i. vs c.

[thierry bouchet]

Could someone explain me what is the difference between i.variable and c.variable?

[Carlo Lazzaro]

Thierry:
welcome to this forum.
Short reply: -i.- means categorical variable, whereas -c.- means continuous variable.
Longer (and possiby) more helpful reply: take a comprhensive look at -fvvarlist- help file and related entry in Stata .pdf manual.

[William Lisowski]

The output of help factor variables has a good explanation of Stata's "factor variable" notation. While you're at it, the output of help tsvarlist will explain the similar "time series" notation like l. and f. .

[Sarah Mj]

c. for continuous variables do not distinguish between normal distribution and not normal distribution and testing it only takes a t-test regardless of the normality.
is there something other than c. distinguish between them?

[Carlo Lazzaro]

Sarah:
1) this is not what c. was developed for;
2) if you fear that the samples of your -ttest- do not come from normal distributions, you can compare the results of the parametric -ttest- with the ones of a boostrap -ttest- (see -bootstrap entry in Stata .pdf manual);
3) that said, the parametric -ttest- is robust to departure from normality of the samle under comparison.

[Maxence Morlet]

In addition to Carlo's great response, you may want to consider other comparison tests, which compare the median I believe, eg. Kruskal-Wallis or Wilcoxon if my memory serves. For correlational analysis, Spearman correlation is (I believe) nonparametric and may be interesting as well, in a non-normality context.