-
Notifications
You must be signed in to change notification settings - Fork 0
/
11-exp-int.Rmd
85 lines (68 loc) · 7.16 KB
/
11-exp-int.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
# Expectations and Intentions {#exp-int}
```{r 11-setup}
library(tidyverse)
library(lubridate)
library(papaja)
library(BayesFactor)
library(kableExtra)
library(ggthemes)
Group1to4 <- readRDS("chapter_3_data/Group1to4.rds")
group_5 <- readRDS("chapter_3_data/Group5.rds")
df <- group_5 %>%
mutate(Assessed = as.factor(case_when(Assessed == 1 ~ TRUE,
Assessed == 0 ~ FALSE))) %>%
filter((firstMLAssessmentDate > SurveyEndDate |
is.na(firstMLAssessmentDate)) &
!is.na(NowAssessIntention))
```
Results of Chapter \@ref(ml-qualitative) suggested that a significant number of candidates registered for the Mountain Leader qualification with the intention of only attending a training course and not going on to be assessed. However, this variable was not selected as an important discriminatory variable in the survey tool pilot work and was therefore dropped from the final survey tool. Given the incongruence of this finding with the results of Chapter 2, we inspected the data collected in the pilot work to better understand this discrepancy.
Candidates in Group 3 and Group 4 of the pilot data ($n = `r nrow(filter(Group1to4, !is.na(v19)))`$) were asked, "Candidates who have registered for the ML may have different aspirations. Below is a list of common aspirations. Please tick the option which best reflected your aspirations at registration, you may only choose one." The response options were: (a) Mountain Leader training only, (b) Becoming a Mountain Leader, (c) Going onto higher walking qualifications, and (d) Going onto higher mountaineering qualifications. Interestingly, just `r printnum(sum(Group1to4$v19==0, na.rm = T))` candidates (`r printnum((1-Group1to4$v19 %>% mean(na.rm = TRUE))*100)`%) from Groups 3 and 4 selected "Mountain Leader Training only." Further, `r numbers2words(printnum(nrow(filter(Group1to4, v19 == 0 & Assessed_18months == TRUE))))` of these candidates had been assessed 18 months after their training course. We would suggest that the rarity of the intention to only attend a training course in the pilot data is the reason it was not carried forward to the final survey tool. Furthermore, this rarity challenges the assumption of the participants in Chapter 2 as very few of the respondents to the pilot survey stated that they only intended to attend a training course and nearly a third of those who did state that had been assessed 18 months after their training course.
When considering this and the "context" findings of Chapter \@ref(ml-pra),^[Which suggest the relative importance of becoming a Mountain Leader and, for male candidates, the strength of intention to be assessed at the end of a training course are important variables from discriminating candidates who are assessed 18 months after their training course from those who are not.] it appears that the strength of intention is more important that the intention itself and we would hypothesise that the greater the strength of intention to be assessed, the greater the likelihood of being assessed. As well as asking candidates in the final survey group (i.e., those trained in 2017 and 2018) about the strength of their of intention to be assessed at registration, the start of training, and the end of training, we asked candidates who had not been assessed when they completed the survey ($n = `r nrow(filter(group_5, (firstMLAssessmentDate > SurveyEndDate | is.na(firstMLAssessmentDate)) & !is.na(NowAssessIntention)))`$) to rate the strength of their intention to be assessed at that point in time on a scale from *no intention of being assessed* (0) to *every intention of being assessed* (100). This variable was not included in any of the pattern recognition analyses (see Table \@ref(tab:exp-int-demographics) for summary statistics).
```{r exp-int-demographics}
df %>%
group_by(Assessed) %>%
summarise(n = n(),
`Female (\\%)` = printnum(mean(if_else(SexId == 2, 1 ,0) * 100)),
`$M_{Age} ± 1 SD$` =
paste(printnum(mean(TrainAge, na.rm = TRUE)),
"±",
printnum(sd(TrainAge, na.rm = TRUE))),
`$M_{years\\,since\\,training}$ ± 1 SD` =
paste(printnum(mean(interval(firstMLTrainingDate, SurveyEndDate)/years(1),
na.rm = TRUE)),
"±",
printnum(sd(interval(firstMLTrainingDate, SurveyEndDate)/years(1),
na.rm = TRUE))),
`$M_{intention\\,now} ± 1 SD$` =
paste(printnum(mean(NowAssessIntention, na.rm = TRUE)),
"±",
printnum(sd(NowAssessIntention, na.rm = TRUE)))) %>%
kable(caption = "Summary statistics for candidates who had not been assessed when completing the survey.",
digits = 2, booktab = TRUE, align = "c", escape = FALSE) %>%
kable_styling(bootstrap_options = c("striped", "hover", "condensed"),
latex_options = "scale_down",
full_width = FALSE)
```
```{r exp-int-analysis}
m1 <- anovaBF(formula = NowAssessIntention ~ Assessed, data = df)
```
We performed a Bayesian t-test using the `BayesFactor` package [@R-BayesFactor] in `R` [@R-base], using the default settings, to test if there was a difference in the mean strength of intention to be assessed when completing the survey between those who had been assessed six months post-survey and those who had not been. Results of this prospective analysis showed strong evidence for their being a difference in the mean intention of being assessed between groups`r apa_print.BFBayesFactor(m1)$full_result`. Figure \@ref(fig:exp-int-plot) shows the distribution of scores for both those who had been assessed and those who had not been. It is important to note that only three of the 45 candidates who had been assessed rated their intention lower than 90.
The analyses presented in this appendix suggest that most candidates do intend to be assessed but this intention must be strong for them to get to assessment. A potential implication of this finding could be that if Mountain Training wish to increase the number of candidates getting to assessment, course staff should aim to increase the strength of candidates' intentions to be assessed. Future studies could examine the efficacy and mechanism of any such intervention.
```{r exp-int-plot, fig.cap="Distribution of intention to be assessed when completing the survey by outcome with individual data points overlaid, grouped by sex." }
group_5 %>%
mutate(Assessed = case_when(Assessed == 1 ~ TRUE,
Assessed == 0 ~ FALSE),
SexId = case_when(SexId == 2 ~ "Female",
SexId == 1 ~ "Male")) %>%
filter((firstMLAssessmentDate > SurveyEndDate |
is.na(firstMLAssessmentDate)) &
!is.na(NowAssessIntention)) %>%
ggplot(aes(x = Assessed, y = NowAssessIntention, colour = Assessed)) +
geom_violin(aes(fill = Assessed), alpha = .5) +
geom_jitter(alpha = 1, height = 0) +
ylab("Intention to be assessed when completing the survey.") +
xlab("Assessed within six months of completing the survey.") +
theme_few(base_family = font) +
theme(legend.position = "none") +
facet_grid(~SexId)
```