random sample of English novels, 1800-1836 (Garside et al. bibliography)

novels.r

library(ggplot2)
library(arm)

###############################################################################
## load data
###############################################################################

data_raw = "status	year	author	further_editions
private/for-profit scan	1822	Isabel HILL	0
public scan	1827	Sarah Wilmot WELLS	0
public scan	1808	Ellen Rebecca WARNER	0
public scan	1831	REYNOLDS, Frederick	0
public scan	1823	George JONES	0
public scan	1833	[BULWER LYTTON, Edward George]	1
public scan	1828	Thomas Henry LISTER	1
public scan	1822	Jean Charles L_onard SIMONDE DE SISMONDI	1
public scan	1819	Adelaide O'KEEFFE	1
public scan	1824	Hannah Maria JONES	0
public scan	1826	Sir Walter SCOTT	1
private/for-profit scan	1819	Elizabeth BENNETT	1
public scan	1823	Lady Caroline LAMB	1
public scan	1835	CAUNTER, {J}[ohn] Hobart	1
public scan	1836	[HOOK, Theodore Edward]	1
private/for-profit scan	1801	ANON	0
public scan	1835	[DEACON, William Frederick]	1
private/for-profit scan	1806	ANON	0
public scan	1835	[SULLIVAN, Arabella Jane]; DACRE, Lady [Barbarina] (editor)	1
no scan, copies survive	1808	ANON	0
private/for-profit scan	1800	Susannah GUNNING	1
no scan, copies survive	1801	Mary CHARLTON	1
public scan	1833	[TONNA], Charlotte Elizabeth	1
public scan	1805	Isaac D'ISRAELI	1
no scan, no copy survives	1826	Friedrich August SCHULZE	0
no scan, copies survive	1825	ANON	0
public scan	1835	[BANIM, John and Michael; and MARTIN, Harriet Letitia]	1
no scan, copies survive	1813	Anne Louise Germaine de STA_L-HOLSTEIN	0
public scan	1829	Catharine HEAD	1
public scan	1821	Pierce EGAN	1
public scan	1814	ANON	1
public scan	1804	Elisabeth GU_NARD	0
public scan	1802	August Heinrich Julius LAFONTAINE	0
public scan	1832	[JAMES, George Payne Rainsford]	1
public scan	1817	Anne Julia Kemble HATTON	0
public scan	1832	ST. LEDGER, [Francis] Barry [Boyle]	0
no scan, copies survive	1803	Robert COUPER	0
no scan, copies survive	1805	August Heinrich Julius LAFONTAINE	0
no scan, copies survive	1823	ANON	0
no scan, copies survive	1800	ANON	0
no scan, copies survive	1818	ANON	1
no scan, copies survive	1808	Augusta Amelia STUART	0
public scan	1832	[CHAMIER, Frederick]	1
public scan	1826	Mrs N. W. OLIVER	1
public scan	1803	Sydney OWENSON [afterwards MORGAN, Lady Sydney]	1
public scan	1816	Thomas Love PEACOCK	1
no scan, copies survive	1801	Ann WINGROVE	0
no scan, copies survive	1830	COATES, Mr [H	0
no scan, copies survive	1810	Mary HOUGHTON	1
public scan	1833	[GORE, Catherine Grace Frances]	0
no scan, copies survive	1819	M. SMITH	0
public scan	1806	Thomas Pike LATHY	0
public scan	1831	[TAYLOR, Isaac]	0
public scan	1801	ANON	0
public scan	1823	Grace KENNEDY	1
no scan, copies survive	1812	ANON	0
public scan	1829	Timothy EAST	0
public scan	1830	COOPER, James Fenimore	1
no scan, copies survive	1816	ANON	0
no scan, copies survive	1802	Marian MOORE	0
private/for-profit scan	1803	J.-J.-M. DUPERCHE	0
no scan, copies survive	1807	ANON	0
no scan, copies survive	1813	Miriam MALDEN	0
no scan, copies survive	1804	Mary TUCK	0
public scan	1829	Catharine HEAD	1
public scan	1834	[MARRYAT, Frederick]	1
no scan, copies survive	1830	ANON	0
private/for-profit scan	1812	Amelia BEAUCLERC	0
no scan, copies survive	1811	ANON	0
no scan, copies survive	1815	Adrien de SARRAZIN	1
no scan, copies survive	1808	Mrs A. DUNCOMBE	1
public scan	1828	Elizabeth Caroline GREY	1
no scan, copies survive	1802	Anna MILLIKIN	0
no scan, copies survive	1812	ANON	0
public scan	1835	HOWITT, William	0
no scan, copies survive	1836	ANON	0
public scan	1828	Benjamin, Earl of Beaconsfield DISRAELI	1
public scan	1817	Thomas Love PEACOCK	1
public scan	1801	Elizabeth HELME	1
public scan	1835	[SHEE, Sir Martin Archer]	1
public scan	1806	ANON	0
public scan	1834	JONES, Hannah Maria	1
public scan	1832	[GALT, John]	1"

df = read.table(textConnection(data_raw), header=T, sep="\t", quote="")

###############################################################################
## make bar chart showing states
###############################################################################
# reorder levels
df$status = relevel(df$status,"private/for-profit scan")
df$status = relevel(df$status,"public scan")
# make plot
g = ggplot(data=df, aes(x=factor(""), fill=status)) +
  geom_bar(position="fill") +
  scale_y_continuous("",formatter="percent") +
  scale_x_discrete("") +
  coord_flip()

png("scan-status.png", units = 'in', res = 300, width=8, height=8 * 2/(1+sqrt(5)))
print(g)
dev.off()

###############################################################################
## logistic model
###############################################################################
# create the response variable
df$scanned = (df$status == 'public scan')

# check for NA values
stopifnot(!is.na(df$year))
stopifnot(!is.na(df$scanned))
stopifnot(!is.na(df$scanned))
stopifnot(!is.na(df$further_editions))

fit_0 = glm(scanned ~ 1, family=binomial(link="logit"), data=df)
display(fit_0)

fit_1 = glm(scanned ~ I(year-1800), family=binomial(link="logit"), data=df)
display(fit_1)

fit_2 = glm(scanned ~ I(year-1800) + further_editions,
            family=binomial(link="logit"), data=df)
display(fit_2)

anova(fit_2, test="Chisq")

###############################################################################
## visualizations
###############################################################################

### basic plots using R 
fit = fit_2
png("scan-model.png", units = 'in', res = 300, width=8, height=8 * 2/(1+sqrt(5)))
curve(invlogit(coef(fit)[1] + coef(fit)[2]*(x-1800) + coef(fit)[3]*0),
      1800, 1836,
      ylim=c(0,1),
      ylab="Proportion of novels with publicly accessible scans",
      xlab="Year",
      lty = 1)
curve(invlogit(coef(fit)[1] + coef(fit)[2]*(x-1800) + coef(fit)[3]*1),
      1800,
      1836,
      col=2,
      lty=2,
      add=TRUE)
legend("topleft", 1.5, c("no further edition", "further edition(s)"),
       col=1:2, lty=c(1,2), cex = 1.1, bty="n")
dev.off()