# What value to enter?

#### Barnzilla

##### New Member
A friend of mine is entering data that she has collected into a database and has the following question:

The sensitivity of reporting for insulin is no lower than 2 ulU/ml. If
it is lower than this, they simply report <2. What do you think is the
best way to input this as a number value? The range of our entire
insulin data is from <2 to 34.6. Only one individual has insulin as
low as <2. Her insulin has a much smaller range at <2 to 3.85. I
don't know if taking <2 as insignificant (ie. 0) or as an average (ie.
1) would be best.
Any ideas on what she should enter for this person into the database?

#### Statstastic

##### New Member
Hi,

I'm no expert, but in this situation I'd be tempted to use the value 1 over the value 0.

My intuitive reason would be that - assuming observations are uniformly distributed throughout the interval 0 to <2 - the probability that 0 is lower than her actual value is 1 (or close to it), while the probability that the value 1 is higher or lower than her actual value is 0.5.

I'm a bit rusty on my stats theory, but there's probably some stats proof for taking the mid point of an interval. Maybe it's the maximum likelihood estimator (?) or something... #### Barnzilla

##### New Member
Hi,

I'm no expert, but in this situation I'd be tempted to use the value 1 over the value 0.

My intuitive reason would be that - assuming observations are uniformly distributed throughout the interval 0 to <2 - the probability that 0 is lower than her actual value is 1 (or close to it), while the probability that the value 1 is higher or lower than her actual value is 0.5.

I'm a bit rusty on my stats theory, but there's probably some stats proof for taking the mid point of an interval. Maybe it's the maximum likelihood estimator (?) or something... Thanks so much for the input. I've never had to deal with this problem before, so I'm completely at a loss as to how to go about it.