How to t test

# How to t test
### Peter Higgins

---

### We will start with the tidy version from the {infer} package: t_test()

#### Notice that you have to state the order of the two levels of your grouping variable
---

```r
*prostate
```
]
 
.panel2-ttest1-auto[

```
# A tibble: 316 x 20
   RBC.Age.Group Median.RBC.Age   Age    AA FamHx  PVol  TVol T.Stage   bGS
           <dbl>          <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>   <dbl> <dbl>
 1             3             25  72.1     0     0  54       3       1     3
 2             3             25  73.6     0     0  43.2     3       2     2
 3             3             25  67.5     0     0 103.      1       1     3
 4             2             15  65.8     0     0  46       1       1     1
 5             2             15  63.2     0     0  60       2       1     2
 6             3             25  65.4     0     0  45.9     2       1     1
 7             3             25  65.5     1     0  42.6     2       1     1
 8             1             10  67.1     0     0  40.7     3       1     1
 9             1             10  63.9     0     0  45       2       1     1
10             2             15  63       1     0  67.6     2       1     2
# … with 306 more rows, and 11 more variables: `BN+` <dbl>,
#   OrganConfined <dbl>, PreopPSA <dbl>, PreopTherapy <dbl>, Units <dbl>,
#   sGS <dbl>, AnyAdjTherapy <dbl>, AdjRadTherapy <dbl>, Recurrence <dbl>,
#   Censor <dbl>, TimeToRecurrence <dbl>
```
]

---
count: false
 
Tidy version from the {infer} package
.panel1-ttest1-auto[

```r
prostate %>%
* t_test(TVol ~ AA,
*        order = c("0", "1"))
```
]
 
.panel2-ttest1-auto[

```
# A tibble: 1 x 6
  statistic  t_df p_value alternative lower_ci upper_ci
      <dbl> <dbl>   <dbl> <chr>          <dbl>    <dbl>
1    -0.683  79.2   0.496 two.sided     -0.271    0.132
```
]

---
count: false
 
Tidy version from the {infer} package
.panel1-ttest1-auto[

```r
prostate %>%
  t_test(TVol ~ AA,
         order = c("0", "1"))
# Interpreting the results
# The t statistic is first
# followed by degrees of freedom
# then the p value
# the default alternative: two.sided
# then the confidence bounds
# output is a tibble so that it is
# easy to use these results
```
]
 
.panel2-ttest1-auto[

```
# A tibble: 1 x 6
  statistic  t_df p_value alternative lower_ci upper_ci
      <dbl> <dbl>   <dbl> <chr>          <dbl>    <dbl>
1    -0.683  79.2   0.496 two.sided     -0.271    0.132
```
]

---

### Now we will use the baseR version: t.test()

#### Notice that you have to use   `data = .`
#### It is not quite as pipe-friendly
#### But you do not have to specify the order of the two levels.
---

```r
*prostate
```
]
 
.panel2-ttest2-auto[

---
count: false
 
Base R Version
.panel1-ttest2-auto[

```r
prostate %>%
* t.test(TVol ~ AA, data = .)
```
]
 
.panel2-ttest2-auto[

```

Welch Two Sample t-test

data:  TVol by AA
t = -0.6833, df = 79.184, p-value = 0.4964
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 -0.2708947  0.1324320
sample estimates:
mean in group 0 mean in group 1 
       2.081712        2.150943 
```
]

---
count: false
 
Base R Version
.panel1-ttest2-auto[

```r
prostate %>%
  t.test(TVol ~ AA, data = .)

# Interpreting the results
# The t statistic is first
# followed by degrees of freedom
# then the p value
# then the alternative hypothesis
# then the confidence bounds
# then the mean mpg for each group
```
]
 
.panel2-ttest2-auto[

```

Welch Two Sample t-test

---

### Which version of the t test do you like better?

Discuss.