07-references.Rmd

# References

<div id="refs"></div>

# (APPENDIX) Appendices {-} 

# Timeline of events


## Spring 2019

This is believed to be Eagle I.O's initial definition of engagement:

>A state of personal immersion in work characterized by enthusiasm, dedication, and personal investment, expressed cognitively, affectively, and behaviorally in the proactive pursuit of advancing organizational goals.

This definition was created by Eagle IO Spring 2019, and modified by Kulas and Renata the following Fall semester of 2019 to include the *four* dimensions of Fulfillment, Absorbtion, Dedication, and Vigor. 

## Spring 2020 

We considered removing fulfillment.

*Fulfillment*: finding meaning in one's work, while having a sense of autonomy, growth, usefulness, achievement, and feeling appreciated by org. [Satisfaction(?)]

It was decided to operationalize fullfillment as an *outcome* of engagement rather than a definitional element 

### February 4, 2020

We lost the document where we had saved the citations for the creation of our engagement dimensions. we found it today (02/04/2020). 

Three out of the four dimensions (Dedication, Vigor, and Absorbtion) came from @schaufeli_measurement_2002, and we are trying to find where Fulfillment came from. 
We are also trying to improve the definition of each domain by looking at the current items and conducting a Modified Q sort (not correct name) to create piles of items that have commonalities within each domain. 

### February 24, 2020

Definitional amendment notes and considerations:

*Absorption*: being fully concentrated and happily immersed in ones work [time passes quickly and has difficulty detaching from ones work; @schaufeli_measurement_2002]

*Dedication/Commitment*: being strongly involved in one's work and experiencing a sense of enthusiasm, inspiration, pride, and challenge. [@schaufeli_measurement_2002] Identifying as an organizational member/ambassador

This includes conceptual elements of identification with the organization, a sense of "oneness", seeking continuous learning and improvement. We got rid of "challenge" altogether and moved the "inspiration" and "pride" elements to other categories. 

*Dedication*: seeking continous imporvement and demonstrating initiative 

*Vigor*: investing consistent effort, persistence, energy, and mental resilience while working [@schaufeli_measurement_2002]
maybe add enthusiasm here as well 

*Vigor*: Experiencing persistent levels of energy and enthusiasm while working 

### Structural considerations

It was considered to potentially change from a focus on affective, cognitive, and behavioral to whether their engagement stems from content/satisfaction with the organization or the people they work with. This idea was however abandoned Spring 2020.

After completing individual Q-sorts (Kulas and Renata) we decided to revisit the originally specified substantive definitions and build them up a little to make the difference between them more noticeable (and we also retained the affect, behavior, cognition focus). 

## Summer 2020

Definitions as of 5/19/2020:

*Absorption*: Being fully immersed in one’s work, where time passes quickly and one has difficulty detaching from work tasks

*Vigor*: Experiencing persistent levels of energy, effort, and enthusiasm while working

*Dedication*: Experiencing pride and challenge in ones work, as well as strong feelings of support from and loyalty toward the organization

The development team "content validation" as well as the broader content validation to inform the pilot administration (see Chapter \@ref(CValid)) retained these definitions we ordered all the items according to the ones we individually selected for each category and created an item bank with the remaining items. Together we placed the items in the bank into the agreed upon categories. 

# Pilot conditions {#pilot2}

Our four orderings of items were randomized within dimension (Affective, Behavioral, Cognitive or Dedication, Absorption, Vigor), "block" (Qualtrics designation for groupings of items), and item. The elements that were randomized are identified in the following tables by randomized element (A, B, C, or D):

```{r conditions, echo=FALSE, warning=FALSE, message=FALSE}


cond1 <- read.csv("Data/Conditions/cond1.csv")
cond2 <- read.csv("Data/Conditions/cond2.csv")
cond3 <- read.csv("Data/Conditions/cond3.csv")
cond4 <- read.csv("Data/Conditions/cond4.csv")

library(kableExtra)

kable(cond1, caption = "Pilot administration ordering Condition 1") 


```

```{r cond2, echo=FALSE, warning=FALSE, message=FALSE}

kable(cond2, caption = "Pilot administration ordering Condition 2") 

```

```{r cond3, echo=FALSE, warning=FALSE, message=FALSE}

kable(cond3, caption = "Pilot administration ordering Condition 3") 

```

```{r cond4, echo=FALSE, warning=FALSE, message=FALSE}
 
kable(cond4, caption = "Pilot administration ordering Condition 4")

