difftab

Syntax

    difftab store name , varlist(fvvarlist) [i(numlist)]

    difftab write namelist [using filename] [, esttab_options ]

Options

• varlist(fvvarlist) 3-way interaction factor variable list (e.g. var1##var2##var3).
• i(numlist) List of levels for each variable in varlist(fvvarlist) to be considered “on”. Default is i(1 1 1).
• esttab_options Any valid options specified in esttab.

Description

difftab is a post-estimation command that takes the results of a regression command containing varlist(fvvarlist) and returns a 2x2x2 table containing the predicted value of Y for every possible linear combination of factor levels, and their differences. For example, see Duflo (2001), Table 3.

difftab store is the equivalent post-estimation command to eststo, storing the factor combinations in matrix name.

difftab write writes matrices prepared using difftab store using esttab. Any valid esttab or estout options can be added to difftab write. Users can specify a file with {hepl using}, or choose not to specify a filename (this will write the table to the Stata output window). difftab write is, at its core, a wrapper for esttab.

Formatting

difftab write will return a default table with the following equation, variable, and model labels:

    . reg y var1##var2##var3
    . difftab store est1, varlist(var1##var2##var3)
    . difftab write est1

    +----------------------------------------+
    |               1.var1  0.var1  D.var1   |
    |----------------------------------------|
    | 1.var3                                 |
    |   1.var2                               |
    |   0.var2                               |
    |   D.var2                               |
    |----------------------------------------|
    | 0.var3                                 |
    |   1.var2                               |
    |   0.var2                               |
    |   D.var2                               |
    |----------------------------------------|
    | D.var3                                 |
    |   1.var2                               |
    |   0.var2                               |
    |   D.var2                               |
    +----------------------------------------+ Where #.varname represents variable varname evaluated at #, and D.varname represents the difference, (1.varname - 0.varname).  difftab write will display the point estimate, as well as standard errors and p-values by defatult.

To change the labels for var1, use the mtitles() option in esttab. For example:

​

. difftab write est1, mlabels("On" "Off" "Difference")

To change the labels for var2, use the varlabels() option in esttab.

 . difftab write est1, varlabels(1.var2 "On" 0.var2 "Off" D.var2 "Difference")

To change the labels for var3, use the eqlabels() option in esttab.

    . difftab write est1, eqlabels("On" "Off" "Difference")

To remove the lines between panels, use the nolines option in esttab

    . difftab write est1, nolines

Examples

Load NLSW data:

    .  sysuse nlsw88, clear

Regress wage on 3 different demographic indicators:

    .  reg wage married##collgrad##union, r

Store results and write a basic table:

    .  difftab store est1, varlist(married##collgrad##union)
    .  difftab write est1

We can combine results from multiple models:

    .  reg hours married##collgrad##union, r
    .  difftab store est2, varlist(married##collgrad##union)
    .  difftab write est1 est2

Author

Aaron Wolf, Yale University aaron.wolf@yale.edu

References

Duflo, E. (2001). Schooling and labor market consequences of school construction in Indonesia: Evidence from an unusual policy experiment. American Economic Review, 91(4), 795–813. https://doi.org/10.1257/aer.91.4.795