Reporting Data: Numerical Flashcards
What are the two rules we adhere to during hypothesis tests?
- Must Report exact p-values
-Never use inequalities(e.g., p = 0.125 vs p > 0.05; p = 0.012 vs p < 0.05)
-unless p is very small(e.g., p = 0.0000001 can be shown as p < 0.001) - Must co-report that affect p-values
-Sample size (e.g., n = 25)
-Effect size (e.g., differences in means)
-Precision (confidence intervals)
Bright-line threshold?
It is p<0.05
A bright-line rule (or bright-line test) is a clearly defined rule or standard, composed of objective factors.
What is an alternative to bright-line hypothesis?
Another approach is to let the reader interpret the result
CI relationship to Precision?
CI is wide and thus decreases precision
CI is narrow and thus increases preciison
Three ways to report results?
- Test Statistic(t-test)
- Analysis of variance(ANOVA)
- Simple Linear Regression
What kind of errors does Bright-line analysis bring?
“Bright-line” analysis brings in type I and type II errors
Type I error?
When the null hypothesis is true, and you reject it
Type II error?
When the null hypothesis is false, and you fail to reject it
Probability of type I/II error to satistical significance?
Probability of a Type I error is equal to the level set for statistical
significance
Probability of a Type II error is dependent on the statistical power
Accuracy vs Precision?
Image result for accuracy vs precision
Precision and accuracy are two ways that scientists think about error. Accuracy refers to how close a measurement is to the true or accepted value. Precision refers to how close measurements of the same item are to each other.
What is precision as high as?
Precision is only as high as that of the instrument used to make the measurement
When is precision inflation seen?
Precision “inflation” often seen when making calculations
What is precision limited by?
When making calculations, precision is limited by the instrument with the lowest precision
What do we limit the precision to?
Limit the level of precision to what is useful to the reader
Reporting results and SI units?
Use SI units unless there is a historical unit that is used by convention
Reporting results and large numbers?
Do not split large numbers using commas
Reporting results and commas/decimals?
Do not use commas instead of decimal places
Reporting results and summary measurement?
Indicate how summary measures are presented
Anatomy of a good figure for a continuous vs continuous scale? SCATTER PLOTS
Figure Heading
Legend
Axis
Axis Scale
Axis length
Key site
Have no figure heading
The legend goes under the figure
Use appropriate axis labels with measurement units
Use the appropriate axis scale. It should not be too cluttered.
Set axis length to avoid too much white space
Key within the plot area(unless unavoidable) or directly label the data
Anatomy of a good figure for a continuous vs categorical scale? BAR CHARTS
Plot to use
Figure head position
App for box-and-whisker plot
Box-and-whisker plot
Positioned below
Ms Excel is useless at drawing box-and-whisker plots. Rather use the free app: poltr
Should we use dynamite plots for bar charts
NO
They hide important information
When do use bar charts?
For categorical scale: Frequency or proportion plots
What should you use when you are trying to get across exact figures?
Choose a table over a figure
Anatomy of a good table?
Vertical Lines
Horizontal Lines
Information in column
Extra information
Title
Numbers
No vertical Lines
Few horizontal lines. No horizontal lines in body
Always put things you want to compare in columns
Extra information must be put in the footnote under the table
Put the short title at top of the table
Rightly justify numbers. Do not use commas or non-divisible figures if not necessary