The number of registrants for the course is currently 2711 from 91 countries.
Hover the pointer over the leftmost point in each group to see the variable label and overall statistics.
f <- ~ statlit + bayesexp + bayes + degree + position + area + clinres +
govfund + way + country + how + dtime
des <- describe(f, data=d, descript='')
p <- plot(des)
p$Categoricalp$ContinuousClick elements of the legend to make the indicated quantities disappear/reappear. Drag the scrollbar to see more of the legend.
w <- subset(d, way == "by watching the live stream with live chat on Friday mornings U.S. Eastern time")
swr <- function(w, ...)
sapply(strwrap(w, ..., simplify=FALSE),
function(x) paste(x, collapse='<br>'))
levels(d$way) <- swr(levels(d$way), width=25)
with(d, histboxp(x=dtime, group=way))f <- subset(as.data.frame(table(d$country)), Freq > 0)
names(f) <- c('Country', 'Frequency')
print(f, row.names=FALSE) Country Frequency
Argentina 8
Armenia 1
Australia 58
Austria 8
Bangladesh 12
Belgium 9
Benin 1
Bolivia 1
Brazil 50
Cameroon 1
Canada 154
Chile 6
China 15
China - Hong Kong / Macau 4
Colombia 11
Congo (DRC) 1
Croatia 2
Cyprus 2
Czech Republic 1
Denmark 42
Egypt 5
Estonia 1
Ethiopia 1
Finland 8
France 15
Gabon 3
Gambia 1
Germany 46
Ghana 6
Great Britain 209
Greece 7
Guatemala 2
Hungary 4
Iceland 8
India 56
Indonesia 3
Israel and the Occupied Territories 5
Italy 30
Jamaica 1
Japan 9
Kenya 20
Korea (South) 5
Kosovo 1
Kuwait 2
Latvia 1
Lebanon 2
Malawi 3
Malaysia 6
Mexico 18
Monaco 1
Mozambique 2
Myanmar/Burma 8
Namibia 1
Nepal 2
New Zealand 5
Nigeria 19
Norway 23
Pakistan 5
Paraguay 1
Peru 4
Philippines 8
Poland 4
Portugal 4
Puerto Rico 2
Qatar 7
Russian Federation 6
Rwanda 1
Saudi Arabia 54
Serbia 3
Sierra Leone 3
Singapore 23
Slovak Republic (Slovakia) 1
South Africa 27
Spain 61
Sri Lanka 1
Sweden 46
Switzerland 15
Tanzania 1
Thailand 3
Netherlands 26
Trinidad & Tobago 2
Turkey 16
Uganda 18
United Arab Emirates 2
United States of America (USA) 1336
Uruguay 3
Vietnam 55
Virgin Islands (UK) 4
Zambia 8
Zimbabwe 1
other 34em <- tolower(d$email)
w <- strsplit(em, '.', fixed=TRUE)
suffix <- sapply(w, function(x) x[length(x)])
table(suffix)suffix
ae ai ar at au be br ca cat cc ch cl co com con ddu
1 1 2 3 33 2 10 97 2 1 4 1 5 1204 2 1
de dk edu es eu eus fi fr gov gr id ie il in info int
29 31 548 14 5 1 5 7 32 1 1 14 1 5 1 1
is it jp ke me mil mm mx net ng nl no nz org pe ph
7 18 7 1 1 2 1 3 30 1 10 14 3 342 1 2
pk pl qa ru sa se sg to uk vn za
3 1 1 1 4 33 5 1 145 2 7 w <- strsplit(em, '@', fixed=TRUE)
server <- sapply(w, function(x) x[length(x)])
tab <- table(server)
cat('\nTop 30 servers\n\n')
Top 30 serverstab <- (-sort(-tab))[1:30]
tabserver
gmail.com vumc.org hotmail.com yahoo.com
918 192 83 61
vanderbilt.edu duke.edu vai.org fda.hhs.gov
59 33 29 25
umn.edu jhmi.edu clin.au.dk kcl.ac.uk
25 24 21 19
celgene.com outlook.com nhs.net ucl.ac.uk
17 17 13 13
ucr.uu.se utsouthwestern.edu ttuhsc.edu mgh.harvard.edu
13 13 12 11
beaumont.org brown.edu manchester.ac.uk ucalgary.ca
10 10 10 10
virginia.edu wustl.edu emory.edu mail.utoronto.ca
10 10 9 9
med.cornell.edu musc.edu
9 9 Let’s see how various participant descriptors relate to the degree of acceptance of a Bayesian approach to statistics. Only participants with a self-graded understanding of Bayesian methods exceeding 5 on a 100-point scale were asked to provide their Bayesian acceptance 0-100 scale. Start by getting ordinary unadjusted stratified means by the categorical descriptors. Note: The course does not concentrate on Bayesian methods; this information is primarily for designing future educational efforts.
label(d$bayes) <- 'Acceptance of Bayes for Data Analysis & Inference'
summaryDp(bayes ~ degree + area + clinres, data=d)Now consider a multivariable regression analysis. Degree of Bayesian understanding, modeled with a restricted cubic spline function, is one of the predictors. The semiparametric proportional odds ordinal logistic model is used for this analysis. First, the ordinary Spearman rank correlation between Bayes knowledge and Bayes acceptance is given. The first graph below shows the relative predictive information of each of the competing predictors.
with(d, spearman(bayesexp, bayes)) rho
0.3389622 options(prType='html')
d <- subset(d, select=-survey)
dd <- datadist(d); options(datadist='dd')
f <- orm(bayes ~ rcs(bayesexp, 5) + rcs(statlit, 5) + degree + area +
clinres, data=d)
M <- Mean(f) # derive R function to translate log odds to mean
print(f, coefs=FALSE)
orm(formula = bayes ~ rcs(bayesexp, 5) + rcs(statlit, 5) + degree +
area + clinres, data = d)
Frequencies of Missing Values Due to Each Variable
bayes bayesexp statlit degree area clinres
1287 467 0 0 0 0
| Model Likelihood Ratio Test |
Discrimination Indexes |
Rank Discrim. Indexes |
|
|---|---|---|---|
| Obs 1424 | LR χ2 277.23 | R2 0.177 | ρ 0.390 |
| Distinct Y 94 | d.f. 24 | g 0.863 | |
| Y0.5 65 | Pr(>χ2) <0.0001 | gr 2.370 | |
| max |∂log L/∂β| 1×10-7 | Score χ2 295.56 | |Pr(Y ≥ median)-½| 0.135 | |
| Pr(>χ2) <0.0001 |
plot(anova(f))p <- plotp(Predict(f, fun=M), vnames='names',
ylab='Mean Bayes Acceptance Score', ncols=1)
p$Continuousp$Categorical