0.991 (0.471)	0.830 (0.490)
Range	0.080 - 2.657	0.002 - 2.050	0.015 - 2.825	0.002 - 2.825
shuckedWeight					< 0.001¹
Mean (SD)	0.446 (0.199)	0.191 (0.128)	0.433 (0.223)	0.360 (0.222)
Range	0.031 - 1.488	0.001 - 0.773	0.006 - 1.351	0.001 - 1.488
visceraWeight					< 0.001¹
Mean (SD)	0.231 (0.098)	0.092 (0.063)	0.216 (0.105)	0.181 (0.110)
Range	0.021 - 0.590	0.000 - 0.440	0.003 - 0.760	0.000 - 0.760
shellWeight					< 0.001¹
Mean (SD)	0.302 (0.126)	0.128 (0.085)	0.282 (0.131)	0.239 (0.139)
Range	0.025 - 1.005	0.002 - 0.655	0.005 - 0.897	0.002 - 1.005

Linear Model ANOVA

Plot of Abalone Age by shuckedWeight

Plot of Abalone Age by Shucked Weight - by sex

Create a plot of abalone age by shucked weight in g Show the plot by sex - either add a color by sex or a facet_wrap().

Linear Regression - Abalone Age by Shucked Weight (model 1)

Regression of Abalone Age by Shucked Weight
term	estimate	std.error	statistic	p.value
(Intercept)	9.243295	0.0861858	107.24850	0
shuckedWeight	6.110826	0.2039592	29.96102	0

Linear Regression - Abalone Age by Shucked Weight, adjusted for sex (model 2)

Regression of Abalone Age by Shucked Weight adjusted for Sex
term	estimate	std.error	statistic	p.value
(Intercept)	10.9143080	0.1285277	84.917946	0.0000000
shuckedWeight	3.8482792	0.2295312	16.765820	0.0000000
sexI	-2.2489742	0.1239592	-18.142855	0.0000000
sexM	-0.3749078	0.1057687	-3.544601	0.0003975

Compare Models - piecemeal steps

The change in R2 for the 2 models is 0.0657824 with a p-value of 4.3055706^{-76}.

Compare models - use `gtsummary` package

Model 1 output

Characteristic	Beta	95% CI¹	p-value
(Intercept)	9.2	9.1, 9.4	<0.001
shuckedWeight	6.1	5.7, 6.5	<0.001
¹ CI = Confidence Interval

Model 2 output

Characteristic	Beta	95% CI¹	p-value
(Intercept)	11	11, 11	<0.001
shuckedWeight	3.8	3.4, 4.3	<0.001
sex
F	—	—
I	-2.2	-2.5, -2.0	<0.001
M	-0.37	-0.58, -0.17	<0.001
¹ CI = Confidence Interval

Put models side by side

Characteristic	Model 1			Model 2
Characteristic	Beta	95% CI¹	p-value	Beta	95% CI¹	p-value
(Intercept)	9.2	9.1, 9.4	<0.001	11	11, 11	<0.001
shuckedWeight	6.1	5.7, 6.5	<0.001	3.8	3.4, 4.3	<0.001
sex
F				—	—
I				-2.2	-2.5, -2.0	<0.001
M				-0.37	-0.58, -0.17	<0.001
¹ CI = Confidence Interval

The `stargazer` package - works for HTML and PDF

WARNING This does NOT work for WORD documents. However, you can create the HTML output and then “cut-and-paste” the HTML table into WORD.


	Dependent variable:

	age
	(1)	(2)

shuckedWeight	6.111^***	3.848^***
	(0.204)	(0.230)

sexI		-2.249^***
		(0.124)

sexM		-0.375^***
		(0.106)

Constant	9.243^***	10.914^***
	(0.086)	(0.129)


Observations	4,169	4,169
R²	0.177	0.243
Adjusted R²	0.177	0.242
Residual Std. Error	2.924 (df = 4167)	2.805 (df = 4165)
F Statistic	897.663^*** (df = 1; 4167)	445.716^*** (df = 3; 4165)

Note:	p<0.1; p<0.05; p<0.01

Creating APA style tables

This package makes nice output but saves each table in a separate output DOC file. This does work for WORD documents.

To embed the apaTable output inside the Rmarkdown document, we can do the following…

Pull out the key parts of the output object. Make a nice table for the formatted output for apa3 and then add the footnote using inline r code.

