Conduct a two-way ANOVA

Problem Set 1: The Two-Factor ANOVA for Independent Measures

Research Scenario: A clinical psychologist is interested in whether a new training program will impact the severity of trauma symptoms in first responders. He is also interested in whether different types of first responders experience different levels of trauma, and whether there is a difference in how the treatment programs help these responders. To help answer this question, he recruits paramedics, firefighters, and police officers who are entering their first year of service. He divides each group in half: one half receives the original training program, and the other half receives the new training program. After their first year in service, the psychologist administers the Davidson Trauma Scale (DTS; Davidson, 1997). The DTS is a self-report instrument that indicates the presence and severity of trauma symptoms. It is often used to detect PTSD, with scores ranging from 0 to 136, and a mean in the general population of 11 (Davidson, Tharwani, and Connor, 2002). The results for all groups are in the table below. Due to unequal samples, some groups had more participants than others.

Conduct a two-way ANOVA to determine the effect of training program and type of first responder on DTS scores. Create a multiple line graph that shows the difference in DTS scores among the groups. (Note that the table continues on the next page.)

Original Training Program

New Training Program

Paramedics

25

18

17

14

20

17

10

21

9

19

16

11

Firefighters

6

10

27

13

11

9

12

9

9

5

6

10

8

Police Officers

7

12

15

6

10

15

14

13

5

8

14

11

7

Problem Set 2: The Two-factor ANOVA for Independent Measures

Research Scenario: Counselors at a substance abuse clinic are interested in the differential effects of opioid dependence counseling for different types of clients. In particular, they have read that there may be differences in treatment response based on whether prescription opioid-dependent clients have a history of chronic pain and/or heroin use (i.e. Weiss, Potter, Fiellin, Byrne, Connery, Dickinson, et al., 2011). To test this theory in their own clinic, they divided new clients based on whether or not they had a history of chronic pain (> 6 months) and whether or not they had a history of heroin use (used at least 1 time). All clients were treated over 16 weeks with a combination of buprenorphine-naloxone stabilization/treatment plus twice weekly individual opioid dependence counseling. Each client was asked to submit urine and report how many times they had used opioids during a 1-week period, each week for 16 weeks. As a first pass through the data, the counselors added the 16 weekly reports for each client to get one overall count of opioid use (i.e. if a 20 is recorded, it means the client reported using opioids 20 times over the 16-week period). The results are in the table below. Based on the data, is there an effect of history of pain or heroin use on subsequent opioid use during treatment?

Using this table, enter the data into a new SPSS data file and run a two-way ANOVA to test whether there is an effect of pain history or heroin use on treatment outcome, as measured by the number of times using opioids during treatment. Create a multiple line graph to show the difference among these scores.

No history of chronic pain

History of

chronic pain

No history of heroin use

10

13

8

18

11

17

11

7

19

11

16

10

6

9

14

12

History of heroin use

19

13

27

18

31

23

28

17

38

31

42

27

12

14

6

16