cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
構造方程式モデルを用いた経時的軌跡モデリング_Laura Castro-Schilo(2021-JA-45MP-US01)

レベル:中級

 

【発表者】

Laura Castro-Schilo, JMP Sr. Research Statistician Developer, SAS Institute Inc.

 

【発表概要】

構造方程式モデル(SEM)は、観測された変数と観測されていない潜在的な変数の間の関連性を研究できるフレームワークです。SEMの多くのアプリケーションは、クロスセクションデータを使用しており、このフレームワークは縦断的データのモデリングにも非常に柔軟性があります。

 

本発表では、経時的な成長の軌跡を特徴づけるための柔軟なツールとして、潜在成長曲線モデル(LGCM)について説明します。まず、SEMの基本的な概念を簡単に説明した後、成長の理論を検証するために平均値をどのように分析に組み込むかご紹介します。本研究では、新型コロナウイルスのパンデミック発生時に個人が報告した「不安」と「健康への不満」のデータにモデルを当てはめることで、LGCMを説明します。分析の結果、レジリエンスは、不安と不定愁訴の軌跡における独自の変化パターンを予測することがわかりました。

 

※このサイトでビデオを再生して画面右下の「CC」をクリックして「English」を選びますと英語の字幕が表示されます。

本発表の発表資料やデータのダウンロードや、日本語トランスクリプトと共に視聴されたい方は、こちらをクリックしてください。
Discovery Summit Americas 2021 Presentationの各発表のページに移動します。

(SAS Profileへのログインを求められますので予めご了承ください。)

Comments
madhu

Hi Laura. This is another excellent presentation on SEM.  Is this possible to get the data you used in this presentation? This is to practice SEM with your analysis. 

LauraCS

@madhu, Thank you for your kind note! I'm glad you enjoyed it. I uploaded the data from the example here, although it's worth stating that the data and analyses I conducted in my presentation were for demonstration of the SEM functionality only. The data are more complex than what they appear (for example, there's selection bias among those who continue participating in the study over time, there are non-Gaussian distributions, etc) and you could learn more about them in these links:

https://www.sheffield.ac.uk/psychology-consortium-covid19

https://osf.io/v2zur/

 

HTH,

~Laura

madhu

Dear Laura

Thank you for your reply and directing me to the data source of your presentation. I have checked the website could not locate the right article/study. Following the closest studies I have found.

 

OSF | STUDY: Modelling changes in clinically relevant anxiety, depression, COVID-19 related traumati...

 

OSF | STUDY: A longitudinal assessment of depression and anxiety in the Republic of Ireland during t...

 

OSF | STUDY: Trajectories of change in internalizing symptoms during the COVID-19 pandemic: A longit...

 

Is the study you used your presentation still available on their website?

Kindest regards

Madhu Acharyya 

madhu

Hello Laura Castro-Schilo 

I have now collected the original data for three waves (W1, W2 and W3) that the researchers used in the study. However, I guess that you compressed the datasets and only used a number of selected variables in your JMP analysis. 

Watching your recording, I can note only 21 variable names as follows:

1. pid

2. Present at W1

3. Present at W2

4. What is your age?

5. What is your gender?

6. W1_Education Variable Binary

7. N Children under 18

8. What is your occupation?

9. Resilience

10. Anxiety March

11. Anxiety April

12. Anxiety June

13. Health Complaints March

14. Health Complaints April

15. Health Complaints June

16. Depression March

17. Total GAD Score (7 items)

18. Loneliness March 

19. Depression April

20. Total GAD Score (7 items)

21. Loneliness April

 

It seems that you have created two columns:

22. Match plag

23. Select

However, I do not understand the type and formula you used to create these two columns

It seems that there are altogether 26 variables in your data set. Are the remaining variables are:

24. Depression June

25. Total GAD Score (7 items)

26. Loneliness June

 

What is Total GAD Score? What formula have used to calculate the score?