1808 summary updates

corestats.Rproj

@@ -14,7 +14,3 @@ LaTeX: pdfLaTeX
 
 AutoAppendNewline: Yes
 StripTrailingWhitespace: Yes
-
-PythonType: conda
-PythonVersion: 3.9.12
-PythonPath: ~/opt/miniconda3/bin/python3.9

materials/cs1_practical_one-sample.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -909,9 +909,11 @@ In terms of choosing between the two test we can see that both meet their respec
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: callout-note
 -   One-sample tests are used when you have a single sample of continuous data
 -   The t-test assumes that the data are normally distributed and independent of each other
 -   A good way of assessing the assumption of normality is by checking the data against a Q-Q plot

materials/cs1_practical_two-samples-paired.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -704,9 +704,11 @@ As Jeremy Clarkson [would put it](https://www.quotes.net/mquote/941330):
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: {.callout-note}
 -   Paired t-tests are used when you have two paired samples of continuous data, which are normally distributed and have equal variance
 -   A good way of assessing the assumption of normality is by checking the data against a Q-Q plot
 - We can check equality of variance (homoscedasticity) with Bartlett's (normal data) or Levene's (non-normal data) test

materials/cs1_practical_two-samples.qmd

@@ -17,8 +17,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -888,9 +888,11 @@ This gives us exactly the same conclusion that we got from the two-sample t-test
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: {.callout-note}
 -   Student's t tests are used when you have two samples of continuous data, which are normally distributed, independent of each other and have equal variance
 -   A good way of assessing the assumption of normality is by checking the data against a Q-Q plot
 - We can check equality of variance (homoscedasticity) with Bartlett's (normal data) or Levene's (non-normal data) test

materials/cs2_practical_anova.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -1012,9 +1012,11 @@ Then we should probably have just skipped the one-way ANOVA test entirely and ju
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: callout-note
 -   We use an ANOVA to test if there is a difference in means between multiple continuous response variables
 -   We check assumptions with diagnostic plots and check if the residuals are normally distributed
 -   We use post-hoc testing to check for significant differences between the group means, for example using Tukey's range test

materials/cs2_practical_kruskal-wallis.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -414,9 +414,11 @@ In this case we should not be doing any post-hoc testing, because we did not det
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: callout-note
 -   We use a Kruskal-Wallis test to see if there is a difference in medians between multiple continuous response variables
 -   We assume parent distributions have the same shape; each data point is independent and the parent distributions have the same variance
 -   We test for equality of variance using Levene's test

materials/cs3_practical_correlations.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -626,9 +626,11 @@ Well done, [Mr. Spearman](https://en.wikipedia.org/wiki/Charles_Spearman).
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: callout-note
 -   Correlation is the degree to which two variables are linearly related
 -   Correlation does not imply causation
 -   We can visualise correlations by plotting variables against each other or creating heatmap-type plots of the correlation coefficients

materials/cs3_practical_linear-regression.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -661,9 +661,11 @@ Neither of these solutions can be tackled with the knowledge that we have so far
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: callout-note
 -   Linear regression tests if a linear relationship exists between two or more variables
 -   If so, we can use one variable to predict another
 -   A linear model has an intercept and slope and we test if the slope differs from zero

materials/cs4_practical_linear-regression-grouped-data.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -983,9 +983,11 @@ So now we know that `yarrow` is a significant predictor of `yield` and we're hap
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: callout-note
 - A linear regression analysis with grouped data is used when we have one categorical and one continuous predictor variable, together with one continuous response variable
 - We can visualise the data by plotting a regression line together with the original data
 - When performing an ANOVA, we need to check for interaction terms

materials/cs4_practical_two-way-anova.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -921,9 +921,11 @@ So we do appear to have a significant interaction between `water` and `shade` as
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: callout-note
 - A two-way ANOVA is used when there are two categorical variables and a single continuous variable
 - We can visually check for interactions between the categorical variables by using interaction plots
 - The two-way ANOVA is a type of linear model and assumes the following:

materials/cs5_practical_model-comparisons.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -562,9 +562,11 @@ Our minimal model is thus:
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: {.callout-note}
 - We can use Backwards Stepwise Elimination (BSE) on a full model to see if certain terms add to the predictive power of the model or not
 - The AIC allows us to compare different models - if there is a difference in AIC of more than 2 between two models, then the smallest AIC score is more supported
 :::

materials/cs5_practical_multiple-linear-regression.qmd

@@ -16,8 +16,8 @@ import shutup;shutup.please()
 exec(open('setup_files/setup.py').read())
 ```
 
-::: callout-tip
-## Learning outcomes
+::: {.callout-tip}
+#### Learning outcomes
 
 **Questions**
 
@@ -1082,9 +1082,11 @@ The diagnostic plots here look rather similar to the ones we generated for the a
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: {.callout-note}
 
 - We can define a linear model with any combination of categorical and continuous predictor variables
 - Using the coefficients of the model we can construct the linear model equation

materials/cs6_practical_power-analysis.qmd

@@ -21,7 +21,7 @@ np.seterr(all='ignore')
 ```
 
 ::: {.callout-tip}
-## Learning outcomes
+#### Learning outcomes
 
 **Questions**
 
@@ -1185,9 +1185,11 @@ We get a denominator df of 116, which means that we would need at least 123 part
 :::
 :::
 
-## Key points
+## Summary
+
+::: {.callout-tip}
+#### Key points
 
-::: {.callout-note}
 - Power is the capacity of a test to detect significant results and is affected by
     1. the effect size
     2. sample size