Two-sided p-value in anova

[Fahmida Huq]

Hello,
I need to make similar table, showing pesticide concentration by demographic characteristics in a population of pregnant women. I was wondering how to get the corresponding p-values for each of the categories individually? For example, how can I get p-values for age 18-24 years, 25-29 years, 30-34 years separately, not just as a summary p-value for the overall age group?

Thanks,
Fahmida.

Table. ΣPBDE serum concentrations (ng/g lipids) around the 27th week of gestation by demographic characteristics in a population of pregnant women
Characteristic	No. (%)a	Geometric mean (GSD)
Age (years)
18–24	129 (48)	27.3 (2.9)
25–29	90 (33)	23.8 (2.5)
30–34	34 (13)	27.2 (2.7)
35–45	17 (6)	35.6 (2.7)
Race/ethnicity
Caucasian	7 (3)	84.1 (4.3)**
Latino	257 (95)	25.3 (2.6)
Other	6 (2)	54.3 (2.8)
Education
≤ 6th grade	110 (41)	20.8 (2.3)#
7–12th grade	97 (36)	26.8 (2.8)
≥ High school diploma	63 (23)	39.5 (2.9)

[Matt Warkentin]

A one-way ANOVA performs an omnibus test for the between vs. within group variances for all groups. Thus, a single p-value (and F-statistic) are provided to show that all the groups are equal/not equal. To get individual group-wise (pair-wise) p-values, you can add the options bonferroni, sidak, or scheffe to the oneway command in Stata to get pairwise comparisons and p-values.

[Fahmida Huq]

Hello,

Did you mean this type of comparisons?

Comparison of log10-transformed ppDDE (ng/g lipids - impute 0 values)
by Child Sex
(Bonferroni)
Row Mean-|
Col Mean | Boy
---------+-----------
Girl | .067668
| 0.176
In that case, can you please give me any links/docs to interpret the results? I'm not very clear about the concept. Thanks.

Fahmida

[Carlo Lazzaro]

Fahmida:
if you're interested in age groups only, you can type:

oneway <depvar> age-group, bonferroni *if you do not have it yet, you should create a variable -age-group- that collects the age categories of your patients*

Otherwise, if you're interested in all the predictors reported in table you pasted in your original post, you should go -regress-.
The results you posted mean that there's no evidence of a statistically significant difference in the mean of -log10-transformed ppDDE- for boys and girls (please note that this stuff is covered in any decent statistics textbook).
As an aside, for the future please use CODE delimiters to share what you typed and what Stata gave you back. Thanks.