User talk:Sammy.kritzer/sandbox

• The p value (“Sig.” in SPSS) is less than .05.

• We rule out sampling error as the cause of any difference that we see between the groups that we’re comparing (because the probability is less than 5%).

• There is a real difference / relationship / effect.

• We reject the null hypothesis.

In the nutshell The average score on the RDI in a sample of college students is 50. Is the mean of the RDI.pre scores for the 300 individuals in this study statistically significantly different from the mean for college students? Test: Single Sample t test Answer: No (t[299]=-0.260, p = .795).

Do NFTR participants differ from non-participants with respect to RDI.post? Test: Independent-Sample t test Answer: Yes (t[298] = -4.22, p < .001).

Do RDI scores differ between pre- and post-program? Test: Repeated-Measures t test Answer: Yes (t[299] = -59.192, p < .001).

Does the NFTR sample contain more men than women or does it show equal numbers of individuals in each gender? Test: Chi-Square Answer: Yes (χ2 = 33.33, p < .001).

Do NFTR participants with inpatient, outpatient or no treatment prior to participation in this study differ on RDI.post scores? Test: ANOVA Answer: Yes (F[2, 297] = 626.96, p < .001).

Which previous treatment categories show pairwise differences? Test: post hoc Answer: All of them.

Is there a relationship between RDI.pre scores and RDI.post scores? Test: Correlation Answer: Yes (r = .935, p < .001).

Does gender, program participation, and Repeat.DWI predict RDI.post scores? Which of the variables are significant predictors of RDI.post scores (if any)? Test: Regression Answer: Yes, the model is good (F[3, 296] = 6.147, p < .001). Only Program is a statistically significant predictor (ϐ = .230, p < .001).

Writing Up Results

The t Test

The participants showed a statistically significant increase in test scores after being exposed to the treatment, t(8) = 6.00, p < .05, r2 = .30 (or 30%).

ANOVA

Participants with previous inpatient treatment showed statistically significant differences in RDI post-test scores, F(2, 297) = 626.96, p < .001. Post hoc Bonferroni tests showed differences between inpatient and outpatient (p < .001), no treatment and inpatient (p < .001), and no treatment and outpatient (p<.001).

Chi-Square

There is a statistically significant difference between the NFTR sample and the US population in gender distribution, χ2 = 33.33, p < .001. The NFTR sample is predominantly male.

Correlation

There is a significant, strong, and positive relationship between RDI pre-test scores and RDI post-test scores, r = .935, p < .001.

Regression

The model of gender, program participation and repeat DWI predicting RDI post-test scores is good (F[3, 296] = 6.147, p < .001, r2 = 0.59). Only Program is a statistically significant predictor (β = .230, p < .001).