---
title: "Analysis of Academic Information System Acceptance in Telkom University"
author: "Nikolaus Aloysius - 29018020"
date: "29/4/2019"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

## Library
```{r}
library(lavaan)
library(semPlot)
library(semTools)
```

## Upload Data
```{r}
tam_dataset <- read_xlsx("~/Downloads/dataskripsi.xlsx")
names(tam_dataset)
```

## variabletable
```{r}
varTable(tamdata, ov.names = names(tamdata), ov.names.x = NULL, 
    ordered = NULL, factor = NULL, as.data.frame. = TRUE)
```

## Construct Model
```{r}
tam <- " experience                =~ EXP1
               perceived_ease_of_use     =~ PEOU1 + PEOU2 + PEOU3 + PEOU4 + PEOU5 + PEOU6      
               perceived_usefulness   =~ PU1 + PU2 + PU3 + PU4 + PU5 + PU6
               information_system_usage  =~ ISU1 + ISU2 + ISU3 + ISU4 + ISU5 "
```

## Model Analysis
```{r}
tam_cfa <- cfa(tam, data=tam_dataset)
lavInspect(tam_cfa, what = "free", add.labels = TRUE, add.class = TRUE,
           list.by.group = TRUE,
           drop.list.single.group = TRUE)
```

## Result
```{r}
summary(tam_cfa, standardized = T, fit.measures = T, rsquare = T)
```

**Manifest Indicator Analysis**  
According to the result, 8 out of 18 manifest indicator score was below standard 0.7. Those variables are PEOU2, PEOU3, PEOU4, PU2, PU3, PU4, PU5, ISU3, ISU4, and ISU5. To do recovery, those variables must be drop from the equation.

## Re-construct Model
```{r}
tam2 <- " experience                =~ EXP1
               perceived_ease_of_use     =~ PEOU1 + PEOU5 + PEOU6      
               perceived_usefulness   =~ PU1 + PU6
               information_system_usage  =~ ISU1 + ISU2 "
```

## Model Re-analysis
```{r}
tam2_cfa <- cfa(tam2, data=tam_dataset)
lavInspect(tam2_cfa, what = "free")
```

## Result
```{r}
summary(tam2_cfa, standardized = T, fit.measures = T, rsquare = T)
```

**Goodness of Fit Analysis**

- The score of Comparative Fit Index (CFI)) should be were 0- 1. if the score was close to 1, it could be state that the construct model has good of fitness. in this estimation, CFI score score is 0.806 it means the model goodness of it is accepted.

**Badness of Fit analysis**

- Root mean square error of approximation (RMSEA) for assess badness of fit, the p-value must be under or equal to 0,05. the result show that score of p-value is 0.000 and it is below alpha 0.05. it means that the model is fit with no badness according to the analysis.

## PEOU Model
```{r}
se.model <- ' perceived_ease_of_use  =~ PEOU1 + PEOU5 + PEOU6 '
peou_model <- cfa(se.model, data=tam_dataset)
reliability(peou_model)
```
- The result of AVE score in perceived_ease_of_use is about 0.5519494. Rules of AVE state that value must be less than 0.5. It means the latent in this model is reliable.

## PU Model
```{r}
se.model <- ' perceived_usefulness  =~ PU1 + PU6 '
pu_model <- cfa(se.model, data=tam_dataset)
reliability(pu_model)
```
- The result of AVE score in Depresiveness is about 0.5646177 Rules of AVE state that value must be less than 0.5. It means the latent in this model is reliable.

## ISU Model
```{r}
se.model <- ' information_system_usage  =~ ISU1 + ISU2 '
isu_model <- cfa(se.model, data=tam_dataset)
reliability(isu_model)
```
- The result of AVE score in impulsiveness is about 0.6634211 rules of AVE state that value must be less than 0.5. It means the reliability for this latent variable is not acceptable.

## Plot of Structural Model
```{r}
semPaths(tam2_cfa, whatLabels = "std", edge.label.cex = 1)
```

## Structural Equation Modelling
**Model Specification**
```{r}
test.model <- "experience               =~ EXP1
               perceived_usefulness      =~ PU1 + PU6
               perceived_ease_of_use     =~ PEOU1 + PEOU5 + PEOU6      
               information_system_usage  =~ ISU1 + ISU2 
               perceived_usefulness ~ experience
               perceived_ease_of_use ~ experience
               perceived_usefulness ~ perceived_ease_of_use
               information_system_usage ~ perceived_ease_of_use
               information_system_usage ~ perceived_usefulness"
```

**Model Analysis**
```{r}
testfit <- cfa(test.model, data=tam_dataset)
```

**Diagram Model**
```{r}
semPaths(testfit, whatLabels = "std", edge.label.cex = 1)
```