Predictor	b	b_95%_CI	sr2	sr2_95%_CI	Fit	Difference
(Intercept)	9.24**	[9.07, 9.41]
shuckedWeight	6.11**	[5.71, 6.51]	.18	[.16, .20]
					R2 = .177**
					95% CI[.16,.20]

(Intercept)	10.91**	[10.66, 11.17]
shuckedWeight	3.85**	[3.40, 4.30]	.05	[.04, .06]
sexI	-2.25**	[-2.49, -2.01]	.06	[.05, .07]
sexM	-0.37**	[-0.58, -0.17]	.00	[-.00, .00]
					R2 = .243**	Delta R2 = .066**
					95% CI[.22,.26]	95% CI[.05, .08]

Note. A significant b-weight indicates the beta-weight and semi-partial correlation are also significant. b represents unstandardized regression weights. beta indicates the standardized regression weights. sr2 represents the semi-partial correlation squared. r represents the zero-order correlation. Square brackets are used to enclose the lower and upper limits of a confidence interval. * indicates p < .05. ** indicates p < .01.

Visualize Regression Coefficients

Here is an example plot of the coefficients from Model 2 (lm2) using the packages from easystats - namely the parameters and see packages.

Learn more at https://easystats.github.io/easystats/.

Here is another example using the sjPlot package. Note: I also had to install/update the associated sjstats package. Learn more at http://www.strengejacke.de/sjPlot/reference/plot_model.html.

Logistic Regression of `adult` by Shucked Weight and Diameter

NOTE: The adult variable is currently a “character” class variable. So, let’s create a 0/1 coded variable.

Use parameters package to get model coefficients table.

Parameter	Coefficient	SE	CI	CI_low	CI_high	z	df_error	p
(Intercept)	0.0365486	0.0110825	0.95	0.0200387	0.0657934	-10.913014	Inf	0e+00
shuckedWeight	217.0126805	139.5074649	0.95	62.5988462	778.3652734	8.368861	Inf	0e+00
diameter	540.2870048	637.2057215	0.95	53.9313991	5495.9308663	5.335075	Inf	1e-07

Another option using the gtsummary package.

Characteristic	OR¹	95% CI¹	p-value
shuckedWeight	217	62.6, 778	<0.001
diameter	540	53.9, 5,496	<0.001
¹ OR = Odds Ratio, CI = Confidence Interval

Compute the AUC for the model and plot the ROC curve.

## 
## Call:
## roc.formula(formula = abalone$adult01 ~ pred_glm1, plot = TRUE,     print.auc = TRUE)
## 
## Data: pred_glm1 in 1335 controls (abalone$adult01 0) < 2834 cases (abalone$adult01 1).
## Area under the curve: 0.8507

Abalone Report

making tables and regression analysis

Melinda Higgins

09/06/2023

Summary Statistics of Abalones’ - Dimensional Measurements

Summary Statistics of Abalones’ - Weight Measurements

Abalone Dimensional Measurements by Sex - default statistical tests

Abalone Dimensional Measurements by Sex - change statistical test

Abalone Weight Measurements by Sex

Plot of Abalone Age by shuckedWeight

Plot of Abalone Age by Shucked Weight - by sex

Linear Regression - Abalone Age by Shucked Weight (model 1)

Linear Regression - Abalone Age by Shucked Weight, adjusted for sex (model 2)

Compare Models - piecemeal steps

Compare models - use `gtsummary` package

The `stargazer` package - works for HTML and PDF

Creating APA style tables

Visualize Regression Coefficients

Logistic Regression of `adult` by Shucked Weight and Diameter

Abalone Report

making tables and regression analysis

Melinda Higgins

09/06/2023

Summary Statistics of Abalones’ - Dimensional Measurements

Summary Statistics of Abalones’ - Weight Measurements

Abalone Dimensional Measurements by Sex - default statistical tests

Abalone Dimensional Measurements by Sex - change statistical test

Abalone Weight Measurements by Sex

Plot of Abalone Age by shuckedWeight

Plot of Abalone Age by Shucked Weight - by sex

Linear Regression - Abalone Age by Shucked Weight (model 1)

Linear Regression - Abalone Age by Shucked Weight, adjusted for sex (model 2)

Compare Models - piecemeal steps

Compare models - use gtsummary package

The stargazer package - works for HTML and PDF

Creating APA style tables

Visualize Regression Coefficients

Logistic Regression of adult by Shucked Weight and Diameter

Compare models - use `gtsummary` package

The `stargazer` package - works for HTML and PDF

Logistic Regression of `adult` by Shucked Weight and Diameter