If I need to validate 900 results, what percentage must I check to have 95% confidence the results are correct?

To determine the percentage of results you need to check to have 95% confidence that the overall results are correct, you can use a statistical technique called sampling. Let me explain how to calculate it step by step:

Step 1: Determine the margin of error:
The margin of error represents the tolerable deviation from the true population parameter. It is typically expressed as a percentage. The commonly used margin of error for estimating proportions is usually around 5%.

Step 2: Determine the desired confidence level:
The confidence level represents the level of certainty you want to have in your results. In this case, you want to have a 95% confidence level.

Step 3: Calculate the sample size:
To calculate the required sample size, you need to use a formula that takes into account the margin of error, confidence level, and the total population size.

The formula is:
n = (Z^2 * p * (1-p)) / e^2

Where:
n = sample size
Z = Z-score corresponding to the desired confidence level (e.g., for 95% confidence level, Z ≈ 1.96)
p = estimated proportion of success (if unknown, you can assume 0.5 for a conservative estimate)
e = margin of error (expressed as a decimal)

Step 4: Plug in the values and solve for sample size:
In your case, the total population size is 900 (as you mentioned). Let's assume the conservative estimate of the proportion of success (p) is 0.5, and the desired margin of error (e) is 0.05 (5%). The Z-score for a 95% confidence level is approximately 1.96.

Using the formula above:
n = (1.96^2 * 0.5 * 0.5) / 0.05^2
n ≈ 384.16

Since you cannot have a fractional number of results, you would need to round up to the nearest whole number. Therefore, you would need to check at least 385 results from the total population of 900 to have a 95% confidence that the results are correct.

So, to answer your question, you must check approximately 43% (385 out of 900) of the results to have 95% confidence that the overall results are correct.