Below we provide two basic examples that use the R package condir. The first example regards data coming from a single group, and the second example data coming from two groups. More details about the package condir are described in the paper by: Krypotos, A. M., Klugkist, I., & Engelhard, I. M. (2017). Bayesian hypothesis testing for human threat conditioning research: An introduction and the condir R package. European Journal of Psychotraumatology, 8.
You can install condir via cran using the following command:
For loading condir, use the line below. Please note that for using condir you have to load it at every new R session.
Here we show a single group example. Specifically, we first simulate data from a normal distribution for two stimuli, the cs1 and the cs2.
These data correspond to the conditioned responses during the presentation of each stimulus (i.e., the cs1 and the cs2). For comparing the two stimuli, we can use the csCompare function in condir as follow.
$descriptives
vars n mean sd min max range se
cs1 1 50 4.21 4.66 -5.23 18.35 23.58 0.66
cs2 2 50 5.46 5.29 -7.11 17.27 24.38 0.75
$freq.results
method alternative WG1 WpG1 WG2 WpG2 null.value LCI
1 Paired t-test two.sided 0.9521536 0.04185158 0 0 0 -3.386249
HCI t.statistic df p.value cohenD cohenDM hedgesG hedgesGM
1 0.886061 -1.176018 49 0.2452697 -0.251108 small -0.2472448 small
$bayes.results
LNI HNI rscale bf10 bf01 propError
1 -Inf Inf 0.707 0.2944811 3.395803 0.0004878148
$res.out
$descriptives
vars n mean sd min max range se
cs1 1 49 3.92 4.23 -5.23 12.32 17.55 0.60
cs2 2 49 5.63 5.20 -7.11 17.27 24.38 0.74
$freq.results
method alternative WG1 WpG1 WG2 WpG2 null.value LCI
1 Paired t-test two.sided 0.9708112 0.2607728 0 0 0 -3.676637
HCI t.statistic df p.value cohenD cohenDM hedgesG hedgesGM
1 0.2543265 -1.750466 48 0.08642769 -0.3613867 small -0.3557105 small
$bayes.results
LNI HNI rscale bf10 bf01 propError
1 -Inf Inf 0.707 0.6383185 1.566616 0.0003070762
attr(,"class")
[1] "csCompare"
attr(,"class")
[1] "csCompare"
The data can be plotted with the csPlot function.
In order to make a basic report of the data, we use the csReport function.
We performed a two sided paired t-test. The results are t (48) = -1.75, p = 0.086, Cohen's d = -0.361 (small effect size).
We performed a two sided Bayesian t-test, with a Cauchy prior, with its width set to 0.707. The BF01 was: BF01(0.707) = 1.567. The BF10 was: BF10(0.707) = 0.638.Main analyses
We performed a two sided paired t-test. The results are t (49) = -1.176, p = 0.245, Cohen's d = -0.251 (small effect size).
We performed a two sided Bayesian t-test, with a Cauchy prior, with its width set to 0.707. The BF01 was: BF01(0.707) = 3.396. The BF10 was: BF10(0.707) = 0.294.
**Outliers report**
We performed a two sided paired t-test. The results are t (48) = -1.75, p = 0.086, Cohen's d = -0.361 (small effect size).
We performed a two sided Bayesian t-test, with a Cauchy prior, with its width set to 0.707. The BF01 was: BF01(0.707) = 1.567. The BF10 was: BF10(0.707) = 0.638.
Lastly, the csSensitivity function can be used for a sensitivity analysis, with the csRobustnessPlot function plotting the results.
The results are now reported with the csReport function.
We performed a Sensitivity Analysis using the scaling factors: 0.707, 1, 1.41. The results for BF01 were: BF01(0.707) = 1.567, BF01(1) = 2.079, BF01(1.41) = 2.827 respectively. The results for BF10 were: BF10(0.707) = 0.638, BF10(1) = 0.481, BF10(1.41) = 0.354 respectively.Sensitivity analyses
We performed a Sensitivity Analysis using the scaling factors: 0.707, 1, 1.41. The results for BF01 were: BF01(0.707) = 3.396, BF01(1) = 4.62, BF01(1.41) = 6.378 respectively. The results for BF10 were: BF10(0.707) = 0.294, BF10(1) = 0.216, BF10(1.41) = 0.157 respectively.
**Sensitivity analyses - Outliers report**
We performed a Sensitivity Analysis using the scaling factors: 0.707, 1, 1.41. The results for BF01 were: BF01(0.707) = 1.567, BF01(1) = 2.079, BF01(1.41) = 2.827 respectively. The results for BF10 were: BF10(0.707) = 0.638, BF10(1) = 0.481, BF10(1.41) = 0.354 respectively.
