A Type II error is a decision error made about the data. Additionally, there must be a hypothesis of some kind present. A measure, instrument, or procedure does not contain a hypothesis.

Perhaps your advisor is getting at something like this: If a study has flawed methodology, this will affect the data, which could result in a Type II error (miss). So, one's methodology could indirectly lead to a Type II error, but the error itself reflects an erroneous decision/conclusion about the data - not the measure or procedure used to collect the data.

In practice -- if you know your methodology is wrong or flawed, you shouldn't be concluding ANYTHING from the data

In addition to Blaine Tomkins, I could also imagine two explanations what your supervisor meant:

1) If you use unreliable instruments, the assessed relationship between variables may underestimate the true strenght of relationship (and is overly confident in the prediction). Typically, regression models assume that predictors are measured without error, therefore SEM may be a better approach to estimate the error directly.

2) I you choose an inappropriate instrument, which is per se less related to your outcome variable. This is not a Type II error in a narrow sense, but if you fail to identiy/hypothesize the smaller relationship, you will conduct an underpowered study (since you did a power analysis on false premisses, althoug you should assess the smallest effect size of interest and not an average expected effect size, but this is another topic).

Hello! I immediately apologize for my poor English.

"Type II error" in the field of metrology is also often used when a decision is made about the conformity / non-conformity of an object. For example, during the verification of the measuring means, if an unsuitable standard is selected, or the work is performed by incompetent personnel, etc., a suitable device can be mistakenly "rejected". It is possible to calculate the probability of an "Type II error" from the ratio of the accuracy characteristics of the object and the standard.

Type ll error is actually an error based on the method of analysis applied on the data. For instance, if you are supposed to carry out a regression analysis of a Null hypothesis, fail to do so, type ll error is said to have occurred. Type lll error is giving a wrong interpretation of the outcome of an analysis. So, if you are to accept the Null hypothesis, but rejected it, type lll error is said to have occurred. Same is when a Null hypothesis should have been rejected, but, rather, accepted in error, this error is a type lll error.

Abel Igah Simeon Not necessarily. One could conduct an appropriate test with valid data and still commit a Type II error.

Yes, you are correct. Type ll error focuses on the method of analysis applied and the tools used for the analysis. Type ll error is avoided when the right method is applied and the right statistical tools are used.

Yes, your Supervisor is correct. Type II error relates to both methodology and the test statistics. Therefore this error can be avoided when the right method is applied leading to the right statistical tool for testing. God bless.