In Richard McElreath's Statistical Rethinking, he talks about the dangers of "stargazing". That is, running a regression, and looking only at the *'s, which signify the level of statistical significance of a coefficient. To help you avoid this sort of analysis, I've recruited the help of several shruggies ( ¯\_(ツ)_/¯), with various arm lengths to remind you looking only at the *'s isn't a very good idea. (Original idea due to Kevin Reuning.)
You can find the function in ShruggieRegression.R.
For example, below, you'd be tempted to stare at all those stars:
> data("iris")
> summary(lm(Sepal.Length ~ . , data = iris))
Call:
lm(formula = Sepal.Length ~ ., data = iris)
Residuals:
Min 1Q Median 3Q Max
-0.79424 -0.21874 0.00899 0.20255 0.73103
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 2.17127 0.27979 7.760 1.43e-12 ***
Sepal.Width 0.49589 0.08607 5.761 4.87e-08 ***
Petal.Length 0.82924 0.06853 12.101 < 2e-16 ***
Petal.Width -0.31516 0.15120 -2.084 0.03889 *
Speciesversicolor -0.72356 0.24017 -3.013 0.00306 **
Speciesvirginica -1.02350 0.33373 -3.067 0.00258 **
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 0.3068 on 144 degrees of freedom
Multiple R-squared: 0.8673, Adjusted R-squared: 0.8627
F-statistic: 188.3 on 5 and 144 DF, p-value: < 2.2e-16
Fortunately, we have our friends the shruggies to help "clarify" your regression output:
> ShruggieRegression(summary(lm(Sepal.Length ~ . , data = iris)))
Call:
lm(formula = Sepal.Length ~ ., data = iris)
Residuals:
Min 1Q Median 3Q Max
-0.79424 -0.21874 0.00899 0.20255 0.73103
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 2.17127 0.27979 7.760 1.43e-12 ¯\_(ツ)_/¯
Sepal.Width 0.49589 0.08607 5.761 4.87e-08 ¯\_(ツ)_/¯
Petal.Length 0.82924 0.06853 12.101 < 2e-16 ¯\_(ツ)_/¯
Petal.Width -0.31516 0.15120 -2.084 0.03889 ¯\___(ツ)___/¯
Speciesversicolor -0.72356 0.24017 -3.013 0.00306 ¯\__(ツ)__/¯
Speciesvirginica -1.02350 0.33373 -3.067 0.00258 ¯\__(ツ)__/¯
---
Signif. codes: 0 ‘¯\_(ツ)_/¯’ 0.001 ‘¯\__(ツ)__/¯’ 0.01 ‘¯\___(ツ)___/¯’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 0.3068 on 144 degrees of freedom
Multiple R-squared: 0.8673, Adjusted R-squared: 0.8627
F-statistic: 188.3 on 5 and 144 DF, p-value: < 2.2e-16
Isn't that better?
(Yes, this is a joke, and I make no claims to the accuracy or correctness of the code. I used regex to modify the outputs instead of actually rewriting all the summary methods for goodness' sake.)