Data Analysis

This week marked the second and final week of data collection for my research. I was able to procure the necessary patient consent and analyze all 24 results like I intended to. Overall, I was able to get 12 adults and 12 children. As I stated last week, I was only able to get 10 males and 14 females, but from going through some previous sources and my literature review, I have found evidence that shows that the severity or type of ADHD does not differ depending on gender. Females and males alike are affected by ADHD the same way, so having a disproportionate number of each did not affect my results (or should not). After gathering all the data I needed, I met with Mrs. Haag and was able to work out a plan for data analysis. The majority of my analysis was conducted over this weekend and I was able to answer my question and find the greater conclusion of the study. So the rest of the week will be dedicated to finding some other key trends and outlining my results section. I also need to definitely present my conclusion in a cleaner way than I have it in my spreadsheet like through some tables and graphs to show the distribution of scores.

From conducting the paired t-test on the adults and children group (both by hand and by graphing calculator), I found the t-score for the children's group to be 4.95 and the t-score for the adults to be 5.53. I did this by taking into account the difference between the final and initial ADHD scores (from the ADHD diagnostic tests) for all the patients in each group. This was the procedure of how I did it:

1. Add up each category for first child patient to find initial ADHD score and final ADHD score.

1. Highest possible score: 300
2. Lowest possible score: 50

2. Determine if the results are a normal distribution for the children group, meaning when ADHD scores are plotted on an x,y graph (one for initial and one for final scores), there are less results in the extremes and more results for the moderate scores, forming an approximate bell curve.

2. Determine it is a normal distribution.

3. Calculate the difference by subtracting final score from initial score.

4. Take the absolute value of this difference.

5. Repeat for every child patient.

6. Add up all these difference values.

7. Square each difference value for every patient.

8. Add up all the squared difference values.

9. Find t-score using this equation:

For which, ΣD is the sum of the differences between final and initial scores.

ΣD2 is the sum of the squared differences

(ΣD)2 is the sum of the differences squared.

N is the number of samples

A t-score is the ratio between the difference between the initial and final score for each child and the difference within the total children data set. A larger t-value means the scores are more different, meaning a higher trend of improvement.

11. Find the degrees of freedom by subtracting 1 from total children sample size (12).

12. Then find the p-value in the standardize t-table with 11 degrees of freedom.

a.  The null hypothesis is established as follows: the distribution of results has a mean equal to 0.  

b.  The p-value will determine whether the results are due to chance or due to neurofeedback therapy sessions.

c.  The t-score will have a designated p-value that will show what is the probability that the results are due to random chance. The lower the p-value, the more likely the results are due to neurofeedback therapy sessions rather than chance, thus rejecting the null hypothesis.

12. Repeat steps 1-12 for adult group. 

13. After p-value for children group and adult group are checked to see if the difference is significant, then compare the t-scores.

14. Find the percent difference between adult and child value to see if significant difference.
a.    T-scores represent trend of improvement. The larger the t-score the larger the trend of improvement.
b.    Compare adult and child t-score to see which improved more.
c.     Take into account which has a lower p-value to see which value can be attributed less to chance and more to neurofeedback therapy sessions.

15. Conclude which group, if any, has a significantly higher trend of improvement. 

a.              Adult group t-score- child group t-score/adult group t-score
b.              According to standardized rules of statistics, if this number is higher than 5%, then percent difference is significant.

I used the t-scores to represent the trend of improvement of each group. The children's group p-value was 0.000218, meaning that, because it is less than 0.05, that the results are significant and not due to chance but rather that the neurofeedback therapy sessions did make an impact on reducing the ADHD scores. Similarly, for the adult group the p-value was 0.000089, which is less than 0.05, so the results are significant and not due to chance. On average, adult ADHD scores reduced by 75.6 and child ADHD scores reduced by 59.5. These results confirmed previous studies conclusions that neurofeedback therapy sessions do make a difference in treating ADHD patients.

To find which group improved significantly more than the other group, I calculated the percent difference. I found the adult's group to improve more than the children's group by 10.5% (more than 5%, so a significant difference). So, overall adults improved more than children by a significant percentage. This disproved my hypothesis. Using evidence from previous two studies conducted on neurofeedback therapy, I had predicted that neurofeedback would be more influential in treating inattention, a primarily childhood-associated symptom, than hyperactivity, an adult ADHD symptom, so this would mean children would improve more than adults. I was surprised by the actual results. This conclusion that adults improve more seems to counter the theory of neuroplasticity that brains can change throughout life, forming new connections between neurons and rewiring and reconfiguring existing connections based on experiences and learning. Children have been most evidenced to have increased neuroplasticity, as brains that are younger are still in the growth stage and have not undergone the "synaptic pruning" maturation stage, which is the altering of the brain structure and the reduction of connections that happens as we develop and age. Adults improving more than children, however, may not be due to children's lack of neuroplasticity, but might be due to the fact that it is hard to get children to sit down and concentrate. It takes more persuasion than it does for adults, who are motivated and realize that they have to consciously work every day for the sessions they pay for to pay off. This conclusion was really interesting and has brought forth more possibilities in other facets of neuroscience. I would like to include these additional ties in my conclusion as possible explanations of the outcome, but then I might have to rework/add it to my literature review.

So, now that I have the main conclusion I can start to look at other trends within the data. I would like to conduct additional statistical analyses to see whether there is a higher trend of improvement for younger ages of the children group more than the older ages (teenagers). I would also like to see if increased neurofeedback therapy sessions result in a greater reduction of symptoms, and thus a lower ADHD score, indicating more improvement from the sessions. I could see if my results confirmed the numerous studies that show there is no difference between males and females with their exhibition of ADHD and determine if one group possibly has more severe ADHD or improves more. I could also see if the longer the patient came to the clinic had any effect on reducing symptoms more. For example, if a patient spaced out their sessions to be just once a week and thus came to the clinic once a week for 6 months v. a patient that had a session every day that allowed them to come to the clinic every day for only a month. I could also see which particular category's score (response time, d prime, or variability) was more influential in reducing the overall ADHD score. Or if more children or adults had their ADHD scores changed from severe (over 200) to average (100-200) or more just reduced within the average range. Since there are so many smaller conclusions and other key trends I can find from the data, I think the hard part would be seeing which are more significant, more related to my research, and more impactful/helpful to the sphere of neuroscience to include in my results section.

So, that has been my week so far. This week I hope I can gain greater insight into my study and find the more nuanced differences that answering my question may yield. This has probably been the most interesting part of my research and the most rewarding in a sense. The fact that the data I chose to analyze and procure actually means something and there is a real-life conclusion coming from my research is truly exciting.

(1510)