Panel Survey

[Ismail Boujnane]

Hi, I would like to request your expertise on my panel survey study.

I have some cross-sectional (categorical) data collected from a questionnaire in 2021, which are integrated into a longitudinal dataset collected at different points in time for a period of 6 years, from 2015 to 2020. Knowing that the sample is the same for both data collection method, and my categorical data (institutional support, corporate governance) are dynamic, not static, I want to know if integrating them into my panel data is feasible.

I would really appreciate your return on this.

[Carlo Lazzaro]

Ismail:
does your sample vary across the data waves?

[Ismail Boujnane]

Carlo Lazzaro thank you for your return. Let me clarify the context of my research: the sample consists of 100 SMEs in the manufacturing sector.
Both data collection methods were implemented for the same sample. So, does that mean they vary across the data waves? It is the same sample so we can say that it does not change. My concern is if I am integrating the categorical variables (institutional support, corporate governance) captured at one point in time (2021) into a panel dataset (2015 - 2020). Is the methodology correct? What would be the justification?

[Carlo Lazzaro]

Ismail:
the issue with the categorical variables is that they were measured only once.
Exception made for the case in which they do not change over time (and you should have robust justification about that), I do not think that a panel data analysis is feasible with these data.

[Ismail Boujnane]

Carlo Lazzaro Thanks very much for your justification. Allow me to ask you: What model do you think would be suitable in this case? Do you think the cross-lagged panel model would be applicable for testing the categorical data effect since they were captured in 2021 and the firm data were collected up to 2020? I wonder if I could use the first lag of categorical variables through CLPM and test their effects.

[Carlo Lazzaro]

Ismail:
I think I would go panel data regression getting rid of those variables that were collected in 2021 (I assume that you have other predictors that were colected during 2015 2020).

[Ismail Boujnane]

Carlo Lazzaro Indeed I have other predictors collected between 2015 - 2020 period. However, the categorical data form an essential part of my thesis research objective as I aim to assess their moderation impact on my dependent variable. So I don’t think excluding them would be favourable. What would be the adequate solution ?

[Carlo Lazzaro]

Ismail:
assuming that the predictors you're interested in are time-invariant, paste them for the 2015-2020 data waves (provided that you have robust justification about this methodological choice).
However, assuming that your dependent variable is continuous, you will get their coefficients with -xtreg,re- ony.