```

# Correlations {#corrs}

## Inter-Item correlations

```{r correlate, echo=FALSE, warning=FALSE, message=FALSE}

mcor<-round(cor(together[2:37], use="na.or.complete" ),2)
lower<-mcor
lower[lower.tri(mcor, diag=TRUE)]<-""

library(magrittr)
library(kableExtra)

kable(mcor, "html") %>%
  kable_styling(font_size=7) %>%
  scroll_box(width = "100%", height = "900px")

```

## Scale-level correlations {.tabset .tabset-pills}

Because of the large inter-construct covariance estimates, we constructed standard unit-weighted scale aggregates (interactive plot of unit-weighted aggregates located below)

### Substantive scales

```{r correlations, echo=FALSE, warning=FALSE, message=FALSE}

testing <- CFAdata

testing$Absorption <- rowMeans(testing[1:12], na.rm=TRUE)
testing$Vigor <- rowMeans(testing[13:24], na.rm=TRUE)
testing$Dedication <- rowMeans(testing[25:36], na.rm=TRUE)

## cor(testing[37:39], use="complete.obs")

# library(plotly)
plot_ly(testing, x=~Absorption, y=~Vigor, z=~Dedication, type="scatter3d", mode="markers",
        marker=list(size=3, color = "deepskyblue"))


```

### Attitudinal scales

```{r correlations2, echo=FALSE, warning=FALSE, message=FALSE}

testing$Cognitive <- rowMeans(testing[c(1:4,13:16,25:28)], na.rm=TRUE)
testing$Affective <- rowMeans(testing[c(5:8,17:20,29:32)], na.rm=TRUE)
testing$Behavioral <- rowMeans(testing[c(9:12,21:24,33:36)], na.rm=TRUE)

## cor(testing[37:39], use="complete.obs")

# library(plotly)
plot_ly(testing, x=~Cognitive, y=~Affective, z=~Behavioral, type="scatter3d", mode="markers", 
        marker=list(size=3, color = "springgreen3"))

```

## {-}

Correlation matrix of 3 unit-weighted substantive and 3 unit-weighted attitudinal scales:

```{r correlations3, echo=FALSE, warning=FALSE, message=FALSE}

##cor(testing[37:42], use="complete.obs")
psych::pairs.panels(testing[37:42])

```

# Corrected item-total correlations {#rdrops}

## Substantive scale (marginal level)

```{r rdrop1, echo=FALSE, warning=FALSE, message=FALSE}

library(kableExtra)
kable(affect$item.stats[-c(2:4)], digits=2, caption="Affective Items")

```


```{r rdrop2, echo=FALSE, warning=FALSE, message=FALSE}

kable(behave$item.stats[-c(2:4)], digits=2, caption="Behavioral Items")

```


```{r rdrop3, echo=FALSE, warning=FALSE, message=FALSE}

kable(cogni$item.stats[-c(2:4)], digits=2, caption="Cognitive Items")

```


```{r rdrop4, echo=FALSE, warning=FALSE, message=FALSE}

kable(absorb$item.stats[-c(2:4)], digits=2, caption="Absorption Items")

```


```{r rdrop5, echo=FALSE, warning=FALSE, message=FALSE}

kable(vigor$item.stats[-c(2:4)], digits=2, caption="Vigor Items")

```


```{r rdrop6, echo=FALSE, warning=FALSE, message=FALSE}

kable(dedic$item.stats[-c(2:4)], digits=2, caption="Dedication Items")

```

## Substantive scale (cell level)

```{r rdrop7, echo=FALSE, warning=FALSE, message=FALSE}

kable(aff.abs$item.stats[-c(2:4)], digits=2, caption="Affective - Absorption Items")

```

```{r rdrop8, echo=FALSE, warning=FALSE, message=FALSE}

kable(aff.vig$item.stats[-c(2:4)], digits=2, caption="Affective - Vigor Items")

```

```{r rdrop9, echo=FALSE, warning=FALSE, message=FALSE}

kable(aff.ded$item.stats[-c(2:4)], digits=2, caption="Affective - Dedication Items")

```

```{r rdrop10, echo=FALSE, warning=FALSE, message=FALSE}

kable(cog.abs$item.stats[-c(2:4)], digits=2, caption="Cognitive - Absorption Items")

```

```{r rdrop11, echo=FALSE, warning=FALSE, message=FALSE}

kable(cog.vig$item.stats[-c(2:4)], digits=2, caption="Cognitive - Vigor Items")

```

```{r rdrop12, echo=FALSE, warning=FALSE, message=FALSE}

kable(cog.ded$item.stats[-c(2:4)], digits=2, caption="Cognitive - Dedication Items")

```

```{r rdrop13, echo=FALSE, warning=FALSE, message=FALSE}

kable(beh.abs$item.stats[-c(2:4)], digits=2, caption="Behavioral - Absorption Items")

