Reporting proc surveylogistic results

[westbestern]

I have survey data and I want to do a bivariate logistic regression of age categories (18-29, 30-39, 40-49, 50+) on PrEP usage (yes, no). Here is my code below

proc surveylogistic data=dataname;
class AgeCat;
model takePrEP= AgeCat;
run;

I want to report the p-value, odds ratios, and confidence intervals for the AgeCat variable. Here is my output for this code. My questions are: what p-value would I report? Am I missing something in my code? 

Model Convergence Status
Convergence criterion (GCONV=1E-8) satisfied.

Model Fit Statistics
Criterion	Intercept Only	Intercept and Covariates
AIC	147.059	150.391
SC	150.252	163.163
-2 Log L	145.059	142.391

Testing Global Null Hypothesis: BETA=0
Test	F Value	Num DF	Den DF	Pr > F
Likelihood Ratio	0.89	3	177	0.4479
Score	0.85	3	177	0.4666
Wald	0.86	3	177	0.4646

Type 3 Analysis of Effects
Effect	F Value	Num DF	Den DF	Pr > F
AgeCat	0.86	3	177	0.4646

Analysis of Maximum Likelihood Estimates
Parameter		Estimate	Standard Error	t Value	Pr > |t|
NOTE: The degrees of freedom for the t tests is 179.
Intercept		1.6129	0.2857	5.65	<.0001
AgeCat	1	0.5619	0.3722	1.51	0.1329
AgeCat	2	-0.0442	0.3787	-0.12	0.9071
AgeCat	3	-0.4089	0.5506	-0.74	0.4587

Odds Ratio Estimates
Effect	Point Estimate	95% Confidence Limits
NOTE: The degrees of freedom in computing the confidence limits is 179.
AgeCat 1 vs 4	1.956	0.359	10.664
AgeCat 2 vs 4	1.067	0.193	5.882
AgeCat 3 vs 4	0.741	0.096	5.694

Association of Predicted Probabilities and Observed Responses
Percent Concordant	41.0	Somers' D	0.182
Percent Discordant	22.7	Gamma	0.287
Percent Tied	36.3	Tau-a	0.044
Pairs	3875	c	0.591

[PaigeMiller]

---

@westbestern wrote:

I have survey data and I want to do a bivariate logistic regression of age categories (18-29, 30-39, 40-49, 50+) on PrEP usage (yes, no). Here is my code below

proc surveylogistic data=dataname;
class AgeCat;
model takePrEP= AgeCat;
run;

I want to report the p-value, odds ratios, and confidence intervals for the AgeCat variable.

---

The p-value is 0.4646 in this line

AgeCat	0.86	3	177	0.4646

There is no odds ratio for AgeCat; there are odds ratio for the different levels of AgeCat vs level 4 of AgeCat.

You want confidence intervals for what statistic regarding the AgeCat variable? You need to specify.

[westbestern]

I want the confidence intervals for the odds ratio of each level of the AgeCat variable. For example, I want to see if an age like 18-29 is significant compared to age 50+

[PaigeMiller]

Those are given here:

Odds Ratio Estimates
Effect 	Point Estimate 	95% Confidence Limits
NOTE: The degrees of freedom in computing the confidence limits is 179.
AgeCat 1 vs 4 	1.956 	0.359 	10.664
AgeCat 2 vs 4 	1.067 	0.193 	5.882
AgeCat 3 vs 4 	0.741 	0.096 	5.694

[westbestern]

There are four categories for age but there seems to be only three in the output. Is the intercept the first age category?

[sbxkoenk]

Every category is "compared" to the reference category, so that makes 3 comparisons (3 odds ratio's).

And you have also 3 parameter estimates indeed.

The way to interpret them depends on the coding-scheme that was used for your age categories: effect coding or pure dummy coding or ...

Koen

[westbestern]

So the first category (ages 18-29) is the reference category, right?

[sbxkoenk]

No, I think the last category is the reference category (by default).

Use REF=FIRST if you want the 1st category to be the reference category instead.

From the doc (you can make any category the reference category) :

REFERENCE=’level’ | keyword
REF=’level’ | keyword

specifies the reference level for PARAM=EFFECT or PARAM=REFERENCE. For an individual (but not a global) variable REF= option, you can specify the level of the variable to use as the reference level. For a global or individual variable REF= option, you can use one of the following keywords. The default is REF=LAST.

FIRST

designates the first-ordered level as reference

LAST

designates the last-ordered level as reference

Koen