By admin


Get to Know Margin of Error and Level of Confidence Better

Categories : General

The survey is a valued appraisal tool, where a sample is chosen and data collected from the sampling is generalized with a large population. Surveys are similar to wine tasting, where a couple of sips tell what the entire barrel full of wine tastes like.

The legitimacy of a survey is uncertain. Just like the wine that represents the entire barrel has to be swirled, the sampling group needs stirring before you choose the respondents. Choose respondents randomly so researchers can generalize the survey results to the entire population.

Two crucial statistics gauge the represented population.

  • Margin of error
  • Confidence level

These two tell how well the chosen respondents represent the entire population. The margin of error is also termed a confidence interval. It reveals how much to expect from survey results to echo overall population views.

Small ME = More confidence level in the results

Big ME = straying away from total population views

The margin of error values ranges above and below actual survey results. For example, a 70% ‘Yes’ response has a 5% margin of error means the assumed “yes’ answer from the entire population lies between 65% – 75%. On Ovation MR, try this simple to operate margin of error calculator for any survey.

Sample size and its impact on the margin of error

An example,

A small sample size of 50 respondents has 14% ME, while 1,000 respondents have 3% ME. If the respondents [sampling size] doubles to 2,000 then the ME decreases from minus or plus 3% to minus or plus 2%.

The confidence level according to the industry standard is 95% but in some situations 90% is sufficient. Researchers can gain a 90% confidence level with small samples, which means a less costly survey. For gaining a 3% margin of error at a 90% confidence level the sample size is around 700 respondents. For 95% confidence level sample size will be around 1000.

The margin of level shows the inbuilt sketchiness within the survey data. It offers a range and not a particular number. If surveys are conducted every 5 to 6 months to gain an insight of improvement in customer service and the percentage of response of ‘very good’ reduces from 49% to 46% within 6 months then both are accurate because they fit the margin of the error range.

Statistically, the decrease is trivial. Alternatively, if the percentage rises from 49% to 53%, there is a small increase in customer service.

Biases about confidence levels

A confidence level of 95% means 5% survey goes wacky. It means if 100 surveys using the same question are made then 5 respondents will provide wacky answers. Researchers are not concerned about the 5% as they will not repeat the same questions, so they assume they will receive results from 95%.

If the same question is repeated, then the researchers will obtain unexpected numbers like 50%, 49%, 54%, etc. If 20% surfaces in one period and 45% in the next, then it is safe to assume that 20% belongs to the 5% wacky portion.

Now, you know about the margin of error calculation and its impact on the results. Therefore, use a sample size calculator available online to determine the respondent’s essential to take a survey.