```

```{r rdrop14, echo=FALSE, warning=FALSE, message=FALSE}

kable(beh.vig$item.stats[-c(2:4)], digits=2, caption="Behavioral - Vigor Items")

```

```{r rdrop15, echo=FALSE, warning=FALSE, message=FALSE}

kable(beh.ded$item.stats[-c(2:4)], digits=2, caption="Behavioral - Dedication Items")

```

# Qualitative item characteristics {#qualitative}

Item reading level information was performed via the package `quanteda` version `r packageVersion("quanteda")` [@R-quanteda] in `r R.version.string`. Two indices were investigated: "Flesch-Kincaid" is the same grade level index that's currently used by Microsoft Word [@kincaid_derivation_1975] whereas "Dale.Chall" reflects *N~wd~* ["difficulty" of words; @chall_dale_1995]. These indices are primarily influenced by two qualitative characteristics: [*N~wf~*](https://quanteda.io/reference/textstat_readability.html), which is the number of words matching the Dale-Chall List of 3000 "familiar words", and  [*N~wd~*](https://quanteda.io/reference/textstat_readability.html), which is the number of "difficult" words not matching the Dale-Chall list of "familiar" words. 

```{r quanteda, echo=FALSE, warning=FALSE, message=FALSE}
library(quanteda)
## library(quanteda.textstats)

data2 <- corpus(cond1, docid_field = "Condition1",
  text_field = "Condition1")

tab2 <- textstat_readability(data2,
                     measure = c("Flesch.Kincaid", "Dale.Chall"))

temp <- cbind(tab2,cond1)

temp$combo <- paste(temp$Substantive,temp$Attitudinal)

meanfk <- mean(tab2$Flesch.Kincaid, na.rm=TRUE)
meandc <- mean(tab2$Dale.Chall, na.rm=TRUE)
sdfk <- sd(tab2$Flesch.Kincaid, na.rm=TRUE)
sddc <- sd(tab2$Dale.Chall, na.rm=TRUE)

```

The average Flesch-Kincaied (e.g., reading grade) was `r round(meanfk,2)` (*sd* = `r round(sdfk,2)`). The average Dale-Chall index was `r round(meandc,2)` (*sd* = `r round(sddc,2)`).

## Frequency distributions by dimension

```{r scale, echo=FALSE, warning=FALSE, message=FALSE}

library(ggplot2)

ggplot(temp, aes(x = Dale.Chall, fill = combo)) +
  geom_density(alpha = .5) +
  xlab("Dale Chall Index") + labs(fill = "Att/Engage\nInteraction") +
  ggtitle("Infrequent word incidence across substantive/attitudinal dimension combinations")

ggplot(temp, aes(x = Dale.Chall, fill = Substantive)) +
  geom_density(alpha = .5) +
  xlab("Dale Chall Index") + labs(fill = "Engagement\nDimension") +
  ggtitle("Infrequent word incidence across substantive dimensions")

ggplot(temp, aes(x = Dale.Chall, fill = Attitudinal)) +
  geom_density(alpha = .5) +
  xlab("Dale Chall Index") + labs(fill = "Attitudinal\nDimension") +
  ggtitle("Infrequent word incidence across attitudinal dimensions")

ggplot(temp, aes(x = Flesch.Kincaid, fill = combo)) +
  geom_density(alpha = .5) +
  xlab("Flesch-Kincaid Index") + labs(fill = "Att/Engage\nInteraction") +
  ggtitle("Item reading level estimates across substantive/attitudinal dimension combinations")

ggplot(temp, aes(x = Flesch.Kincaid, fill = Substantive)) +
  geom_density(alpha = .5) +
  xlab("Flesch-Kincaid Index") + labs(fill = "Engagement\nDimension") +
  ggtitle("Item reading level estimates across substantive dimensions")

ggplot(temp, aes(x = Flesch.Kincaid, fill = Attitudinal)) +
  geom_density(alpha = .5) +
  xlab("Flesch-Kincaid Index") + labs(fill = "Attitudinal\nDimension") +
  ggtitle("Item reading level estimates across attitudinal dimensions")

```

## Tables of qualitative indices

```{r freqtables, echo=FALSE, warning=FALSE, message=FALSE, fig.height=3}

temp2 <- temp[,c(1,5,6,2:3)]

tab3 <- temp2[order(-temp2$Flesch.Kincaid),]
knitr::kable(tab3, digits=2, caption = "Organized by Flesch-Kincaid aka Reading Level")

tab4 <- temp2[order(temp2$Dale.Chall),]
knitr::kable(tab4, digits=2, caption = "Organized by Dale Chall aka includes Difficult Words")

```