ttest code multiple variables that define group

[Imtiaz Bhayat]

Hello.
I cant seem to work out the coding that replicates the following function, which when copied from the 'command copied from clipboard', looks like:
'time_ann_dif PriceSensitive, sort : ttest Volume, by(Pre_or_Post_Colo) unequal'. Does someone know how this can be done?

Essentially I want to run a ttest but with multiple variable, then save the output using putexcel. I know how to use putecel but cant get the code working for multiple variables.
As an FYI time_ann_dif has values that range from -30 to 30, while PriceSensitive has values that are TRUE or FALSE.

Thanks

[Carlo Lazzaro]

Imtiaz:
welcome to this forum.
As far as the first part of your query is concerned, do you mean something along the following lines?

. sysuse auto, clear
. egen wanted=group(foreign rep78) if rep78==3

. ttest price, unequal by(wanted)

Two-sample t test with unequal variances
------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
       1 |      27    6607.074    704.6111    3661.267    5158.725    8055.423
       2 |       3    4828.667    742.2489    1285.613    1635.027    8022.306
---------+--------------------------------------------------------------------
combined |      30    6429.233    643.5995     3525.14    5112.924    7745.542
---------+--------------------------------------------------------------------
    diff |            1778.407    1023.431               -655.8428    4212.658
------------------------------------------------------------------------------
    diff = mean(1) - mean(2)                                      t =   1.7377
Ho: diff = 0                     Satterthwaite's degrees of freedom =  6.80377

    Ha: diff < 0                 Ha: diff != 0                 Ha: diff > 0
 Pr(T < t) = 0.9365         Pr(|T| > |t|) = 0.1271          Pr(T > t) = 0.0635

.

In addition, the code you copied and pasted should probably be:

by time_ann_dif PriceSensitive, sort : ttest Volume, by(Pre_or_Post_Colo) unequal'

[Imtiaz Bhayat]

Hi Carlo,

Thanks for your reply.
Essentially I would like to perform multiple ttests and capture the p values to excel. The code below works however I now want to add an element to the test where for each 'i', I would like to further stratify the data by adding an additional criteria which is PriceSensitive (True and False).

So in essence the test below runs 61 ttest (-30 to 30), I now want to double that with the inclusion of the additional PriceSensitive component (True and False).

When I use the Statistics > Summaries, Tables, Tests > Classic tests of hypothesis > t test (mean comparison test)....I can do the test I need which represents the code you noted ie
by time_ann_dif PriceSensitive, sort : ttest Volume, by(Pre_or_Post_Colo) unequal'

I now don't know how to add the PriceSensitive component to the code below so that it replaces the task performed when using the dropdowns.

------------------------------------------------------------------------------------------------------
**ttest Volume, by (Pre_or_Post_Colo), by min unequal
putexcel set VolumeTtest
forvalues i = -30(1)30 {
ttest Volume if time_ann_dif == `i', by(Pre_or_Post_Colo) unequal
local counter1 = `i' + 31
putexcel A`counter1' = `r(t)'
putexcel B`counter1' = `r(p)'
}
encode

[Imtiaz Bhayat]

Hi again Carlo,

It took me a while to absorb your suggestion and understand the grouping concept.

When I try and create a new var I get a type mismatch r109 message. Is it because the time_ann_diff data type is a double and PriceSensitive is a type long?

[Carlo Lazzaro]

Imtiaz:
It's difficult to say wthout knowing your data and codes.
Anyway, the definition of -error109- is the following one (source: -error- entry, Stata .pdf manual)

109. type mismatch

In an expression, you attempted to combine a string and numeric subexpression in a logically impossible way. For

instance, you attempted to subtract a string from a number or you attempted to take the substring of a number.

[Imtiaz Bhayat]

Thanks Carlo. Your egen suggestion works up to the point of the if statement, in my code.

When I add the following if PriceSensitive == FALSE or even the code if PriceSensitive == 2 that's when the 109 error appears.

Data types: the time_ann_diff data type is a double (integer) and PriceSensitive is a type long (visibly written as true or false but I can see that the system reflects this as 1 and 2).

I apologise in that its probably a really basic question but I am literally 2 weeks into my Stata journey!

[Carlo Lazzaro]

Imitiaz:
it would be really useful if you could share what you typed and what Stata gave you back, along with an excerpt of your dataset via -dataex- (as per FAQ). Thanks.
In addition, I fail to get how yoo coded -FALSE- in your -if- qualifier clause.

[Imtiaz Bhayat]

Carlo - I really appreciated your advice.
It all worked fine