The same steps as above are used for the two group example. The only difference is that we now have to define the group allocation by using the group argument – see the first line below. Apart from that, the code is the same as in the example above. That is why we provide the code for this example in a single chunk of code.
$descriptives
$descriptives$`1`
vars n mean sd min max range se
cs1 1 25 3.07 4.86 -5.23 18.35 23.58 0.97
cs2 2 25 5.16 4.89 -4.45 17.27 21.72 0.98
cs3 3 25 -2.08 6.85 -12.36 21.34 33.70 1.37
$descriptives$`2`
vars n mean sd min max range se
cs1 1 25 5.34 4.23 -1.87 12.32 14.18 0.85
cs2 2 25 5.76 5.74 -7.11 14.79 21.90 1.15
cs3 3 25 -0.42 8.18 -16.53 17.40 33.93 1.64
$freq.results
method alternative WG1 WpG1 WG2 WpG2
1 Welch Two Sample t-test two.sided 0.8810523 0.007265277 0.9687348 0.6131729
null.value LCI HCI t.statistic df p.value cohenD
1 0 -5.961606 2.628604 -0.7807465 46.56861 0.4389025 -0.2509847
cohenDM hedgesG hedgesGM
1 small -0.249059 small
$bayes.results
LNI HNI rscale bf10 bf01 propError
1 -Inf Inf 0.707 0.3630444 2.754484 7.354763e-05
$res.out
$descriptives
$descriptives$`1`
vars n mean sd min max range se
cs1 1 24 2.44 3.76 -5.23 9.17 14.40 0.77
cs2 2 24 5.50 4.69 -4.45 17.27 21.72 0.96
cs3 3 24 -3.06 4.91 -12.36 7.07 19.43 1.00
$descriptives$`2`
vars n mean sd min max range se
cs1 1 25 5.34 4.23 -1.87 12.32 14.18 0.85
cs2 2 25 5.76 5.74 -7.11 14.79 21.90 1.15
cs3 3 25 -0.42 8.18 -16.53 17.40 33.93 1.64
$freq.results
method alternative WG1 WpG1 WG2 WpG2
1 Welch Two Sample t-test two.sided 0.9759581 0.8113906 0.9687348 0.6131729
null.value LCI HCI t.statistic df p.value cohenD
1 0 -6.522944 1.237837 -1.376793 39.59676 0.1763113 -0.3612266
cohenDM hedgesG hedgesGM
1 small -0.3583972 small
$bayes.results
LNI HNI rscale bf10 bf01 propError
1 -Inf Inf 0.707 0.6155059 1.62468 7.331263e-05
attr(,"class")
[1] "csCompare"
attr(,"class")
[1] "csCompare"
We performed a two sided Welch two sample t-test. The results are t (39.597) = -1.377, p = 0.176, Cohen's d = -0.361 (small effect size).
We performed a two sided Bayesian t-test, with a Cauchy prior, with its width set to 0.707. The BF01 was: BF01(0.707) = 1.625. The BF10 was: BF10(0.707) = 0.616.Main analyses
We performed a two sided Welch two sample t-test. The results are t (46.569) = -0.781, p = 0.439, Cohen's d = -0.251 (small effect size).
We performed a two sided Bayesian t-test, with a Cauchy prior, with its width set to 0.707. The BF01 was: BF01(0.707) = 2.754. The BF10 was: BF10(0.707) = 0.363.
**Outliers report**
We performed a two sided Welch two sample t-test. The results are t (39.597) = -1.377, p = 0.176, Cohen's d = -0.361 (small effect size).
We performed a two sided Bayesian t-test, with a Cauchy prior, with its width set to 0.707. The BF01 was: BF01(0.707) = 1.625. The BF10 was: BF10(0.707) = 0.616.
We performed a Sensitivity Analysis using the scaling factors: 0.707, 1, 1.41. The results for BF01 were: BF01(0.707) = 1.567, BF01(1) = 2.079, BF01(1.41) = 2.827 respectively. The results for BF10 were: BF10(0.707) = 0.638, BF10(1) = 0.481, BF10(1.41) = 0.354 respectively.Sensitivity analyses
We performed a Sensitivity Analysis using the scaling factors: 0.707, 1, 1.41. The results for BF01 were: BF01(0.707) = 3.396, BF01(1) = 4.62, BF01(1.41) = 6.378 respectively. The results for BF10 were: BF10(0.707) = 0.294, BF10(1) = 0.216, BF10(1.41) = 0.157 respectively.
**Sensitivity analyses - Outliers report**
We performed a Sensitivity Analysis using the scaling factors: 0.707, 1, 1.41. The results for BF01 were: BF01(0.707) = 1.567, BF01(1) = 2.079, BF01(1.41) = 2.827 respectively. The results for BF10 were: BF10(0.707) = 0.638, BF10(1) = 0.481, BF10(1.41) = 0.354 respectively.