topic Re: Finding first level risk associated with t-test in SAS Procedures

Finding levels of risks associated with t-test

Mutua — Tue, 05 Jan 2016 22:40:07 GMT

Dear SAS community,

I have a simple question that came from a reviewer of my mnuscript, but i can´t get my head around it. I performed a t-test where i compared two means. I found that there are signficant differences between the means. Therefore i created a table of each comparison and indicated the signficance difference using asterisks. However the editor of the journal has constantly asked me to state the first level risk associated wth the results i gave in the table of the t-test comparisons.

My questions is what is the reviewer asking for? that i state its type one error and the risk levels are less than 0.05? (p-value limit)?

Your support is highly appreciated.

Best regards,

Peter.

Re: Finding levels of risks associated with t-test

ballardw — Tue, 05 Jan 2016 22:54:19 GMT

Have you read the publication where someone else did a t-test? I would expect that to provide a pretty concrete example.

Re: Finding levels of risks associated with t-test

Mutua — Tue, 05 Jan 2016 23:00:41 GMT

Dear ballardw,

Yes i have read, but i have not come across that term before. Please kindy explain.

Best regards,

Peter

Re: Finding levels of risks associated with t-test

Reeza — Tue, 05 Jan 2016 23:00:50 GMT

We don't have enough information to answer your question. It depends on what type of analysis you undertook and what kind of data you have. The reviewer may be asking for an odds ratio or relative risk.

Given you said you made a table of "each comparison", I'm guessing that maybe T-Test isn't appropriate, since you would then be comparing multiple means independently. ANOVA may be more appropriate...but again, we don't have enough information.

Re: Finding levels of risks associated with t-test

Mutua — Tue, 05 Jan 2016 23:12:41 GMT

Thankyou Reeza for the reply.

To explain i had a treatment where i added a biological control and the pathogen in the soil where potato plant was planted. The other treatment was where i added the pathogen alone and also a potato plant was included (all in pots). Each treatment was replicated 10 times. At the end of the experiment i extracted the pathogen and determined the population. The main intention was to detect if the biological control agent had any influence on the pathogen. I then performed a Ttest to compare the means between those treatments with pathogens and those without. The results indicated that adding the biocontrol led to higher pathogen numbers and therefore significantly different from that without the biological control.

The reviewer states that i should state the level of risks associated with the signficant differences between the means (if i get it right).

Its my hope that the question is abit clearer. Looking forward to your feedback.

Best regards,

Peter

Re: Finding levels of risks associated with t-test

ballardw — Tue, 05 Jan 2016 23:19:21 GMT

@Mutua wrote:

Dear @ballardw,

Yes i have read, but i have not come across that term before. Please kindy explain.

Best regards,

Peter

In my admittedly limited reviewed article publications it appears that different fields, or sometimes even a single publication, have some of their own jargon. I don't recognize this particular phrase either which is why I would look in the publication to find something similar and see what I missed. Or see if the commenting editor would point to an example.

Re: Finding levels of risks associated with t-test

Reeza — Tue, 05 Jan 2016 23:21:09 GMT

Is it level of risk or "first level of risk"? I would run your article by a statistician, I would have assume some sort of analysis to take advantage of the replication would have been used, instead of straight t-test on the means.

Re: Finding levels of risks associated with t-test

Mutua — Tue, 05 Jan 2016 23:24:24 GMT

Dear Reeza,

its "the first level of risk" that the editor is asking for.

Re: Finding first level risk associated with t-test

Mutua — Tue, 05 Jan 2016 23:32:03 GMT

Dear Reeza,

What type of additioal analysis should i have performed to get this details on "first level risk"?

Regards,

Peter.

Re: Finding first level risk associated with t-test

Reeza — Tue, 05 Jan 2016 23:50:51 GMT

I'm not familiar with the term either and haven't come across it before and Google isn't helpful here either.

I suspect either it's

1) Translation or Typo issue

2) Something in the analysis/methodology section of your table/data that wasn't specified.

If you can, ask the editor for clarifications.

Re: Finding levels of risks associated with t-test

Ksharp — Wed, 06 Jan 2016 02:44:03 GMT

Maybe Reviewer is thinking you are using ANOVA and ask you what is your reference/base level .

Re: Finding first level risk associated with t-test

Mutua — Wed, 06 Jan 2016 03:43:57 GMT

Thanks Reeza for the feedback. I really appreciate.

I will follow it up with the editor and see what can be done or if there are issues i missed in the table.

Best regards,

Peter.

Re: Finding levels of risks associated with t-test

Mutua — Wed, 06 Jan 2016 03:46:22 GMT

Thanks Keshan for the reply. I will follow up with the editor for further guidance on the issue.

Best regards,

Peter