Creating a graph with hours worked and wage

[Chris Rooney]

Hello all,

Let's say I have two variables - weekly hours worked (hourly_work) and hourly wage (hourly_wage).

What I'd like to do is plot the mean hours of work by wage level (basically across all wage levels), with mean hours of work on the y-axis and hourly wage on the axis.

What I'm struggling with is creating a mean variable for each wage level - of course there an infinite number of wage levels and sometimes the mean of a wage level will be a sample of one. But that is not my concern. I can create a mean with a group but not sure how to do it on an individual level.

Thanks

[Maarten Buis]

// open example data
sysuse nlsw88, clear

// group wage in 10 deciles
xtile wage_gr = wage, nq(10)

// for plotting assign them the middle wage
bys wage_gr (wage) : replace wage_gr = (wage[_N] - wage[1])/2 + wage[1]

// create mean hours per wage group
bys wage_gr : egen hours_m = mean(hours)

// plot the two variables
twoway scatter hours_m wage_gr

[Nick Cox]

Maarten's advice is always good, but I don't quite share his enthusiasm for decile bins.

But additionally I'd use a scatter plot smoother. If your dataset is moderate or large in size, expect results to be a little slow.

I have struggled often to move away from lpoly's defaults but you may want to try mine!

See http://www.statalist.org/forums/foru...ial-regression